Silicon Investor - Kirk's Market Thoughts

[nicewatch] All FYI only. Still unsure what to make of quantum computing in its current form...

nicewatch — 9/12/2026 8:04:10 PM

All FYI only. Still unsure what to make of quantum computing in its current form despite several small retail meme stock bubbles in it this decade. :-) Some of the interesting private "quantum" companies I know of aren't even on their list. Please don't ask me what they are although one is easy to figure out if you research recent physics nobel laureates working on quantum computing startups. :-)
_____________________________________________________________________________________________________________________________________________________________

[Investment Map] 23 Companies in the Quantum Cycle: From Materials to Logical Qubits $IONQ $QNT $XNDU $GFS $NVDA and 18 more | Quantum computing value chain investment map

PhotonCap Sep 10, 2026

Over the course of 2024, the four quantum pure plays gained an average of 1,063 percent. Then the following year, on January 8, 2025, the day after Jensen Huang made a shocking remark at CES, saying useful quantum computers were 15 years away on the early side and 30 years on the late side, Rigetti dropped 45.4 percent in a single session [1]. Hamamatsu Photonics rose 2.4 percent that same day. From then until now the pure plays are down about 5 percent on average, while the component and metrology layer beneath them is up 129 percent. Most of that 129 percent, though, was built by AI. This piece is an investment map that reprices the quantum supply chain AI has been covering up, across 23 listed companies and five axes. And for each of the 23 names I measured how much of a quantum stock it actually is, two different ways, instead of eyeballing it. One is the single-day move on that January day. The other is a quantum beta over the last 64 trading sessions. As far as I know, nobody has published these numbers name by name. With the US and China accelerating the race for quantum supremacy as the thing that comes after AI, and since May 2026, when the US Commerce Department attached equity conditions to funding for nine quantum companies, I think the downside in this industry has changed shape.

TL;DR

The Defiance Quantum ETF rose 54.9 percent over the past year while the median of the four quantum pure plays was negative. Most of the return came from outside the obvious quantum stocks.
On September 8, 2026, the US Commerce Department turned the first three May letters of intent into definitive agreements, putting $100 million each into D-Wave, Rigetti, and Quantinuum while taking equity stakes. Six of the nine remain pending.
I measured “quantum purity” across 23 listed companies two ways: the Jensen shock and a 64-session quantum factor beta. Only nine show statistically significant quantum sensitivity.
The highest-ranked company on this map is not a quantum pure play. Strip out optical interconnect, the signature axis of my own framework, and the same company still ranks first.
Government capital lowers financing risk, not technology risk. It can keep funding from drying up. It cannot make the quantum computer work.

Contents

Two eras: 1,063 percent and minus 5 percent
Why I am the one drawing this map
Background one: qubits, errors, and logical qubits
Background two: five platform branches
The farther you go, the more light wins
The government attached conditions
Three frameworks
23 companies and who works with whom
The five axes
Rankings, and quantum purity
The lag between tiers, and what I have not confirmed
Catalyst calendar: the next 12 months
Group mapping and hidden cards
Company cards
Scenarios
Where this framework breaks
Watchlist: private and adjacent
Closing
References and sources

1. Two eras: 1,063 percent and minus 5 percent

Over 2024, Quantum Computing Inc (QUBT) rose 1,713 percent, Rigetti (RGTI) 1,449 percent, D-Wave (QBTS) 855 percent, and IonQ (IONQ) 237 percent. Simple average across the four: 1,063 percent. More than a tenbagger.

In the same year Coherent rose 118 percent, Tower Semiconductor 69 percent, FormFactor 5 percent, MKS and Keysight 1 percent each, while GlobalFoundries fell 29 percent and Hamamatsu fell 39 percent. Average across those seven: 18 percent.

Look only at 2024 and the answer seems obvious. If you want to invest in quantum, buy quantum companies. Why would you buy component makers.

But I think that rally was mostly hype with nothing inside it. Prices ran ten times over, to levels the revenue and product scale of the time could not explain. What changed that year was attention, not technology. Which is why one sentence from Jensen was enough to break it.

At the CES Q and A on January 7, 2025, Jensen Huang said this. “We are probably five or six orders of magnitude off. If you said 15 years for very useful quantum computers, that would probably be on the early side. If you said 30, it is probably on the late side. If you picked 20, a whole bunch of us would believe it” [2]. The next day Rigetti fell 45.4 percent, Quantum Computing Inc 43.3 percent, IonQ 39.0 percent, and D-Wave 36.1 percent [1][3].

That same day the tech names held up: Coherent fell only 3.9 percent, Keysight only 0.7 percent, and Hamamatsu Photonics actually rose 2.4 percent [3].

Single-session moves on January 8, 2025, four quantum pure plays versus seven component-layer names

One sentence from one person took 45 percent off some assets in a day, and other assets in the same industry went up on that same sentence. What that difference is, is where this piece starts.

That one-day move gets used again later. As a ruler for how much of a quantum stock a given name really is. This was the largest quantum sentiment shock on record, so how far a company moved that day is its quantum purity. The broad market barely moved: SMH, the semiconductor ETF, was down 0.7 percent. Meaning it was not a day when the whole market sold off.

Look at what came after that day and the direction flips. From January 7, 2025, the day before CES, through September 4, 2026, the four pure plays are down 5 percent on average. Over the same window the seven component and metrology names are up 129 percent. Tower Semiconductor 329 percent, Coherent 189 percent, MKS 131 percent [3].

Cut it to the last year alone and the picture holds. IonQ minus 4.1 percent, Rigetti minus 0.2 percent, D-Wave plus 7.2 percent, Quantum Computing Inc minus 45.2 percent. Average across the four is minus 10.6 percent, median minus 2.2 percent. IBM is flat too at minus 4.3 percent. The seven component and metrology names, meanwhile, are up 156.1 percent on average, 161.4 percent at the median [3]. Those seven are the ones just counted: Coherent, Tower, FormFactor, MKS, Keysight, GlobalFoundries and Hamamatsu. AXT is excluded because the InP cycle is what drives it.

One more number belongs here. Over that same year, the Defiance Quantum ETF (QTUM) rose 54.9 percent [3]. Which means that while a quantum ETF was up 55 percent, the median quantum pure play was down. So where did that 55 percent come from. As you would guess, the answer was outside the pure plays.

Something has to be said honestly here. The real driver that pushed the component layer into triple digits is, surprisingly, not quantum but AI. What took Coherent up 189 percent was optical demand from AI datacenters, and what took Tower up 329 percent is the same cycle. Quantum revenue at these companies is small enough that it does not get broken out in filings.

So this is as far as the data proves anything. Assets tied only to quantum expectations collapsed on one sentence, and assets whose revenue turns on a different cycle held up. The next question, which is where the quantum exposure at these companies turns into actual orders, is what this map is trying to answer.

Let me flag one exception up front. Not every pure play broke together. D-Wave is up 74 percent since CES and still sits 48 percent above its March 2025 high [3]. There was an acquisition this year that changed what the company is, and first-half bookings grew sharply. That bookings number has a few things worth picking apart, though. I get into it in the company card behind the paywall.

2. Why I am the one drawing this map

This is the sixth in the investment map series. I have drawn glass substrates, LEO satellites, submarine cable, defense sensors, and humanoids. This one is quantum.

This one is a different animal from the previous five. Those five were cycles where revenue was already turning. Glass substrate pilot lines were going in. Submarine cable had 13 billion dollars of orders on the table. Quantum is not like that. A good share of the companies on this map have no product and effectively no revenue.

Which is why nobody draws this map. It is hard.

I wonder if there is anyone other than PhotonCap who can cover this. Someone fluent not just in optics but in semiconductors, in quantum, and in metrology and packaging, all of it. How many people have designed a semiconductor, laid it out, actually built one, and then measured it. I mean someone who has done all of it. Quantum is the same story. On X, on Substack, across the various online communities, I have not seen anyone other than me covering quantum at anything above trader level.

There is a reason all five of those are needed. Take a quantum computer apart and you find a chip built on a semiconductor process, optics that get light into and out of that chip, equipment that cools it or holds it in vacuum, and metrology that reads out what is happening inside. And in the end it has to go through packaging to become a product. Someone who knows only one of those layers cannot see which of the other four is the bottleneck. And from an investment standpoint, the bottleneck is exactly where the money turns first.

For readers who want to invest in quantum but get stuck at what a qubit even is, I am starting from background. Sections 3 and 4 are for people seeing quantum for the first time, so if you already know this, skip to section 5.

3. Background one: qubits, errors, and logical qubits

The computers we use calculate with 0 or 1. Switch off is 0, switch on is 1. We call it a bit.

A qubit is a switch that can sit somewhere between on and off. It holds a probability of being 0 and a probability of being 1 at the same time, and settles into one of the two at the moment you read it. Physics calls that state superposition.

Every qubit you add doubles the state space you can represent. At 300 qubits that space exceeds the number of atoms in the universe. 2 to the 300th is about 2.04 times 10 to the 90th. There is a common misunderstanding here, though. You cannot read that whole space at once. The moment you measure, it collapses to a single answer. The real advantage comes from specific algorithms that use interference to raise the probability of the answer you want. Molecular bonding calculations and integer factorization are that kind of case.

The problem is that this superposition breaks very easily in practice.

A vibration, a temperature swing, a cosmic ray passing through, and the qubit loses its state. This is called decoherence. How you defend against it differs by platform. For superconducting and some spin qubits, cooling to millikelvin is the core of it. For trapped ions and neutral atoms, ultra-high vacuum and laser control are the core. Photonic runs at room temperature but lives or dies on optical loss and detection. Even with all of that, the best qubits today get it wrong somewhere between one time in a thousand and one time in ten thousand.

One in a thousand sounds fine? A single useful calculation involves billions of operations. The probability of never making a single error is the per-operation success probability of 999/1000 multiplied by itself as many times as there are operations. At a billion operations that is (999/1000) to the billionth power, which is essentially zero. At one error per thousand, the calculation turns to garbage long before an answer comes out.

This is where error correction enters.

The idea itself is simple. Bundle a set of unreliable qubits together and use them as one reliable qubit. The physically existing qubits are physical qubits, and the logical unit you build by bundling many of them is a logical qubit. They keep checking each other so the wrong one can be found and fixed.

The cost is large. It depends on the scheme and on qubit quality, but one logical qubit takes hundreds to thousands of physical qubits. So when you see an announcement that says a company built a 1,000 qubit machine, the first thing to check is whether that is physical or logical. It is about three orders of magnitude of difference.

Here is the part that matters to an investor. To increase logical qubits you have to multiply physical qubits, and when physical qubits go up, the equipment that cools or confines them, the wiring that connects them, the control channels that drive them, and the detectors that read them all go up together. Add one qubit and that much hardware really does get added alongside it. So the moment error correction genuinely starts working, money flows to the objects around the qubit.

There is a further condition: that checking has to happen in real time. You have to find and fix the error before the qubit loses its state, so the verdict has to land faster than the error correction cycle of that platform. That verdict is called decoding, and it is a classical computer’s job. How much time you get differs by platform. Superconducting is already in the regime where microseconds are the constraint, while trapped ion readout itself runs hundreds of microseconds to milliseconds, so the margin is different. Either way it means a fast enough classical computer has to sit next to the quantum computer.

From bits to qubits to logical qubits, and the cooling, wiring, control channels, detectors, and decoding load that multiply alongside one added logical qubit

4. Background two: five platform branches

The camps split on what you make the qubit out of. Based on what is investable in listed markets and what current government programs fund, this piece centers on five branches.

Superconducting. Run current through a circuit chilled to very low temperature to make a qubit. It is also the best known quantum platform. Google, IBM, and Rigetti are here. It reuses semiconductor processes so it is familiar to build, and gate speeds are fast. In exchange it has to be cooled to millikelvin, so a dilution refrigerator is mandatory, and every qubit needs wiring running into the fridge, so wiring saturates before qubit count does. What this camp can and cannot do is something I worked through using Google’s case in Google Quantum AI says it can break Bitcoin in 9 minutes. Should we actually worry?.

Trapped ion. Levitate ions in a vacuum chamber with electromagnetic fields and manipulate them with lasers. IonQ, the bellwether among the quantum pure plays, is here, and so is Quantinuum. Gate fidelity is high and there is no device-to-device variation, because you are using atoms that nature made. In exchange it is slow. And there is a limit on how many ions fit in one trap, so scaling means splitting into modules. You read an ion’s current state by counting the faint fluorescence it emits, and the photomultiplier tubes used for that come back in section 13.

Neutral atom. Pick up chargeless atoms one at a time with optical tweezers made of laser beams and arrange them. QuEra, Atom Computing, and Infleqtion are here, and Auratomic, which took Bezos money, is here too. The strength is that the array can be reconfigured freely. And this approach is structurally a pile of lasers. Holding the qubit, manipulating it, and reading it are all done with light. Infleqtion built revenue in quantum sensing before quantum computing, and I covered that structure in Infleqtion and Quantum Sensing: The Quantum Reality That Arrived Before Computing. It was private then and is now listed as INFQ.

Photonic. Use particles of light themselves as the qubit. Among listed companies there is Xanadu, and among private ones PsiQuantum. The strengths are that it runs at room temperature and can use silicon photonics processes directly. In exchange photons are easy to lose. Loss is the biggest wall for this camp. I went into this in detail on Xanadu in The 1% Loss Wall (Part 1): Where Xanadu ($XNDU) Really Stands, Through the Lens of Three Nature Papers.

Silicon spin. Use the spin direction of a single electron inside silicon as the qubit. Diraq, Quantum Motion, EeroQ, and Silicon Quantum Computing are here. The biggest strength is that it can use existing semiconductor fabs almost as they are. Qubit counts are still low. And in this camp, Diraq and Quantum Motion and EeroQ all use the same process at the same foundry. Which one comes out in section 8.

Five platforms compared, representative companies and the physical condition each camp has to defend against

Five branches, five different physics. So nobody knows who wins right now. If someone says it with confidence, here is something I can say with confidence. That someone does not know.

This is the core difficulty of quantum investing. If you cannot pick the winning modality, how do you pick a stock.

5. The farther you go, the more light wins

Half the answer is already in an earlier piece. My subscribers already know this one, in fact.

Whichever of the five branches you take, meaningfully increasing qubit count means going past the limit of one chip, one fridge, or one trap. Which means splitting into modules, and then reconnecting the modules you split.

What you connect them with is decided by distance. Modules close together inside a fridge can be linked with superconducting wiring and microwave, and there are demonstrations of cryogenic microwave links between separate fridges. But the moment distance grows and the path has to cross a room-temperature section, microwave gets sharply worse. At the scale of rack to rack and room to room, conversion into telecom-band light is the best-validated scaling path so far. More precisely than saying light is the only answer: as distance grows, the scalability of optical conversion wins.

Trapped ion links modules with photons. Neutral atom does the same. For the photonic camp light is the qubit to begin with, so there is nothing to argue. Even in superconducting systems, conversion from microwave to photons for long-distance transport is under active research.

I took this argument apart layer by layer in Superconducting or Trapped Ion, They All Buy the Same Light. Four layers, source and conversion and transport and routing and detection, down to who sits in each.

The conclusion there was to stop waiting for a modality winner and look at the component and foundry layer first. Here is what happened in the three months since.

In April 2026, IonQ announced it had linked two independent commercial trapped ion systems through photonic entanglement. By the company’s own account it is the first case of commercial quantum computers connected by photons [4]. And on July 31, 2026, IonQ closed its 1.8 billion dollar acquisition of SkyWater Technology [5][19]. A trapped ion quantum company bought an entire US semiconductor foundry. In January it also acquired Skyloom, a free-space optical communications company. And Lightsynq, which does quantum memory and photonic interconnect for linking quantum processors, closed earlier than either of those, in June 2025. More than 20 patents and applications covering quantum memory and interconnect came with it [37].

Why would a trapped ion company buy a foundry, an optical comms company, and a quantum memory and interconnect company. Its own qubits are ions floating in vacuum.

Because once you split into modules you have to reconnect them, reconnecting takes light, and someone has to build the chips that handle light. Hand that to someone else and someone else holds the bottleneck.

There is one more layer on top of this. Look at the 11 companies that made Stage 2 of DARPA QBI and you find Photonic Inc, which DARPA itself describes as “optically-linked silicon spin qubits” [6]. Even the silicon spin camp has optical linking baked into its architectural premise.

6. The government attached conditions

On May 21, 2026, the US Commerce Department announced it had signed letters of intent with nine quantum companies totaling 2.013 billion dollars [7]. One billion to IBM, 375 million to GlobalFoundries. Atom Computing, D-Wave, Infleqtion, PsiQuantum, and Quantinuum get 100 million each, Rigetti up to 100 million, and Diraq up to 38 million, the one name at a different order of magnitude.

The conditions matter more than the amounts. The announcement says this. “The Department will take a minority, non-controlling equity stake in each company as a condition of receiving funds, in order to increase returns for the American taxpayer.”

Give funding, take equity. That is an investment structure.

The equity-stake condition sentence in the US Commerce announcement

What Commerce announced is a letter of intent, and the equity is a condition of receiving funds. The announcement contains no confirmation that equity has been taken in all nine. Only one company has disclosed a percentage: GlobalFoundries, which put about 1 percent in its own release [21]. The other eight have to be confirmed at definitive agreements in the fourth quarter. Commerce is aiming to close them then.

Then, right before I filed this, the first three closed. On September 8, 2026, D-Wave and Rigetti and Quantinuum each signed a definitive agreement worth 100 million dollars [38][39][40]. Three deals, 300 million dollars, and Commerce takes a minority non-controlling equity stake in each as a condition of the funding. This is the first case of a letter of intent turning into actual money and an actual stake.

The stated uses differ. D-Wave said it will put the money toward a 100,000-qubit annealing system and a 10,000-qubit gate model system designed to carry 100 logical qubits through more than a million operations [38]. Rigetti committed to three things: compressing readout electronics into a miniaturized package, expanding cryogenic capacity through a new cryostat architecture, and building fabrication capability for high-connectivity chip architectures [39]. Neither is about the qubit itself. Both are about what surrounds the qubit. Section 3 said money flows to the objects around the qubit once error correction starts working. This time that move happened through government funding rather than acquisitions.

Quantinuum’s award names two more companies. GlobalFoundries is named as the fab that will make its next-generation ion traps and control electronics on 300mm wafers, and Monarch Quantum is named for the lasers and optical components [40]. The first name comes back behind the paywall. The second one is on the watchlist.

The other six are still pending. So the verification of this axis is only half done.

Even so, I think a government putting money down with equity as the condition is a hard floor. But the floor sits under financing, not under the share price. That is the answer to the question of why you should not exclude companies with no product and no revenue. The US has no option to give up quantum in its technology competition with China, so capital keeps coming in.

What that floor actually buys you has to be kept straight, though. A government stake raises financing availability and signals government technical validation. It does not guarantee continued listing, or solvency, or the absence of further dilution. Do not blur that distinction.

And this is not just Commerce. In June 2026 the Department of Energy launched Quantum Genesis with a stated goal of building and deploying a scientifically useful fault-tolerant quantum computer by 2028 [8]. DARPA QBI is running 11 companies in Stage 2 with Stage 3 validation to follow. The National Quantum Initiative Reauthorization Act (H.R. 8462) passed the House Science Committee in April 2026 and went to the floor [33]. It is not law yet.

The other side is symmetric. China’s 15th Five-Year Plan put quantum near the top of its future industries list. Origin Quantum raised close to 3 billion yuan in a July 2026 pre-IPO round led by China North Industries Group [18]. A state-owned defense conglomerate put money into a quantum computing company. I covered this shape in The Real Quantum Race: NVIDIA, NQI, and China’s State Capital.

Overlay the nine Commerce names with the 11 DARPA QBI names and something interesting falls out. Only four companies appear on both lists. Which four, and what that means for scoring, is behind the paywall.

7. Three frameworks

That is as far as the picture goes on public information alone.

The market looks at quantum through ordinary financial metrics. Revenue, growth, valuation. In an industry where revenue is effectively zero, those metrics distinguish nothing. You cannot line up five companies with zero revenue and compare P/E ratios.

So this cycle needs its own scorecard. Behind the paywall I apply three frameworks to all 23 companies at once.

The GVM score. Five axes, one to five points each. The axes are government anchor, modality neutrality, physical bottleneck, revenue substance, and convergence proximity. That last axis is this map’s signature, and it measures how tightly a company is attached to the two convergence axes, optical interconnect and real-time decoding. This axis is graded on evidence tier. Having your name in a press release, giving a supporting quote, and actually shipping parts or running a joint demonstration do not get the same score. No primary source, no points.

Phase weighting. Adding the five axes straight up fails to reflect the current phase. In a stretch where revenue does not mean much yet, the revenue substance axis has to weigh less and government anchor, modality neutrality, and convergence proximity have to weigh more. Apply the weights and the ranking changes.

The catalyst calendar. Trigger events scheduled in the next 12 months, placed on a time axis. Each row is marked as a fixed government date, a company-stated target, or my estimate.

And quantum purity. This is the part of this piece I put the most work into. The obvious objection is that Google and IBM and a foundry are not quantum companies, so scoring them on quantum axes alone cannot be right. So I measured how much of a quantum stock each name actually is, two different ways, and published it as its own column. There is plenty of writing that hand-waves at quantum exposure. I have yet to see anyone measure it as a number and attach it name by name. Overlay this column on the ranking table and investment character splits four ways, and it explains why the two hidden cards come out of the quadrant they come out of.

Quantum value chain positioning matrix, vertical axis government capital anchor strength, horizontal axis modality neutrality, company names withheld

Behind the paywall: all 23 companies identified, the foundry and component supply wiring diagram, market cap and returns by tier, the five-axis scorecard, the weighted ranking, group mapping, two hidden cards, the 12-month calendar, scenarios, and monitoring points. Five ownership changes in 2026 alone reshaped this map. One of them is an event where the listed exposure to the hardest physical bottleneck in this industry disappeared entirely, and another happened in late August and actually changed the rankings in this piece.

If you’d like to support my independent research and creative journey, please consider a “pledge.” Your support keeps the photons moving.

Share this with anyone who needs to see the world through a different wavelength.

Share

8. 23 companies and who works with whom

Quantum value chain five-tier cascade, representative companies and the physical bottleneck at each tier

Tier 1, materials and substrates (3 companies). Linde (LIN), Anglo American (AAL.L), AXT (AXTI). Helium-3 and cryogenic gases, Element Six CVD diamond, InP and GaAs compound substrates. What AXT holds I covered separately in AXT Inc. (AXTI) Deep Dive: The Hidden Bottleneck in AI Optical Interconnect.

Tier 2, cryogenics and vacuum (2 companies plus 1 straddler). ULVAC (6728.T), Atlas Copco (ATCO-A.ST), and FormFactor (FORM), which straddles Tier 2 and Tier 3.

There is a place in this tier where I got it wrong and then fixed it, and I am leaving it in.

The physical bottleneck that stops the superconducting and silicon spin camps from moving at all is the dilution refrigerator. The equipment that cools qubits to near absolute zero. There are several commercial suppliers for large system units, Bluefors and Oxford Instruments NanoScience and Leiden Cryogenics and Cryoconcept among them, and almost that entire list is private. Bluefors is private in Finland, and when Oxford Instruments handed its NanoScience division to Quantum Design on January 2, 2026, the one listed name still in this market dropped out [31]. The sale was announced in June 2025 with a headline consideration of 60 million pounds [36], and net proceeds at completion were 48.5 million pounds [31]. Quantum Design, the buyer, is also private.

Oxford Instruments’ announcement of the NanoScience divestment

Read only that far and it looks like the way to buy this bottleneck in listed markets has vanished. That is how I wrote it, too.

But it had not. It was already inside this map.

FormFactor acquired the JanisULT dilution refrigerator product line for 3.4 million dollars in June 2022 [34]. In the company’s own words it is the largest commercial supplier of dilution refrigerators in the United States, with a product range spanning sub-10 millikelvin to 77 kelvin. And in March 2026 it launched Flatiron, a new benchtop dilution refrigerator. Base temperature around 30 millikelvin, and a horizontal layout instead of the traditional vertical chandelier structure, which improves lab access [35].

FormFactor Quantum Cryogenics product page, dilution refrigerator product line

So the accurate sentence is this. The suppliers of dilution refrigerators for large QPU systems are now effectively all private, but listed exposure did not disappear. It sits inside FormFactor, and nobody calls FormFactor a dilution refrigerator company.

This is the last thing I found on this map. I had classified FormFactor as a cryogenic prober company and left Tier 2 with two names in it. I fixed it to a straddle and rewrote the company card.

One distinction has to hold. Flatiron is for benchtop validation, not a large system fridge running a 1,000 qubit machine. Bluefors is still large in that slot, which is why I put FormFactor’s bottleneck score at 4. Strong at the device validation stage, not yet at the system stage, is the accurate position.

Tier 3, control and readout and decode (2 companies plus 1 straddler). NVIDIA (NVDA), Keysight (KEYS), and FormFactor reaches here too. This is where the real-time decoding from section 3 sits. Which slice of optical test Keysight and FormFactor each hold is laid out in The Three Pillars of SiPh Wafer Test: What AEHR, FORM, and KEYS Actually Do.

Tier 4, photonics and detection and foundry (6 companies). GlobalFoundries (GFS), Coherent (COHR), Hamamatsu Photonics (6965.T), MKS (MKSI), Novanta (NOVT), Tower Semiconductor (TSEM). How far these foundries have come on 300mm silicon photonics is in OFC 2026 Part 1: 300mm SiPh Foundry: Who Is Actually Ready? and Everyone Saw a Laser Shortage. The Money Went to the Foundries First..

Tier 5, QPU and platform (9 companies). IonQ (IONQ), Quantinuum (QNT), Rigetti (RGTI), D-Wave (QBTS), Xanadu (XNDU), Infleqtion (INFQ), Quantum Computing Inc (QUBT), IBM, Alphabet (GOOGL).

I took Honeywell out. After the Quantinuum IPO, Honeywell deconsolidated Quantinuum and moved its roughly 48 percent voting interest to equity method accounting, recognizing a 6.629 billion dollar one-time gain in the second quarter [32]. Put HON and QNT side by side and you are counting the same asset twice. Treat indirect exposure through Honeywell as equivalent to buying 48 percent of QNT.

23 companies identified / company, ticker, country, tier, quantum positioning, market cap, 1-year return

The market cap spread is wide. NVIDIA is 5.555 trillion dollars and Quantum Computing Inc is 1.8 billion. More than 3,000 times apart, and they are caught in the same cycle.

2026 was also the year quantum companies listed in bulk. Infleqtion went public on the NYSE via SPAC on February 17 (over 550 million dollars), Xanadu on Nasdaq and Toronto via SPAC on March 27 (about 302 million dollars), and Quantinuum began trading on Nasdaq on June 4 (60 dollars a share, offering closed June 5) [10][11][12]. The first listed neutral atom company, the first listed photonic company, and the largest listed quantum company after IonQ all appeared in one year.

Those three have no one-year history, so they cannot be compared on a one-year return. Against reference prices, Quantinuum is minus 17.3 percent against its 60 dollar offering price, and Xanadu and Infleqtion are minus 2.8 percent and plus 28.5 percent respectively, assuming a 10 dollar SPAC trust reference. The trust reference assumes the SPAC standard value and needs to be confirmed against the merger filings.

8.1 Whose chips get made where

Trading relationships tell you more than the tier split does. Follow where quantum companies get their chips made and the wiring diagram of this industry appears.

Quantum collaboration wiring diagram, from five camps to three foundries

Start with the photonic camp. PsiQuantum fabs at GlobalFoundries. It is a deep 300mm silicon photonics partnership and the relationship is spelled out in the company’s announcement [21]. Xanadu, in the same photonic camp, fabs at Tower Semiconductor. They co-designed a process that puts a Xanadu-specific material stack on top of Tower’s silicon photonics platform, they are optimizing ultra-low-loss silicon nitride and integrated photodiodes, and they have run multiple joint tapeouts [22].

Tower Semiconductor and Xanadu’s expanded collaboration, the joint tapeout and material stack language

The silicon spin camp is all one place. Diraq and Quantum Motion and EeroQ all use GlobalFoundries’ FDX cryogenic CMOS process [21]. I planted the setup in section 4 that these three share a foundry and a process, and this is it.

The partner paragraph in GlobalFoundries’ Quantum Technology Solutions announcement

Trapped ion splits. Quantinuum is at GlobalFoundries, taking cryogenic CMOS plus integrated photonics and advanced packaging [21], while IonQ bought its own foundry. SkyWater, closed on July 31 [5][19].

For superconducting, IBM is building its own fab. That is what the billion dollars from Commerce is for [7].

Count that up and the answer falls out. Five camps fighting with five different physics, and the places that fab the chips narrow to three, with five of those companies concentrated at one of them.

One distinction has to be explicit. Google Quantum AI, Microsoft Quantum, and NVIDIA also appear in the GlobalFoundries announcement. But those are supporting quotes about US quantum manufacturing infrastructure [21]. That is as far as the confirmation goes, and there is no material showing wafers actually being run. This distinction is what actually separates scores in section 9. Treat being named in a press release the same as co-developing a process and the table means nothing.

Names repeat on the component side too, not just chips.

Dilution refrigerators, as covered above, come from several suppliers for large systems, Bluefors and Quantum Design and Leiden among them, and that list is effectively all private, while the device validation product line sits inside FormFactor. Every superconducting and silicon spin camp is immobile without this equipment, and listed exposure narrows to FormFactor alone.

Ion fluorescence readout uses Hamamatsu photomultiplier tubes. Whether an ion is 0 or 1 is read by counting that faint light [23]. Lasers and optical components come from Coherent and MKS and Novanta, selling across the trapped ion, neutral atom, and photonic camps. Real-time control and decoding has 17 hardware builders and 9 national labs on top of NVIDIA’s NVQLink [9], and the controller makers underneath it, Quantum Machines and QBLOX and QubiC and Zurich Instruments, are all private. Metrology is Keysight, which goes into every camp without discrimination.

Overlay the two wiring diagrams and the skeleton of this map shows up. The modalities fight along five branches, but the supplier list underneath keeps repeating the same names, and a good share of them are not listed at all. The listed ones are mostly companies nobody calls quantum.

9. The five axes

The market looks at quantum through revenue and growth and valuation. In an industry where revenue is effectively zero, those metrics distinguish nothing. You cannot line up five companies with zero revenue and compare P/E ratios.

So I built axes specific to this cycle. Five of them, and I will call each by one word.

Government. Has government capital gone in directly. A 5 is a company in both programs, or one where the government put money down with equity as the condition. A 4 is selection into one. A 1 is no contact at all.

The DARPA QBI Stage B list of 11 companies with each qubit modality

Neutrality. Do they sell no matter which camp wins. A 5 is infrastructure you have to buy regardless of camp. A 3 is a company spanning two or more. A 1 is a single-modality QPU company. Not being able to pick the modality winner is the hard problem of this industry, and this axis is where that problem gets routed around.

Bottleneck. Do they hold a position that cannot be substituted. A 5 is a near-monopoly that is physically irreplaceable. A 4 is an oligopoly of two or three. A 1 is a company with many alternative suppliers.

Revenue. Is cash turning right now. A 5 is large commercial revenue, which makes quantum an option on top. A 3 is small but turning. A 1 is effectively zero.

Convergence. How attached are they to the two axes from section 5, optical interconnect and real-time decoding. This is the map’s signature axis, and the score is split by strength of evidence. In one line: no primary source showing a named supply relationship or a joint demonstration, no score above 4.

There is a reason for that threshold. If I turn my own thesis that optics is the convergence axis into an axis, then crown the company that scores high on it, the conclusion was already inside the premise. Raising the threshold cuts the companies favorable to my thesis first. It actually did: Coherent went from 5 to 3, Rigetti from 4 to 2, and the leading candidate for first place in this piece went from 5 to 4. There is residual bias even so, which is why section 10.1 also runs the ranking with the convergence axis removed entirely.

9.1 How the convergence scores split

Convergence axis evidence ladder, 23 companies placed

Only Quantinuum gets a 5. On the optical side it runs integrated photonics with GlobalFoundries [21], and on the decoding side it demonstrated real-time correlated decoding on Helios with an NVIDIA GPU-accelerated decoder [28]. It is the only company that filled both axes with primary sources. And the fact that it came from a QPU company rather than a supplier is the most surprising position on this ladder.

Five get a 4. GlobalFoundries (PsiQuantum 300mm silicon photonics), NVIDIA (NVQLink and the Quantinuum demonstration), Tower Semiconductor (joint tapeouts with Xanadu), IonQ (photonic entanglement between two commercial systems), Xanadu (photons are the qubit itself, plus joint tapeouts with Tower). All of them are confirmed on one axis with the other axis empty, which is why none reached 5.

Four get a 3. Coherent, Hamamatsu, IBM, Quantum Computing Inc. Their products sit on both axes but no named quantum customer has been disclosed. Ten get a 2. Keysight, FormFactor, MKS, Novanta, Anglo American, AXT, Alphabet, Rigetti, Infleqtion, D-Wave. Three get a 1. Linde, ULVAC, Atlas Copco.

Let me explain three of the boundary calls.

Rigetti dropping from 4 to 2 is the largest downgrade in this scoring pass. Its name is on the NVQLink partner list, true, but that is participation, not a joint demonstration. There is no material showing decoding actually run on its own system the way Quantinuum did.

D-Wave also goes 3 to 2. Dual-rail hardware error detection is excellent technology, but what this axis measures is real-time decoding where a classical computer renders a verdict inside microseconds. Catching errors inside the hardware is a different layer.

Infleqtion also goes 3 to 2. It is true that neutral atom is structurally a pile of lasers, but that is light controlling its own atoms, not optical interconnect linking QPU to QPU. Skip that distinction and the axis widens into “do they use optics,” which makes it meaningless.

10. Rankings

Linde and Atlas Copco have large revenue, so they automatically take a 5 on the revenue axis, and that revenue has nothing to do with quantum. So the weights get set to match the current phase. Government, neutrality, and convergence at 1.5x, bottleneck at 1x, revenue at 0.5x. The weights sum to 6.0, so a perfect score is 30 points, and I will call this the weighted score.

23 companies by weighted rank, with the five axis scores alongside

10.1 Does the ranking change if you remove the convergence axis

Ranking with the convergence axis included versus excluded, with movement lines

However hard you set the evidence threshold, the question of why those two axes and not others remains. The only way to answer it is to take the axis out and count again.

Set the convergence weight to zero and a perfect score becomes 22.5. Counted that way, GlobalFoundries is still first at 21.5, and the gap to second is still 3 points. Most of this company’s score already comes from a 5 on government, a 5 on neutrality, and a 5 on revenue. Convergence was stacked on top of that. The result of this check is that removing my signature axis entirely leaves the conclusion unchanged.

Below that, positions move. NVIDIA and IBM tie for second, Keysight fourth, Coherent fifth, Linde sixth. This is where the names that lean hard on the convergence axis show themselves. Tower goes from sixth to ninth and Xanadu falls further. In the other direction, Linde climbs from 11th to 6th and Keysight from 6th to 4th. If you do not buy the optical interconnect thesis, this is the ranking to read.

10.2 But what does this score have to do with the share price

Read this far and an obvious objection comes up. Google and IBM and a foundry are not quantum companies, so is scoring them on quantum axes alone the right thing to do in an investment map. GlobalFoundries alone has legacy foundry and silicon photonics growth running separately.

It is a fair point, which is why quantum weighting is not in the score. Put it in and you end up docking points for being diversified, when that diversification is exactly why these companies do not die when the quantum timeline slips. It should be read as a strength, if anything.

Instead I am publishing it as its own column. And rather than assigning it by judgment, it can be measured. I measured it twice.

Measurement 1. Jensen shock sensitivity

This is the ruler I flagged in section 1. The one-day move on January 8, 2025. It was the largest quantum sentiment shock on record, and the broad market barely moved that day. SMH, the semiconductor ETF, was down 0.7 percent.

Jensen shock sensitivity, single-session moves on January 8, 2025, and the empty middle

The most important thing in this table is the gap.

Everything between minus 36.1 percent and minus 6.6 percent is empty. A 29.5 percentage point hole. The market sorted these 23 names into four quantum stocks and everything else, and built nothing in between.

Look at IBM. On both the Commerce list and the DARPA list, building a billion-dollar quantum fab, and it bought HRL in August. Third on this map. And it moved 0.3 percent on the largest quantum shock ever recorded.

What you call this matters. It is a single-day event sensitivity, not a beta. Building a proper abnormal return would mean estimating a market model over a prior window and subtracting expected return, and I did not go that far. So I am calling it Jensen shock sensitivity.

Measurement 2. The 64-day quantum factor beta

That table has only twenty names. Quantinuum, Xanadu, and Infleqtion are missing, because all three were private in January 2025. Which happens to leave holes at fifth and joint thirteenth on this map. So I measured a second way.

I used the 64 trading sessions from June 5, 2026, when Quantinuum listed, through September 4. Three steps.

First, turn how much the quantum sector moved that day into a single number. It is the average daily return of the four pure plays, with the portion lifted by the broad semiconductor market stripped out using SMH.

Then, regress each name on the quantum factor and SMH at the same time. Control for the common market move, then pull out the quantum factor coefficient alone. That coefficient is the quantum beta. At 1.0, when quantum moves 1 percent this name moves 1 percent. At 0, it does not budge.

There is a trap here worth flagging. The quantum factor just built is the average of the four pure plays. Which means one quarter of that factor is IonQ. Measure IonQ’s beta against that factor and you are partly regressing IonQ on itself. A day when IonQ alone jumped sits on the left side of the equation and inside the right side too, so the beta has to come out higher than the true sector sensitivity. Measure yourself against a factor containing yourself and it biases toward you.

So for each of the four pure plays I rebuilt the factor from the other three and measured against that. Measuring IonQ uses a factor built only from Rigetti, D-Wave, and Quantum Computing Inc. Doing so took IonQ from 0.94 to 0.86, Rigetti from 0.97 to 0.93, and Quantum Computing Inc from 0.88 to 0.79. Self-inclusion really was inflating them, by 0.04 to 0.09 points each.

Quantum beta across 23 companies, with beta, t statistic, and R squared

Only the nine names whose t statistic exceeds 1.96 in absolute value, at the 5 percent significance level, respond to quantum in a statistical sense. The other fourteen are indistinguishable from zero. The line you could see by eye in section 1 gets drawn here as a number.

The three that were missing are filled in now, and once filled in, the three are different from each other.

Infleqtion is at 1.11. That is pure play territory. The market treats this company as the same object as Rigetti or IonQ. And it ranks joint thirteenth on this map. Mid-table score, pure play sentiment sensitivity.

Quantinuum and Xanadu are at 0.45 and 0.43. Both significant, but R squared is low at 0.32 and 0.46. They listed recently, so the sample is short and float is thin, which means the current beta is hard to treat as representative of these companies’ long-run sector sensitivity. They need remeasuring once sample accumulates.

Coherent is the only significant name in the component layer. 0.27 with a t of 2.1. Every other foundry, metrology, and cryogenics name is indistinguishable from zero.

Where the two measurements diverge

The two measurements point the same way on the four pure plays. I am not going to use that as independent validation, though. The pure plays are the raw material of the quantum factor to begin with, so of course they point the same way.

Two names diverge in the component layer.

AXT was down 4.4 percent that day, but its beta is high at 0.74. I do not read that as quantum exposure. AXT is up 1,914 percent over the past year, a momentum name where the same money moves in and out in the same direction as the quantum pure plays. What the regression caught is more likely shared flow than quantum exposure. R squared is low at 0.48, too.

FormFactor is the opposite. It fell the most of any component name that day, minus 6.6 percent, while its beta is 0.06, effectively zero. Meaning it got classified as quantum when the shock came, but does not move with quantum the rest of the time. And as section 8 showed, that is true even though the company actually sells dilution refrigerators.

The two measurements measure different things. The event measure asks whether the market classifies this as quantum when a shock arrives. The beta asks whether it moves with quantum day to day. That some names diverge is itself information.

The limits, stated. The event measure has one observation, so other news could have overlapped that day. The beta has a short 64-day sample and can mistake flow for quantum, as with AXT just now. And Hamamatsu, ULVAC, Atlas Copco, and Anglo American trade in different session hours from the US, so I lagged them by a day, and this kind of non-synchronous trading can bias beta downward. So these two columns are not used to change the ranking, only to read it.

Overlaying the ranking and the sensitivity

Weighted score versus quantum beta, 2x2, character by quadrant

Let me be precise about what these two axes measure. The weighted score measures strategic position in the quantum supply chain. The quantum beta measures how sensitive the current share price is to quantum sentiment.

There is one reading that must not happen. A low beta does not mean the company’s quantum value is absent from the share price. GlobalFoundries having a beta of 0.12 does not support the conclusion that the market values its quantum business at zero. Beta comes out low when some of it is already in the valuation and the company as a whole is simply large. Saying whether something is undervalued requires a third axis, valuation, and this piece does not have one.

Holding that line, the quadrants read like this.

High score, low sensitivity. GlobalFoundries, NVIDIA, IBM, Coherent, Tower, Keysight, Hamamatsu. FormFactor sits just under this box’s boundary at a weighted score of 18.5, with the same character. Top-tier position in the quantum supply chain, share prices that do not move on quantum news. That is as far as the data proves, and I think the two hidden cards are here. The downside is that the company does not die if quantum does not arrive. The upside is that this sensitivity rises once quantum revenue actually starts being recognized, and that second part is still a hypothesis.

High score, high sensitivity. IonQ, with Quantinuum on the border. This is the direct bet on quantum, and it pays the most if it is right. In exchange it takes 40 percent off at the next Jensen moment.

Low score, high sensitivity. Rigetti, Quantum Computing Inc, D-Wave, and Infleqtion. Pure play volatility with five-axis scores that do not keep up. Wide range in both directions with thin grounds to justify that range. Infleqtion landing here is what this measurement pass turned up. Revenue grew 157 percent at that company and the market is still putting it in the same room as Rigetti.

Low score, low sensitivity. Linde, ULVAC, Atlas Copco, Alphabet, Novanta, Anglo American. Not names to buy because of quantum. They are on this map to show the structure of the supply chain, not because they are buy candidates.

One more thing. On January 8, 2025, the Defiance Quantum ETF was down 4.6 percent. A quantum ETF falling only 4.6 percent on a day when pure plays fell 40 percent means, as section 1 showed, that most of what is inside that ETF sits outside the pure plays.

11. The lag between tiers, and what I have not confirmed

In the earlier investment maps there was a 6 to 12 month lag between tiers. Materials sold first, equipment followed, and manufacturers recognized revenue last.

In quantum that lag cannot be measured yet. Because the companies in Tiers 1 through 4 do not break out quantum revenue. Coherent and Hamamatsu and Keysight are selling into quantum labs, true, but there is no way to know what that amounts to as a share of the whole company.

Here is as far as current data goes. The total company revenue of Tiers 1 through 4 and the quantum expectations of Tier 5 are priced in differently. What pulled Tier 1 through 4 share prices was the AI optical and semiconductor cycle. What pulled Tier 5 was quantum expectations. Which is why one sentence from Jensen Huang took 45 percent off only one side.

I cannot go as far as concluding there is no lag between tiers. I will leave it as a hypothesis that gets tested once quantum orders start being disclosed separately. When that separation starts is one of the things to watch in the section 12 calendar.

12. Catalyst calendar: the next 12 months

12-month trigger event timeline, color coded by certainty

The heaviest item is the Q4 conversion of the Commerce letters. And it started ahead of schedule. Three closed on September 8, so six are left. Testing the government axis scores is already half underway.

13. Group mapping and hidden cards

Positioning scatter by group, vertical axis weighted score, horizontal axis 1-year return, tickers labeled

Group A, picks and shovels already priced (6 companies). Coherent, Tower Semiconductor, FormFactor, MKS, Keysight, NVIDIA. High weighted scores and one-year returns in the double to triple digits. The market has already put a price on these.

Group B, the government attached conditions but the share price has not moved (5 companies). Rigetti, D-Wave, Infleqtion, Quantinuum, IBM. Four of them have one-year or reference-price returns that are negative or in single digits. The asymmetry in this group runs both ways, though. Government capital does raise financing availability, but it does not manufacture share price appreciation. What you are buying here is the probability that funding does not dry up. The probability of success is a separate thing.

September 8 was the test of that sentence. On the day the three definitive agreements were signed, D-Wave rose 6.6 percent, Rigetti 4.0 percent and Quantinuum 1.7 percent. The semiconductor ETF was up 1.2 percent the same day, so the excess was 5.4, 2.8 and 0.5 percentage points. Then they gave it back. Measured at the September 9 close against September 4, D-Wave is up 3.3 percent, Rigetti up 0.3 percent, and Quantinuum down 1.5 percent [3]. On the day 100 million dollars was actually committed and the government actually took a stake, these stocks moved single digits and handed more than half of it back within two sessions. It means the May letters of intent were already priced, and the definitive agreement was not a second event. The sentence from section 6, that government capital is a floor under financing rather than a lift under the share price, gets confirmed here as a number.

Look at what else rose that day. Infleqtion was up 6.3 percent, and it is not one of the three [3]. It received nothing, the sector moved, and it moved along with it. That is what the 1.11 beta from section 10.2 looks like in practice.

Group C, the modality-neutral unrecognized layer (7 companies). GlobalFoundries, Hamamatsu, Linde, ULVAC, Atlas Copco, Anglo American, Novanta. Companies that sell no matter which camp wins and do not get classified as quantum stocks. The two hidden cards are here.

Group D, the vertical integration bet (5 companies). IonQ, Xanadu, Alphabet, Quantum Computing Inc, AXT. The side that intends to own part or all of the stack itself.

AXT needs a footnote even inside this group. What produced that 1,914 percent one-year number is the InP substrate cycle. The actual driver is in The More Tents Meta Pitches, the Faster InP Runs Dry and AXT and Lumentum’s Prepayment Deal, which is why convergence is a 2. InP is upstream material for the light source layer, not a component sitting on either convergence axis. I also excluded this name from the component layer averages earlier.

Hidden card 1: GlobalFoundries (GFS) / weighted score 27.5 / 1st

Market cap 24.8 billion dollars (as of 2026-09-04). 1-year +37.0%. Tier 4.

The 5 on government is the 375 million dollars from Commerce plus the equity condition. The company disclosed the stake at about 1 percent [21].

The 5 on neutrality is backed by a list. On the same day as the Commerce announcement, GlobalFoundries launched a separate business unit called Quantum Technology Solutions and disclosed its collaborators [21]. As the section 8 table showed, photonic (PsiQuantum), silicon spin (Diraq, Quantum Motion, EeroQ), and trapped ion (Quantinuum) are confirmed as actual process collaborations. Three of the five branches are in at wafer level. This is the only company on this map that can document modality neutrality to that degree.

Convergence is a 4. The optical axis is clear through PsiQuantum 300mm silicon photonics, but on the decoding axis the announcement carries only a supporting quote from NVIDIA and no confirmed integration project. I cut the company most favorable to my thesis first.

A line from CTO Gregg Bartlett’s announcement compresses the position. “Quantum is at an inflection point, and GF is building QPUs” [21]. A foundry defining itself as a QPU manufacturer. This company’s foundry position itself I have covered in Everyone Saw a Laser Shortage and Samsung’s Silicon Photonics Bet: Too Late Against TSMC, GFS and Tower?.

And yet the stock is up 37.0 percent over a year. While Coherent is up 211 percent and Tower 259 percent. Because the market sees GlobalFoundries as a legacy node foundry. Mature process, low growth, automotive and IoT customers. The identity of a quantum foundry has not yet shown up in this share price’s quantum sensitivity.

The counterargument is just as clear. 375 million dollars from Commerce is 1.5 percent of a 24.8 billion dollar company. It will take a very long time for quantum foundry revenue to move this company’s P&L. So this company’s quantum position runs well ahead of its current quantum revenue, and that position has not yet shown up in this share price’s quantum sensitivity. I am holding the line from section 10.2 here too, though. Low sensitivity does not mean undervalued. Saying that requires a valuation axis, and this piece does not have one. And the point where the sensitivity could rise is tied to IBM’s own foundry coming online, because once IBM is running, exclusivity shrinks.

One more piece of evidence attached itself here. In Quantinuum’s September 8 Commerce announcement, GlobalFoundries is named as the fab that will produce its next-generation ion traps and control electronics on 300mm wafers [40]. The relationship from section 8 is confirmed a second time, this time in a document with government money attached to it. Fabricating control electronics is a different job from building a real-time decoder, though, so the convergence score stays at 4.

The stock was down 0.3 percent that day [3]. Its own customer was up 1.7 percent. Strong position, quiet stock, the whole character of this card, showed up inside a single session.

The passage naming GlobalFoundries as the 300mm ion trap fab in Quantinuum’s Commerce definitive agreement announcement

Hidden card 2: Hamamatsu Photonics (6965.T) / weighted score 20.0 / joint 8th

Market cap 4.5 billion dollars. 1-year +58.0%. Tier 4.

The government axis is a 1. Not a dollar of US government money, because it is a Japanese company. And it still scores the same as IonQ and five points above Rigetti.

Trapped ion fluorescence readout is the most certain piece. The way you find out whether an ion is 0 or 1 is by counting the very faint light it emits, and photomultiplier tubes go in there. The company’s own technical materials present photon counting heads for ion fluorescence counting in exactly this use [23]. The neutral atom camp reads atom arrays through fluorescence imaging, which brings in scientific cameras, and the diamond NV center side uses fluorescence detectors too. Supply agreements with individual customers are not disclosed, though, so convergence is a 3.

The superconducting nanowire single-photon detectors used in photonic quantum computing are other companies’ territory. That is ID Quantique (now an IonQ subsidiary) and Single Quantum and the like. Hamamatsu’s modality neutrality comes from cutting across trapped ion, neutral atom, and NV, and does not extend to the photonic camp.

Convergence is a 3. The readout products are confirmed in company documents but no named quantum customer has been disclosed. Same yardstick, same cut.

The biggest counterargument is this. Buy Hamamatsu to buy quantum exposure and what you actually buy is the semiconductor inspection cycle. What drives this share price is semiconductor inspection and medical imaging. Most of Group C shares that character. That it was the only name that rose on January 8, 2025, means something on its own, though.

14. Company cards

IBM / weighted score 23.0 / 3rd (rank changed in August)

Market cap 221.3 billion dollars. 1-year -4.3%. Tier 5.

This name’s score changed while I was writing this piece. On August 26, 2026, IBM completed its acquisition of HRL Laboratories [25]. What HRL brought is silicon spin qubits, quantum sensing, quantum materials, cryogenics, control electronics, packaging, and interconnect technology. Deal value undisclosed.

The assets HRL brought are broad, reaching into interconnect and control electronics and packaging. But public materials alone cannot pin that down to optical interconnect or real-time QEC decoding. The accurate phrasing is that IBM acquired capability adjacent to the two axes this map measures, and acquired it as a whole company.

Two scores move. Neutrality goes from 2 to 3. It was a single superconducting camp, and silicon spin attached, so it now spans two or more. Convergence goes from 2 to 3. Adjacent capability acquired, but no supply relationship or joint demonstration with a named quantum customer, so no 4. The weighted score goes from 20.0 to 23.0, moving the rank from joint 7th to 3rd.

The 5 on government was already there. It took the largest Commerce award at a billion dollars and is in DARPA QBI, one of the four names on both lists. That billion is for a new quantum-grade superconducting wafer foundry, and once it runs, the bottleneck score goes up and it erodes GlobalFoundries’ edge.

The weakness is the absence of a price signal. 221.3 billion dollar market cap, minus 4.3 percent over a year. Quantum cannot move this share price. That does not mean you get the quantum for free, though. It could equally be that the exposure is small enough inside 221.3 billion dollars that no signal registers, and the two measurements in this piece cannot separate those two cases.

In 2026, four companies started buying the stack outside the QPU

Take IBM’s HRL acquisition on its own and it is just one M&A deal. Line up what happened this year and it reads differently.

TimingAcquirerTargetWhat was bought2026-01-20D-WaveQuantum CircuitsError-corrected gate model superconducting [24]2026-07-31IonQSkyWaterA US semiconductor foundry [5][19]2026-08-26IBMHRL LaboratoriesSilicon spin, cryogenics, control electronics, packaging, interconnect [25]H1 2026Quantum Computing IncNHanced and two othersSemiconductor packaging fab and photonics [26]

The four targets are different from each other. Gate model architecture, a foundry, spin and cryogenics and control, packaging and photonics. But the direction is the same. QPU companies are pulling manufacturing and packaging and interconnect and control in-house. Section 3 said that once error correction starts working, money flows to the objects around the qubit. The QPU companies are making that move on their own balance sheets right now.

I read this as the companies proving this map’s thesis on my behalf. It also explains why the right-hand column of the section 8 wiring diagram keeps shrinking. Every time a QPU company buys foundry and control assets, there are that many fewer independent suppliers left.

NVIDIA (NVDA) / weighted score 24.5 / 2nd

Market cap 5.555 trillion dollars. 1-year +35.1%. Tier 3.

NVIDIA is pushing hard to build an ecosystem in quantum right now, not just in GPUs and AI. So judging only by the pace of ecosystem expansion disclosed so far, I see CUDA-Q and NVQLink as the strongest candidate for the orchestration layer between classical and quantum. Doing this without NVIDIA infrastructure and without help from NVIDIA’s quantum team will not be easy. Even IonQ, for all its full-stack talk, is inside the NVIDIA ecosystem.

Three grounds: real-time decoding, partner breadth, and the equity portfolio.

The real-time decoding from section 3 is a GPU’s job. NVQLink was announced at GTC Washington in October 2025 [9] and extended at GTC 2026 with the cudaq-realtime API [29]. CUDA-Q QEC 0.6 contains two real-time decoders: a RelayBP belief propagation decoder for qLDPC and a neural network predecoder for surface code [30]. On a Dell host server configuration, round-trip latency under 4 microseconds has been validated [29].

Partner breadth is wide. Seventeen hardware builders and nine national labs are attached to NVQLink [9]. On the controller side, QBLOX, QubiC, and Zurich Instruments. On the hardware side, Rigetti, SEEQC, and Silicon Quantum Computing. Quandela, which builds photonic QPUs, has also published low-latency integration work.

There are equity stakes too. NVentures went into QuEra (neutral atom), Quantinuum (trapped ion), and PsiQuantum (photonic) starting in the first half of 2025 [17]. Three modalities covered at once. And since Quantinuum listed in the meantime, NVIDIA is also a shareholder in a listed quantum company. The same pattern of betting on multiple suppliers simultaneously in optical communications is covered in Coherent, Lumentum, Marvell, and Now Corning: NVIDIA’s 4 Photonics Bets.

Two cautions. NVQLink hardware integration is an application-based program, and the cudaq-realtime software API went public in March 2026 [9][29]. And quantum’s contribution to a 5.555 trillion dollar market cap is a rounding error. Buying NVIDIA because of quantum is not rational. NVIDIA’s meaning on this map is a coordinate axis rather than an investment candidate.

Quantinuum (QNT) / weighted score 21.0 / 5th

Market cap 13.1 billion dollars. Minus 17.3% against the 60 dollar offering price. Tier 5.

The only company on this map with a 5 on convergence. It is also one of the four names on both the Commerce and DARPA lists, so government is a 5 as well.

Both axes have primary evidence. Optical: cryogenic CMOS and integrated photonics collaboration with GlobalFoundries [21]. Decoding: a correlated decoding demonstration on Helios using an NVIDIA GPU-accelerated BP+OSD decoder [28].

The weakness is a 1 on neutrality. Single trapped ion camp. And in the three months since the June listing, the market has not accepted the offering price.

IonQ (IONQ) / weighted score 20.0 / joint 8th

Market cap 16.0 billion dollars. 1-year -4.1%. Tier 5.

Convergence is a 4. In April it linked two commercial systems through photonic entanglement [4], it acquired companies in photonic networking and single photon detection back to back, and it closed the 1.8 billion dollar SkyWater acquisition on July 31 [5][19]. The optical axis is certain, but I could not find primary evidence on the decoding axis, so no 5.

The 4 on revenue is real. Second quarter revenue was 80.1 million dollars, up 287 percent year over year [14]. Far and away first on revenue among listed quantum companies.

Two weaknesses. Neutrality is a 2, single trapped ion camp. And government is a 4, because it is not among the Commerce equity-condition names. The company with the largest revenue and the most advanced vertical integration has no government anchor, which you can read as building revenue on its own without government money, or as the government not picking it.

The SkyWater acquisition is itself a risk. Foundries are capital intensive and utilization driven. A quantum company running a foundry while also scaling qubits is doing two hard businesses at once.

Xanadu (XNDU) / weighted score 17.0 / joint 13th

Market cap 3.0 billion dollars. Minus 2.8% assuming a 10 dollar SPAC trust reference. Tier 5.

I covered Xanadu in April in The 1% Loss Wall (Part 1). The argument was that photonic quantum computing gets decided by loss in the end. That gate has passed, so let me go back over it.

In second quarter results on August 5 the company announced it had achieved 0.085 dB per facet of edge coupling loss [12]. That is the loss at the point where light enters and exits the chip, one of the hardest positions in optical packaging. In the August 31 roadmap it said it would bring the loss metric from 24.1x in 2026 down to 1.0x in 2030. Logical qubits: up to 200 in 2029, up to 500 in 2030, more than 1,000 in 2031 [13].

The loss metric and logical qubit targets in Xanadu’s August 31 roadmap

I think my April call was reasonably right. The judgment that loss is this company’s single metric was confirmed when the company made it the axis of its roadmap. There is something I did not see, though. I treated loss reduction as a process problem, and the schedule the company put out was four years long. From 24.1x to 1.0x is a twenty-four-fold reduction, and cash goes out every year in between.

That is the substance of the 1 on revenue. Second quarter revenue of 1.5 million dollars against a net loss of 42.1 million, 62.7 million cumulative for the first half [12]. The funding situation is better than I first read it, though. As of August 31, cash of 312.8 million dollars, and total available funding of about 686 million once you add remaining ATM capacity and committed Canadian government funding [13]. So the risk in this name is continuous dilution through the ATM rather than near-term cash exhaustion. The 4 on convergence lifts it, and the 1 on revenue is exactly that much of a price.

Tower Semiconductor (TSEM) / weighted score 20.5 / joint 6th

Market cap 25.1 billion dollars. 1-year +259.4%. Quantum beta +0.19 (not significant). Tier 4. Government 2 / Neutrality 4 / Bottleneck 3 / Revenue 5 / Convergence 4

Only two foundries on this map are actually running wafers with a named quantum customer: GlobalFoundries and Tower. Tower’s customer is Xanadu.

In a February 19, 2026 announcement, Tower said it had co-designed a process putting a Xanadu-specific material stack on top of the Tower silicon photonics platform. They are jointly optimizing ultra-low-loss silicon nitride and integrated photodiodes and have run multiple joint tapeouts [22]. In Xanadu CEO Christian Weedbrook’s words, the work with Tower “has been instrumental in taking our hardware from concept to prototype to demonstration systems.”

The character differs from GlobalFoundries. GlobalFoundries is the broad side, with three branches confirmed as actual process collaborations, and Tower is the side that went deep into one camp, photonic. Hence neutrality 4 and convergence 4. What position Tower has built in silicon photonics is covered in The Polymer That Tower Chose: LWLG x TSEM and The More Silicon Wins, the More InP Sells.

The weakness is price. Up 259.4 percent over a year, already heavily reflected. Seven points of weighted score separate it from GlobalFoundries, and 222 percentage points of one-year return. Two foundries on the same thesis, only one of them priced, is why hidden card 1 is what it is, and this name is the other side of it.

Quantum beta is 0.19 with a t of 1.5, statistically indistinguishable from zero. On the Jensen shock day it barely moved either, minus 0.2 percent. Meaning this share price does not move on quantum news. Which also means it holds up if quantum slips.

Keysight (KEYS) / weighted score 20.5 / joint 6th

Market cap 55.7 billion dollars. 1-year +98.0%. Quantum beta -0.05 (not significant). Tier 3. Government 3 / Neutrality 5 / Bottleneck 3 / Revenue 5 / Convergence 2

Neutrality is a 5. Whatever the qubit is made of, its state has to be measured, and measurement equipment does not discriminate by camp. Superconducting, trapped ion, neutral atom, photonic, the instruments go in. Which slice of the optical test chain this company holds is something I compared alongside FormFactor and Aehr in The Three Pillars of SiPh Wafer Test.

I lowered convergence from 3 to 2 in the rescoring. Selling control and readout instruments is a different job from building a real-time decoder. That moved it from 3rd to joint 6th.

Take the convergence axis out, though, and this company is 4th. As section 10.1 showed. For a reader who does not buy my optical interconnect thesis, this is the name to look at after GlobalFoundries. Government 3 (national lab deliveries), neutrality 5, and revenue 5 all stand independent of convergence.

Minus 0.7 percent on the Jensen shock day is the same as SMH that day, and the 64-day beta is minus 0.05, not significant. This share price does not respond to quantum sentiment.

Coherent (COHR) / weighted score 21.5 / 4th

Market cap 55.2 billion dollars. 1-year +211.4%. Quantum beta +0.27 (the only significant one in the component layer). Tier 4. Government 2 / Neutrality 5 / Bottleneck 4 / Revenue 5 / Convergence 3

Neutrality 5 and bottleneck 4. Every camp buys lasers and optical components. Trapped ion needs lasers to manipulate ions, neutral atom needs lasers to make optical tweezers, and the photonic camp needs the light source itself. It also holds InP capacity, which keeps the bottleneck score high. The earnings structure I took apart quarter by quarter in Coherent FQ4 2026: InP Capacity, CPO/NPO Content, and CapEx Return.

I lowered convergence from 5 to 3. That every camp buys lasers is industry common knowledge, but I could not find a primary source for a supply agreement with a named quantum customer. This is where common knowledge and evidence got treated differently. If Coherent discloses a supply relationship with a specific quantum company, this score goes to 4 and the rank moves into the top two. It is a trigger event that is not on the section 12 calendar but could actually happen.

The weaknesses are the 2 on government and the price. No direct name in government quantum programs, and 211 percent over a year already reflected. And what built that 211 percent was AI datacenter optics. That it is the only name in the component layer with a statistically significant quantum beta is still worth remembering, though. 0.27 with a t of 2.1.

FormFactor (FORM) / weighted score 18.5 / 10th (a listed company that sells dilution refrigerators)

Market cap 8.1 billion dollars. 1-year +267.8%. Jensen shock sensitivity -6.6%. Quantum beta +0.06. Straddles Tier 2 and Tier 3. Government 2 / Neutrality 4 / Bottleneck 4 / Revenue 5 / Convergence 2

As covered in section 8, this is the name on this map I saw properly last.

I knew about the cryogenic probers. Equipment that cools qubits at the wafer stage to 4 kelvin, down to 50 millikelvin, and measures them, a process step the superconducting and silicon spin camps have to pass through to reach volume production.

What I missed was the dilution refrigerators themselves. FormFactor acquired the JanisULT product line for 3.4 million dollars in June 2022, and the company now describes itself as the largest commercial supplier in the United States [34], and in March 2026 it launched the 30 millikelvin Flatiron [35]. A 3.4 million dollar acquisition changed the tier structure of this map four years later.

Hence the 4 on bottleneck. There is a reason it is not a 5, though. Flatiron is for benchtop validation, and the large system fridges that run 1,000 qubit machines are a separate thing. Bluefors is still large in that slot, and where FormFactor is strong is the device validation stage. That stage is one every company heading to volume production has to pass through, though.

It is also the flagship case of the two measurements diverging. On the Jensen shock day it fell the most of any component name, minus 6.6 percent, while the 64-day beta is 0.06 with a t of 0.6. Meaning the market classifies this company as quantum when a shock hits and does not see it that way at all the rest of the time. Even though it actually sells dilution refrigerators.

The counterargument is clear. Up 267.8 percent over a year, the most in this map’s component layer, and the driver of that move is HBM and CPO test, not quantum. Quantum revenue is not broken out either. The related analysis is in The Three Pillars of SiPh Wafer Test.

Linde (LIN) / weighted score 18.0 / 11th

Market cap 220.2 billion dollars. 1-year +1.4%. Quantum beta -0.08 (not significant). Tier 1. Government 1 / Neutrality 5 / Bottleneck 5 / Revenue 5 / Convergence 1

The only company on this map with a 5 on bottleneck. Because of helium-3.

Helium-3 is not something you mine. It is a byproduct of tritium decay, and tritium comes out of nuclear stockpile stewardship. Which means supply is determined with no relation to quantum demand, and there is effectively no way to increase it when demand rises. The amount that goes into one dilution refrigerator is small, but there is no substitute material. Few positions fit the phrase physical bottleneck more exactly.

And yet convergence is a 1 and the quantum beta is minus 0.08. The day after Jensen’s remark it actually rose 1.2 percent. Quantum sentiment and this share price are completely unrelated.

That combination is what makes Linde interesting. Take out the convergence axis and it climbs five places, from 11th to 6th. If you do not buy the optical interconnect thesis and want to look only at physical bottlenecks, this name is top tier. Buy my thesis and Linde is an industrial gas company unrelated to this cycle. That the same name moves between 6th and 11th depending on the frame is the actual use of the section 10.1 check.

The obvious limit is that quantum is a rounding error inside a 220.2 billion dollar market cap. Buying Linde because of quantum is not rational.

The rest

The remaining 11 companies, ticker, tier, market cap, 1-year, quantum beta, five axes, weighted score, group

D-Wave (QBTS, 15.5, joint 15th, quantum beta 1.21). Its character changed this year. On January 20, 2026 it closed the 550 million dollar acquisition of Quantum Circuits [24]. That is Rob Schoelkopf’s Yale team, and the dual-rail approach puts error detection into the hardware to cut physical resources per logical qubit. A company that only did annealing now has error-corrected gate model superconducting too, so I raised neutrality to 2. Convergence stays at 2, though. Catching errors inside the hardware and having a classical computer render a verdict within microseconds are different layers. The 3 on revenue has to be read carefully. First-half bookings of 35.5 million dollars grew 1,120 percent and RPO of 40.7 million grew 668 percent [20], but 20 million of that is one system sale and 2.3 million is bookings Quantum Circuits signed right before the acquisition closed. Recurring revenue did not go up tenfold. And on September 8 it signed a 100 million dollar definitive agreement with Commerce, which executed the equity condition for real [38]. This map’s government axis rubric treats equity-conditioned funding as a 5. So at the next update D-Wave goes from 4 to 5, its weighted score from 15.5 to 17.0, and its rank from joint 15th to joint 13th. The scores in this table are as of September 4, so I left them as they are.

Rigetti (RGTI, 15.0, joint 17th, quantum beta 0.93). Government is a 5. The government attached conditions and the share price has been flat for a year. The largest downgrade in this rescoring pass. I had given convergence a 4 for being on the NVQLink partner list, but list membership is participation and falls short of a joint demonstration, so it went to 2. That moved it five places, from 12th to joint 17th. Its purity is the highest on this map. Low score with the highest volatility puts it in the least favorable spot in the section 10.2 quadrants.

Infleqtion (INFQ, 17.0, joint 13th). The numbers changed in August. Second quarter revenue of 13.5 million dollars grew 157 percent year over year, and by the company’s description it is entirely quantum-related revenue with 100 percent organic growth [27]. I raised revenue from 2 to 3. In the other direction I lowered convergence from 3 to 2. Neutral atom being a pile of lasers is true, but that is light controlling its own atoms, not optical interconnect linking QPU to QPU. The two adjustments offset, moving it from 10th to joint 13th. The structure where revenue came from quantum sensing before quantum computing is covered in Infleqtion and Quantum Sensing.

Quantum Computing Inc (QUBT, 12.5, last, quantum beta 0.79). Still last, but the content is different. Second quarter revenue was 5.6 million dollars. The year-ago quarter was 61 thousand dollars, so effectively starting from zero, and most of it is photonics product sales. Backlog of 42.5 million dollars, cash and investments of about 1.3 billion [26]. It made three acquisitions this year including NHanced Semiconductors, picking up a semiconductor packaging fab. So I raised revenue from 1 to 3, bottleneck from 1 to 2, and convergence from 2 to 3. It is still last because of the 1 on government and the 2 on neutrality. No name in government programs and a narrow camp. Holding 1.3 billion dollars in cash and starting to generate actual revenue makes it a different company from the one in 2024. Two of the five axes are simply still thin.

MKS (MKSI, 17.5, 12th) / Novanta (NOVT, 15.0, joint 17th). Both have high neutrality and a 2 on convergence. They sell laser and vacuum and photonics components to every camp, and those products are merely adjacent to the two convergence axes. Purity is minus 1.5 percent and plus 0.7 percent, effectively unrelated.

Anglo American (AAL.L, 15.5, joint 15th). Element Six CVD diamond is the material for the diamond NV camp. Bottleneck is a 4, and the camp is narrow so neutrality is a 3. One watch item: the merger with Teck is in progress, so the future owner of the Element Six exposure could change.

ULVAC (6728.T) / Atlas Copco (ATCO-A.ST), both 14.5, joint 19th. Vacuum equipment. It goes into trapped ion and neutral atom ultra-high vacuum and superconducting cryogenic vacuum alike. Neutrality is a 4 but convergence is a 1, which keeps the rank low. Purity is plus 1.9 percent and plus 0.5 percent, completely unrelated.

AXT (AXTI, 14.0, 21st, quantum beta 0.74). What produced that 1,914 percent one-year number is the InP substrate cycle. The actual driver is in The More Tents Meta Pitches, the Faster InP Runs Dry and AXT and Lumentum’s Prepayment Deal. Convergence is a 2 because InP is upstream material for the light source layer, not a component on either convergence axis. I also excluded this name from the component layer averages earlier.

Alphabet (GOOGL, 13.0, 22nd, quantum beta 0.11, not significant). It leads the superconducting camp with Willow and its name is in the GlobalFoundries announcement, but that is a supporting quote, not a wafer production agreement. Quantum is a rounding error inside a 4.12 trillion dollar market cap. What this camp can and cannot do I worked through in Google Quantum AI says it can break Bitcoin in 9 minutes.

All market caps and returns are cross-checked between TradingView and Yahoo Finance, based on the September 4, 2026 close. Purity is the January 8, 2025 close against the January 7, 2025 close [3].

15. Scenarios

Base case: the government schedule holds (qualitatively 55 percent). The DOE keeps its 2028 target and the Commerce letters convert to definitive agreements. On this path Group C moves first. Raising logical qubits requires multiplying physical qubits, and that turns cooling and lasers and detectors and foundry capacity into actual orders. This is also the first stretch where GlobalFoundries’ government score and neutrality score get confirmed in revenue.

Alternative case: the schedule slips (30 percent). The 2028 target moves to 2030 and entry into DARPA Stage 3 comes out narrower than expected. Group B hurts most. A government anchor raises financing availability, it does not produce next quarter’s share price. Companies with a 1 or 2 on the revenue axis have to raise again, and raising is dilution. Xanadu gets caught first on this path. Group C is relatively unbothered, because semiconductors and AI are making these companies’ revenue anyway.

Downside case: a second Jensen moment (15 percent). A technical setback becomes public, or an influential figure pushes the timeline back again. January 8, 2025 repeats. Groups B and D drop 30 to 45 percent in a day and Group C falls single digits or rises. Hamamatsu rose that day. One thing is different between then and now, though. Back then there was no government capital on the table.

Additional risk: a self-contained Chinese ecosystem. If China starts sourcing cooling equipment and lasers and detectors domestically, Group C’s addressable market shrinks to non-China. Tier 2 and Tier 4 are the most directly exposed.

All probabilities are qualitative estimates.

16. Where this framework breaks

The first premise is that modality neutrality is worth something. If one camp pulls decisively ahead, a component network specific to that camp forms separately. If superconducting wins, cooling and superconducting wafers are everything and laser demand shrinks. If photonic wins, dilution refrigerators stop being necessary. In that case a 5 on neutrality flips from a strength into a lack of focus. This signal shows up first as the modality mix of the teams entering DARPA Stage 3 tilting heavily one way. DARPA itself states that QBI is not framed as a competition to pick a few top performers [6], so it should be read as concentration rather than elimination.

The next premise is that government capital is a floor. If the Lofgren objection is upheld and the equity conditions fall through, or if the CHIPS quantum budget is clawed back, Group B loses its grounds. It gets settled at the Q4 conversion to definitive agreements. If Commerce cannot move the letters into definitive agreements, or softens the conditions, the government axis becomes paper from that point on. The first evidence is already in. Three converted to definitive agreements on September 8 and the equity was executed as specified [38][39][40]. Whether the remaining six follow the same path is the other half of this axis.

The last premise is that NVIDIA keeps holding the orchestration layer. If the controller makers, Quantum Machines and QBLOX and Zurich Instruments, establish their own real-time decoder standard not tied to GPUs, half the convergence axis disappears. It would look like names dropping off the GTC 2027 partner list.

And there is one unconfirmed item holding up this whole map. Until the companies in Tiers 1 through 4 break out quantum revenue, nobody knows how much actually flows through this supply chain. The neutrality and convergence scores I assigned measure the existence of a relationship, not an amount. Once separate disclosure starts, this table may have to be rebuilt from scratch.

All three, and the last one too, get their first evidence between this year’s fourth quarter and next year’s first.

17. Watchlist: private and adjacent

SkyWater was delisted as of July 31, 2026 when it became an IonQ subsidiary. A year earlier it would have been an independent name in Tier 4.

18. Closing

If you invested in quantum over the past year you are probably flat or down. The median of the four pure plays is minus 2.2 percent. Over the same period the component and metrology layer beneath them rose 161.4 percent at the median. And the quantum ETF rose 54.9 percent, with most of that return coming from outside the quantum pure plays.

But what built that 161.4 percent was AI and the infrastructure investment returns that came with it, not quantum. What this piece did was pull the quantum exposure out from under that AI return, and sort by one yardstick which companies actually hold a relationship and which are only named. Setting that yardstick hard cut the companies favorable to my thesis first, and first place survived it anyway.

Quantum pure plays stay on the list too. The US government put money down with equity as the condition at nine companies in May 2026, the DOE set a 2028 target, and China wrote quantum near the top of its five-year plan. The US has no way to step off this board. So even with no product and no revenue, these companies do not come off the investment list. What that government capital buys, though, is the probability that funding does not dry up. The probability of success is a separate thing. That distinction is the most important sentence in this piece.

I see two axes for picking winners. Whatever the modality, growing distance means connecting with light. And running error correction in real time means attaching classical decoding faster than that platform’s cycle. Companies with a relationship on those two axes confirmed by primary sources sell regardless of how the camp fight resolves. Which is why first place on this map is not a company that makes qubits but the foundry that fabs them.

Of course I could be reading it wrong. If only one modality survives DARPA Stage 3, the value assigned to neutrality disappears and the ranking has to be rebuilt. If the Commerce letters cannot become definitive agreements, the grounds holding up Group B are gone. If the controller companies establish a decoder standard not tied to GPUs, half the signature axis vanishes. And above all, until these companies break out quantum revenue, what I measured is the existence of relationships, not amounts. All four get their first evidence between this year’s fourth quarter and next year’s first. I will pull this table back out then and compare.

19. References and sources

[1] CNBC, “Quantum stocks like Rigetti plunge after Nvidia’s Huang says the computers are 15 to 30 years away”, Jan 8, 2025.

[2] Constellation Research, “Here’s what Nvidia CEO Jensen Huang said about quantum computing, Project Digits and robotics”, Jan 2025.

[3] Prices, market caps and returns: cross-checked between the TradingView scanner and Yahoo Finance, based on the 2026-09-04 close. One-year returns are TradingView Perf.Y. The September 8 and 9 moves in section 13 and in hidden card 1 are that day’s closes from the same sources.

[4] IonQ, “IonQ Achieves Key Photonic Interconnect Milestone, Demonstrating Networked Quantum Systems Using Entanglement”, Apr 2026.

[5] IonQ, “IonQ Completes Acquisition of SkyWater Technology”, Jul 31, 2026.

[6] DARPA, “Quantum Benchmarking Initiative Stage B selection”, Nov 6, 2025.

[7] NIST and the US Department of Commerce, “Department of Commerce Announces Letters of Intent With 9 Companies for $2 Billion to Accelerate U.S. Leadership in Quantum Computing”, May 21, 2026.

[8] US Department of Energy, “Energy Department Announces Initiative to Create and Deploy the World’s First Scientifically Relevant, Fault-Tolerant Quantum Computers”, Jun 23, 2026.

[9] NVIDIA, “NVIDIA Introduces NVQLink, Connecting Quantum and GPU Computing for 17 Quantum Builders and Nine Scientific Labs”, Oct 28, 2025.

[10] Quantinuum, “Quantinuum Announces Pricing of Upsized Initial Public Offering”, Jun 2026.

[11] Infleqtion, “Infleqtion and Churchill Capital Corp X Complete Business Combination”, Feb 17, 2026.

[12] Xanadu Quantum Technologies, “Q2 2026 results press release, Form 6-K Ex-99.1”, Aug 5, 2026.

[13] Xanadu, “Xanadu Charts Path to Over 1,000 Logical Qubits by 2031”, Aug 31, 2026.

[14] IonQ, “Form 10-Q, Q2 2026, period ended 2026-06-30”, Aug 2026.

[15] Quantum Computing Report, “Oratomic Secures $300M Series A to Build Fault-Tolerant Quantum Computers via Reconfigurable Neutral-Atom Arrays”, Jul 2026.

[16] PsiQuantum, “PsiQuantum Raises $1 Billion to Build Million-Qubit Scale, Fault-Tolerant Quantum Computers”, Sep 2025.

[17] The Quantum Insider, “How is NVIDIA ‘Betting’ on Quantum? A Look at NVentures Quantum Investment Playbook”, Sep 2025.

[18] ChinaTechNews, “China’s Quantum Tech Sector Sees Capital Surge as State Funds and Startups Accelerate Growth”, Aug 9, 2026.

[19] Bloomberg, “IonQ Cleared to Close $1.8 Billion SkyWater Deal After FTC Deadlock”, Jul 29, 2026.

[20] D-Wave Quantum, “D-Wave Reports Second Quarter 2026 Results”, Aug 6, 2026.

[21] GlobalFoundries, “GlobalFoundries launches Quantum Technology Solutions to scale U.S. quantum manufacturing”, May 21, 2026.

[22] Tower Semiconductor, “Xanadu and Tower Semiconductor Deepen Strategic Collaboration to Accelerate Photonic Quantum Hardware Innovation”, Feb 19, 2026.

[23] Hamamatsu Photonics, “Quantum Computing and Communication Questions & Answers”, accessed Sep 7, 2026.

[24] D-Wave Quantum, “D-Wave Completes Acquisition of Quantum Circuits Inc., Creating World’s Leading Quantum Computing Company”, Jan 20, 2026.

[25] IBM, “IBM Completes Acquisition of HRL Laboratories to Accelerate the Future of Quantum”, Aug 26, 2026.

[26] Quantum Computing Inc., “Quantum Computing Inc. Reports Second Quarter 2026 Financial Results”, Aug 2026.

[27] Infleqtion, “Infleqtion Reports Updated Financial Results for Q2 2026 and FY26 Revenue Guidance”, Aug 17, 2026.

[28] Quantinuum, “Real-time error correction at increased scale”, 2026.

[29] NVIDIA, “Introducing cudaq-realtime for programming the Logical QPU”, Mar 16, 2026.

[30] NVIDIA, “CUDA-Q QEC 0.6 Enables Real-Time QEC with NVQLink”, Apr 14, 2026.

[31] Oxford Instruments plc, “Q3 Trading Update, confirming completion of the NanoScience divestment”, Jan 15, 2026.

[32] Honeywell, “Form 10-Q, Q2 2026, period ended 2026-06-30”, Jul 2026.

[33] US House Committee on Science, Space, and Technology, “Full Committee Markup of the National Quantum Initiative Reauthorization Act”, Apr 2026.

[34] Quantum Computing Report, “FormFactor Acquires JanisULT’s Dilution Refrigerator Product Line”, Jun 2022.

[35] FormFactor, “FormFactor Introduces Flatiron Dilution Refrigerator for Benchtop Millikelvin Research and Quantum Hardware Validation”, Mar 2026.

[36] Oxford Instruments plc, “Sale of Oxford Instruments’ quantum business, NanoScience”, Jun 2025.

[37] IonQ, “IonQ Completes Acquisition of Lightsynq, Accelerating Quantum Computing and Networking Roadmap”, Jun 3, 2025.

[38] D-Wave Quantum, “D-Wave Quantum Finalizes Definitive Agreement with U.S. Department of Commerce for up to $100 Million”, Sep 8, 2026.

[39] Rigetti Computing, “Rigetti Signs Definitive Agreement for $100M with U.S. Government to Accelerate R&D for Superconducting Quantum Computing”, Sep 8, 2026.

[40] Quantinuum, “Form 8-K, Quantinuum Finalizes $100 Million CHIPS Award”, Sep 8, 2026.

This article is an independent technical analysis published by PhotonCap, based on an engineering perspective. All content is derived from publicly available information and is intended solely for educational and informational purposes. In other words, nothing in this material should be construed as a recommendation to buy, sell, or hold any specific securities. Please note this carefully.

The author may hold positions in the securities mentioned herein and reserves the right to trade such securities at any time without prior notice. Readers should conduct their own thorough review and research before making any investment decisions.

[#Breeze] HPE Wave Count Update On A Weekly Logarithmic Chart. Breeze [graphic]

#Breeze — 9/12/2026 4:04:32 PM

HPE Wave Count Update On A Weekly Logarithmic Chart. Breeze

[mend] The story is to continue to grow with BE.

mend — 9/12/2026 3:08:11 PM

[#Breeze] BE Wave Count Update. I projected Wave (iii)'s target at 340. Wave v of (iii) is...

#Breeze — 9/12/2026 2:36:41 PM

BE Wave Count Update. I projected Wave (iii)'s target at 340. Wave v of (iii) is in play with a target range of 317 - 359 and that would complete Wave (iii). Breeze

[#Breeze] SIMO Wave Count Udpdate. Minute Wave (circle) iii of Minor Wave 5 is in play wit...

#Breeze — 9/12/2026 2:31:27 PM

SIMO Wave Count Udpdate. Minute Wave (circle) iii of Minor Wave 5 is in play with a target of 358. Breeze

[#Breeze] SNOW Wave Count Update. Another Bingo. [graphic]

#Breeze — 9/12/2026 2:29:08 PM

SNOW Wave Count Update. Another Bingo.

[toccodolce] [X] $BE Enverus’s 4.8 GW assumes a particular technology split inside a 31.6-GW...

toccodolce — 9/12/2026 1:49:46 PM

[X]

$BE
Enverus’s 4.8 GW assumes a particular technology split inside a 31.6-GW BTM generation forecast. Fuel cells are only about 15% of that generation mix.

Enverus currently models just 4.8 GW of U.S. BTM fuel cells through 2030.$BE already has 1.2 GW contracted with Oracle,… pic.twitter.com/aeURO0ESWH
— Equity Climb (@equityclimb1) September 12, 2026

[/X]

[X]

$BE
Rabobank counts more than 130 GW of announced onsite resources but cites an S&P estimate of only about 22–32 GW actually deploying by 2030. Enverus independently forecasts 31.6 GW of generation required for BTM industrial loads, with data centers responsible for 22.5 GW of… pic.twitter.com/nxHky5dp9J
— Equity Climb (@equityclimb1) September 12, 2026

[/X]

[X]

Gas-fired power accounts for ~94% of Enverus’s 31.6 GW U.S. BTM forecast, with reciprocating engines at 7.1 GW, small/medium turbines at 6.2 GW, and fuel cells at 4.8 GW. Cleanview says $CAT leads tracked OEM share at 33% overall, while $BE has risen to 14% just within year an… pic.twitter.com/U0T9LbsfVE
— Equity Climb (@equityclimb1) September 12, 2026

[/X]

[Kirk ©] Thanks for the HPE update. I think there is a cup n handle now that has a targe...

Kirk © — 9/11/2026 7:02:17 PM

Thanks for the HPE update.

I think there is a cup n handle now that has a target near your $73 and it if goes much higher without a pullback, it will make a new one with a neckline at about 63 that might close a lot of ground with Dell's recent gains ($85 target?) !

I also see an iH&S tested from above targeting ~65

[#Breeze] DELL Wave Count Update. Finally a breakout from that bull flag. Minute Wave (cir...

#Breeze — 9/11/2026 2:16:00 PM

DELL Wave Count Update. Finally a breakout from that bull flag. Minute Wave (circle) v of Minor Wave 3 is in play with a target of 606. Breeze

[#Breeze] Spot Silver Wave Count Update. Wave (b) of (ii) is complete with a 61.8% retrace...

#Breeze — 9/10/2026 3:18:41 PM

Spot Silver Wave Count Update. Wave (b) of (ii) is complete with a 61.8% retracement. Wave (c) of Wave (ii) is in play. Target is that 61.8% retracement level at 61. A Delta Low Turning point is due on 9/14 +/- 2 rating days. FOMC is on 9/12. Wave (a) lasted 4 trading days. Breeze

[#Breeze] Tocco here's a rough cut EROC Wave Count. Sub-microwave (2) is in play. If that ...

#Breeze — 9/10/2026 11:31:12 AM

Tocco here's a rough cut EROC Wave Count. Sub-microwave (2) is in play. If that 61.8% retracement level doesn't hold then watch the 76.4% level. This price pattern resembles ones of IPO stocks as they waterfall below IPO price. Breeze

[toccodolce] Hi Breeze, Could you please post a chart on $EROC (Erock Inc.: NYSE)? It's a b...

toccodolce — 9/10/2026 11:03:52 AM

Hi Breeze,

Could you please post a chart on $EROC (Erock Inc.: NYSE)?

It's a behind the meter AI compute. The company has signed contracts with $META and Anthropic.

TIA,

Tocco

[Selectric II] I mentioned DELL as an alternative to SMCI. He's gone heavily into AI servers a...

Selectric II — 9/9/2026 12:03:36 PM

I mentioned DELL as an alternative to SMCI. He's gone heavily into AI servers and data centers; no longer just a PC maker.

Yes to Texas; he pretty much put Round Rock on the map. Enormous campus north of Austin, just down the road from Taylor/Hutto area where Samsung is expanding its second chip factory in the area and Elon plans to build his. There's a (at least one) big data center under construction there, too.

finance.yahoo.com

[Kirk ©] No, but it is probably a good choice, especially before it had its recent run up...

Kirk © — 9/9/2026 11:52:11 AM

No, but it is probably a good choice, especially before it had its recent run up. Its namesake is far less politically toxic than that of Tesla or SpaceX... but he has supported Trump and GOP (he's from Texas, no?) so that could bite him in the a$$ets if we get a change in the White House in 2028.

I have some HPE at a near zero cost basis via working at HP long ago.... and really don't need more tech as I'm happy with the other stocks that are performing far better.

[Selectric II] Have you considered DELL?

Selectric II — 9/9/2026 11:27:51 AM

[Kirk ©] Yeah, just reaching the ATH from $30 or lower would be a Nazaré wave ride! [gra...

Kirk © — 9/9/2026 11:16:23 AM

Yeah, just reaching the ATH from $30 or lower would be a Nazaré wave ride!

[#Breeze] If my wave count is valid then that pull back I'll label as Minute Wave (circle)...

#Breeze — 9/9/2026 10:59:02 AM

If my wave count is valid then that pull back I'll label as Minute Wave (circle) ii, then an explosive Minute Wave (circle) iii will unfold. That's the wave to ride. Or in your case surf.

[Kirk ©] Thanks. It looks like her possible plan will fit your count well. It looks lik...

Kirk © — 9/9/2026 10:46:51 AM

Thanks. It looks like her possible plan will fit your count well.

It looks like my "Cramer Timing" came near a top:

From my notes:

1/30/24 @ $495.67 ($49.57 post split): Warning! Cramer was gushing about SMCI this morning and I believe he had the CEO on his show yesterday.

That is usually a "take profits" warning shot. Message 34554089

[#Breeze] Kirk, here's a SMCI Wave Count Update on a Weekly Logarithmic Chart. Primary Wav...

#Breeze — 9/9/2026 10:35:20 AM

Kirk, here's a SMCI Wave Count Update on a Weekly Logarithmic Chart. Primary Wave (circle) 2 has unfolded a ((A))-((B))-((C))-((D))-((E)) triangle. The current Wave Count is Wave (v) of Minute Wave (i) is in play with a target of 45.79. Once complete expect a 50%-61.8% retracement which means prices will drop within the apex area of the triangle. Breeze

[Elroy] the shares are in a taxable account and the position is relatively small compare...

Elroy — 9/9/2026 9:36:15 AM

the shares are in a taxable account and the position is relatively small compared to my major positions that were in more urgent need of profit taking.

You have shares that were in "urgent need of profit taking"?

Are the shares in the room with you now?

[Kirk ©] So, any guesses if SMCI will break-out of this 2.5 tough years pattern? I wonde...

Kirk © — 9/9/2026 9:27:08 AM

So, any guesses if SMCI will break-out of this 2.5 tough years pattern?

I wonder if y1kate is still in it.

I've never owned it and someone close to me has been buying the dips and sold all this week at $41. My guess is she'll buy it back if she can get a 10 or 20% discount. I worry about the allegations of fraud for selling Nvidia equipment to the Mainland Chinese.

I own HPE instead and have just been holding since the shares are in a taxable account and the position is relatively small compared to my major positions that were in more urgent need of profit taking.

.

[#Breeze] AMD Wave Count Update. Minor Wave 4 unfolded as a W-X-Y. Appears complete. IF so...

#Breeze — 9/8/2026 7:27:03 PM

AMD Wave Count Update. Minor Wave 4 unfolded as a W-X-Y. Appears complete. IF so then Minor Wave 5 is in play with a target of ~736. Breeze

[#Breeze] tocco here's a SIVEF Wave Count Update. Minute Wave (circle) ii appears complete...

#Breeze — 9/8/2026 6:55:13 PM

tocco here's a SIVEF Wave Count Update. Minute Wave (circle) ii appears complete. A price above 3.37 will confirm. Minute Wave (circle) iii in play with a target of 19.27. Breeze

[#Breeze] SIMO Wave Count Update. Minute Wave (circle) iii of Minor Wave 5 is in play with...

#Breeze — 9/8/2026 4:21:35 PM

SIMO Wave Count Update. Minute Wave (circle) iii of Minor Wave 5 is in play with a target of ~358. IF this wave count progression continues then Minor Wave 5 target will get exceeded. Breeze

[#Breeze] tocco, here a SEI Wave Count Update. Wave (iii) is in play with a target of ~84 ...

#Breeze — 9/8/2026 3:55:19 PM

tocco, here a SEI Wave Count Update. Wave (iii) is in play with a target of ~84 and Minor Wave 3 is in play with a target of ~172. Breeze

[#Breeze] A_G, here's a BE Wave Count. Wave (iii) is in play with a target of 340.71 and W...

#Breeze — 9/8/2026 3:49:18 PM

A_G, here's a BE Wave Count. Wave (iii) is in play with a target of 340.71 and Wave iii of (iii) is in play. Breeze

[toccodolce] $SEI is on Fire! $SEI Raised Guidance Solaris Energy Infrastructure stock jump...

toccodolce — 9/8/2026 10:52:11 AM

$SEI is on Fire!

$SEI Raised Guidance

Solaris Energy Infrastructure stock jumps on raised guidance

Investing.com -- Solaris Energy Infrastructure, Inc. ( NYSE:SEI) shares rose 6% on Tuesday following the company’s announcement of raised guidance for 2026 and initial projections for the first quarter of 2027.

The company increased its Adjusted EBITDA guidance for the third and fourth quarters of 2026, citing stronger contributions from its core power services offerings and better-than-expected performance from recently acquired businesses.

For the third quarter of 2026, Solaris now expects Adjusted EBITDA of $110 million to $130 million, up from its prior range of $90 million to $105 million, representing a 23% increase at the midpoint. The fourth quarter guidance was raised to $145 million to $180 million from the previous range of $100 million to $120 million, marking a 48% increase at the midpoint.

The company also initiated guidance for the first quarter of 2027, projecting Adjusted EBITDA between $200 million and $240 million.

This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.

Tocco

[Kirk ©] I think Biden and Powell screwed the pooch, so to speak, when inflation was NOT ...

Kirk © — 9/8/2026 10:28:11 AM

I think Biden and Powell screwed the pooch, so to speak, when inflation was NOT "transitory." They are both to blame. I think Trump is a fool to push for lower rates while he pumps up inflation with his policies.

We have a "problem" of what is now being called a "K-shaped economy" where the string pullers at the DNC (loaded rich people like Soros, Obama, Clooney and Bernie Sanders) are thriving just like most of the savers and Boomers who populate both Democrats and GOP.

Inflation is an AVERAGE that is probably as meaningless to the average person as the "average home price" in my zip code is with top homes going for over 10x what a condo goes for.

Another example, my home will be paid for in a few months so my monthly cost will be property taxes that are locked in with Prop 13 at a rate now about equal to what an owner of a nice condo pays for monthly HOA fees. I'm not counting water, sewer, power, garbage, etc. as those are the same for most, based on use. So, those who own their homes outright or bought 10 or 20 years ago before their values soared, are doing great on paper with monthly home costs locked in and probably refinanced around 3%, give or take 1%.

We do NOT see the same inflation as renters in the Bay Area. FOR ME, this is BY DESIGN as I saw it coming decades ago and invested in far more land and house than I need.

The 20 and 30 somethings who are renters without parents to fund their down payments are highly resentful so we get people like Mamdani in NY and Ro Khanna here in the Silicon Valley proposing wealth taxes and other reactionary nonsense that will probably tank the economy and stock markets.

We could argue the lower arm of the K-economy has been in a recession since Biden stopped handing out free money to get us out of COVID. Also, that free money to the lower earners plus forgiving hundreds of billions in student loans CAUSED even higher inflation than just the wars.

Finally, I'm not sure if the whole US outlawed evicting tenants who didn't pay rent during COVID but here in CA it is still very hard to evict tenants and I think you have to pay them $14,000 or more to move out of a unit if the owner wants to move back in, often after a divorce. With the problems collecting rent, rent control laws and the propensity of those with low credit ratings to skip rent or trash units, it is no wonder "nobody" wants to build low cost housing for the low and no income cohort. Add in Biden flooded those markets with illegal aliens to compete for those low cost housing units and it is no surprise that rents went up.

They're getting $6K a month in SF for 2b/2b rentals with a single parking space in very old buildings! You can get a pretty decent 4b/2.5bath townhouse or "affordable" house here for $6K a month that sells for "only" $3M. The numbers are crazy.

[Sam] That is an interesting blog and post, thank you for posting it. Here is another ...

Sam — 9/8/2026 7:46:06 AM

That is an interesting blog and post, thank you for posting it. Here is another brief excerpt:

>>
He reinforced the Fed’s 2% inflation target as a “firm, fixed target” and stated unequivocally that it was “the Fed’s job to deliver stable prices.”

Have they been delivering it?

Absolutely not, as Warsh plainly acknowledged in saying the Fed bears “responsibility for 65 months of sustained, elevated inflation.”

And unless they are confident that underlying inflation is moving back to 2% at “sufficient speed,” Warsh believes they “have work to do.”
<<
----------------------------------------------------------------

Well, what was going on in 2021 that led to PCE going up? The economy was in the tank due to COVID and the administration and the Fed were both trying to support it with loose money and rates, leading to higher inflation while avoiding the widely expected recession. The rate started coming slowly down a year later. Very slowly in order to avoid killing the economy with too restrictive yields. Recall, everyone was predicting a recession and it never came. The cost of it never coming was inflation higher than it otherwise would have been. But inflation has been gradually coming down since 2023, until being pushed up by the onset of the tariffs and the beginning of the Iran war. The Powell-Biden plan was to avoid recession by tolerating somewhat higher inflation; they judged that inflation wouldn't get out of hand. Warsh seems to not understand that.

[toccodolce] [X] Decrypting Trump’s power play post on Canada’s dollar: This was NOT a mess...

toccodolce — 9/7/2026 4:02:22 PM

[X]

Decrypting Trump’s power play post on Canada’s dollar:

This was NOT a message to regular Canadians. It was a message to Mark Carney. Anyone selling it as a threat to canadians is a liar.
Anyone saying it is a threat to our sovereignty is a liar (also canada is ruled by a king… pic.twitter.com/9np3Vn91fh
— CDN_Karma (@cdn_karma) September 6, 2026

[/X]

[#Breeze] Kirk, here's a DBA - Invesco Agriculture Fund Monthly Chart. Tells all. Working ...

#Breeze — 9/7/2026 9:49:24 AM

Kirk, here's a DBA - Invesco Agriculture Fund Monthly Chart. Tells all. Working its way on two H&S continuation formation. No Wave count needed on this one. Breeze

[#Breeze] MSFT Wave Count Update. Wave (iv) of Minute Wave (circle) i is in play next week...

#Breeze — 9/6/2026 7:52:17 PM

MSFT Wave Count Update. Wave (iv) of Minute Wave (circle) i is in play next weeks trades should be the tell if its complete. Breeze

[#Breeze] MAGS Wave Count Update. Minor Wave 5 of Intermediate Wave (1) is in play with a ...

#Breeze — 9/6/2026 7:43:03 PM

MAGS Wave Count Update. Minor Wave 5 of Intermediate Wave (1) is in play with a target of 91.42. Note breakout from a bull flag/staff formation that has an obj of 78.18. Breeze

[Kirk ©] Via email from Charlie Bilello 3) The Iran War at 6 Months We are now 6 mo...

Kirk © — 9/6/2026 5:53:24 PM

Via email from Charlie Bilello

3) The Iran War at 6 Months

We are now 6 months into the Iran war.

The impact on inflation has been felt by everyone around the world, with the prices of food and energy spiking since the start of the War.

In the US, we saw a painful new record for American drivers: for the first time in history, the national average price of gasoline was above $4.00/gallon every single day in August. And heading into Labor Day at $4.15 per gallon, this will be the highest level ever on the holiday (prior record was $3.82/gallon set in 2012).

So why was the stock market up 13% in the first 6 months of the war?

I’ll repost what I wrote at the start of the war and again at the 3-month mark…

The best we can say in studying past military conflicts is that with the passage of time, the stock market has tended to rise – and the more time that has passed, the more it has risen.

There’s two reasons for this: 1) all wars eventually come to an end, and 2) the economy and earnings, even if impaired in the short run, still tended to grow in the long run despite these conflicts.

More at bilello.blog

[robert b furman] No doubt an ongoing Mega trend. Like Breeze says, "it won't be a straight line"...

robert b furman — 9/5/2026 10:30:10 AM

No doubt an ongoing Mega trend.

Like Breeze says, "it won't be a straight line", but it looks good now.

Bob

[Elroy] It would be interesting to break out what percentage of the semiconductor sales ...

Elroy — 9/4/2026 9:05:49 PM

It would be interesting to break out what percentage of the semiconductor sales growth is memory price increases and what is "everything else".

It's easy to grow sales when prices are going up.

[Kirk ©] Sales growth compounding rate is still increasing... this is incredible. Gl...

Kirk © — 9/4/2026 7:40:19 PM

Sales growth compounding rate is still increasing... this is incredible.

Global Semiconductor Sales Increase 6.4% Month-to-Month in JulyFriday, Sep 04, 2026, 5:00pm
by Semiconductor Industry Association

Worldwide chip sales in July up 135.1% year-to-year; year-to-date sales through July exceed highest-ever annual total
WASHINGTON, Sept. 4, 2026 – The Semiconductor Industry Association (SIA) today announced global semiconductor sales were $146.8 billion during the month of July 2026, an increase of 6.4% compared to the June 2026 total of $137.9 billion and 135.1% more than the July 2025 total of $62.5 billion. Monthly sales are compiled by the World Semiconductor Trade Statistics (WSTS) organization and represent a three-month moving average. SIA represents 99% of the U.S. semiconductor industry by revenue and nearly two-thirds of non-U.S. chip firms.

“The global semiconductor market grew on a month-to-month basis for the 17th consecutive month in July, and the industry has already reached its highest-ever annual global sales total just seven months into the year,” said John Neuffer, SIA president and CEO. “Sales into the Americas, the Asia Pacific, and China continue to push strong year-to-year growth.”

Regionally, year-to-year sales in July were up in the Americas (171.3%), Asia Pacific/All Other (134.9%), China (123.6%), Europe (85.2%), and Japan (50.8%). Month-to-month sales increased in Japan (9.2%), the Americas (8.7%), Asia Pacific/All Other (6.8%), Europe (5.2%), and China (2.9%).

For comprehensive monthly semiconductor sales data and detailed WSTS forecasts, consider purchasing the WSTS Subscription Package. For detailed historical information about the global semiconductor industry and market, consider ordering the SIA Databook.

[ July 2026 chart and graph]
.

[nicewatch] Marvell Drew Its Own Plasmonics Map in June: Two Overlapping Targets and an Unna...

nicewatch — 9/4/2026 6:48:31 PM

Marvell Drew Its Own Plasmonics Map in June: Two Overlapping Targets and an Unnamed Material $MRVL $LWLG | Where POH actually points, the overlap with Celestial AI, and what to check at ECOC

PhotonCap Sep 04, 2026

Somewhere on a Swiss airfield, two points 1.4 km apart were linked over a 226 GHz radio carrier. Net rate 153 Gb/s. The thing that caught that carrier at the receive end and put it back onto light was a plasmonic modulator. The paper is from ETH Zurich and Polariton, and the publication date is the interesting part: it ran four days before Marvell announced the acquisition. But a more important document came out two months later. The Polariton founders, now carrying Marvell business cards, wrote up the application range of this technology themselves. In that post, plasmonics sits on co-packaged optics inside the package and on coherent links between campuses, at the same time. The layers do not divide. They overlap. Which raises a question about the other optical platform this company bought four months earlier for more than three billion dollars. To say where I land before the paywall: I read this overlap as an internal contest that has not been settled yet rather than a failure of division, and I think it gets settled on materials and reliability rather than on bandwidth. The name of that material never appears in anything Marvell has published. It does appear in the August deck of the company that sells it, where Marvell is listed as a customer.

Contents

The paper that ran four days early
Two months later, the company drew its own map
What 1,400 meters proved and what it did not
So where does Celestial AI’s SiGe stand
Two blanks left in the material layer
One sentence from August 11
The ECOC 2026 checklist
Closing

1. The paper that ran four days early

A joint team from ETH Zurich and Polariton Technologies linked two points 1,400 m apart over a 226 GHz carrier [1]. Net 153.2 Gb/s, 184 Gb/s at the line rate. The paper claims this as a record on net-rate times distance. The signal did cross 1.4 km of outdoor path rather than a short bench span. Worth keeping in proportion though: the equipment sat in a container and the modulator was contacted with GSG probes, so this is an outdoor link demonstration and not a finished product.

The architecture is unusual. At the transmit end, two optical tones are mixed to generate the 226 GHz carrier and radiated. At the receive end, an antenna picks that carrier up and feeds it straight onto the modulator electrodes. The resulting electrical signal rides back onto light and goes into fiber. What gets skipped is the electronic downmixer in the wireless receive front end, with the sub-THz signal lifted directly into the optical domain instead. Baseband conversion and signal processing still happen downstream in the coherent receiver, and the RF amplifiers, the antennas and the optical amplifiers are all still there. The part doing the receive-side job is a plasmonic Mach-Zehnder modulator filled with an organic material.

Now the dates. The paper was received on December 26, 2024 and published on April 18, 2026 [1]. Four days after that, on April 22, Marvell announced the Polariton acquisition [2]. Two of the thirteen authors, Marcel Destraz and Wolfgang Heni in the fifth and sixth slots, carry Polariton Technologies affiliations.

What the sequence supports stops there. This result was public before the acquisition was announced. What Marvell actually reviewed during diligence is internal, and dates alone are not a basis for guessing. I covered the acquisition itself and the material question inside the slot in The Truth Behind Marvell-Polariton: Is LWLG the First Candidate, or Just a Backup Bet?, so this piece picks up what the company has said in the four months since.

2. Two months later, the company drew its own map

On June 17, a post explaining plasmonic technology went up on the Marvell blog [3]. Two authors: Claudia Hoessbacher and Wolfgang Heni, listed as Senior Director and Director of Optical Engineering at Marvell. The Polariton founders writing a technical document on Marvell letterhead.

The device description is specific. It explains compressing light below the diffraction limit at a metal-dielectric interface, and states that prototypes measure roughly 10 micrometers in length, 300x to 500x shorter than existing silicon photonics devices, and operate at 1 THz [3]. There is a line on manufacturing too: modified silicon processes with the metal dielectrics added toward the end of production.

Then come the applications. This is the part that matters.

“Plasmonic-based SiPho engines for co-packaged optics for scale-inside or scale-up applications could deliver the same or more bandwidth as conventional devices while minimizing space and power...”

“plasmonic-based SiPho engines for scale-across networks could be used to develop compact, ultra-fast and energy efficient ZR/ZR+ class modules...”

The same technology gets placed on co-packaged optics inside the package and on coherent links between campuses. The April acquisition release named only scale-across, DCI and the ZR family [2]. Two months on, the technical document widened the range upward. It also adds a new class of modules built around massively parallel plasmonic arrays as a possibility, and mentions modules at 3.2T and beyond [3].

The tenses tell you something else. Every sentence describing an application uses could. Able to, not doing. One sentence is in the past tense.

“High-speed SiPho devices capable of operating at 400G/lane containing plasmonic transponders were delivered to customers in 2026.“

400G per lane plasmonic transponders shipped to customers in 2026 [3]. No customer names, no volumes, no distinction between samples and production units. But it is the one completed event sitting among the conditionals.

The same post also sums up what is left.

“The technology is still in the early days of development for commercialization. In the next phase, the focus is on improving the reliability and performance of the devices as well as manufacturing methodologies to make the devices at scale.”

The company calls its own technology early stage for commercialization, and puts reliability first among the next-phase problems. That the company picked that word itself comes back later

3. What 1,400 meters proved and what it did not

With the map in hand, back to the paper. What the April result actually proved is narrower than it looks.

The application is different. The paper states its target as fronthaul and backhaul for 6G mobile networks [1]. Linking base stations over radio instead of fiber. Not datacenter interconnect, and there is no passage in the paper that frames DCI as the target.

The role the device plays is different too. Here the modulator is not loading data onto light at the transmit end. It is an up-mixer at the receive end, lifting an incoming radio carrier into the optical domain. Which changes the drive conditions completely. The paper puts the voltage amplitude at the modulator input at 0.3 V, with the sub-THz power reaching the modulator at minus 5.6 dBm [1]. With a Vpi of 12.3 V, that is small-signal operation at less than a fortieth of Vpi.

This distinction matters because taking Vpi 12.3 V and turning it into “the drive voltage is high, so the power budget is bad” gets it wrong. In this experiment the device was never swung anywhere near Vpi. Receive sensitivity was the thing that mattered. Try to use the same device as a transmit-side data modulator and Vpi comes back as a real burden. Change the role and the number means something else.

Insertion loss is 17.5 dB fiber to fiber [1]. Optical power drops to roughly a sixtieth. The paper reports this as a total and does not break it into components. Metal absorption in the plasmonic structure and the coupling loss of pushing light from a fiber into a slot tens of nanometers wide are both in there somewhere, but how much each contributes is not something this paper answers. In the lab you cover that with optical amplification. In a product, that amplifier shows up on the bill as cost, power and noise.

One comparison deserves care. This paper says thin-film lithium niobate gives low optical insertion loss but that its electro-optic bandwidth drops beyond 100 GHz [1]. That is the plasmonics camp talking, and there is work pointing the other way. A November 2025 paper in the same journal, on thin-film lithium niobate over a quartz substrate, reported 145 GHz at 3 dB and 310 GHz at 6 dB [4]. So 100 GHz is not a wall that cannot be crossed.

Which puts plasmonics somewhere other than winning a bandwidth race. Its claim is the same bandwidth in a far shorter device. The company blog saying 300x to 500x shorter is pointing at exactly that [3].

That is as far as public material takes us. But once the June blog widens the applications in both directions, the targets start to overlap with the other optical platform this company bought four months earlier for considerably more money.

4. So where does Celestial AI’s SiGe stand

Put Marvell’s optical acquisitions in order and there are two inside ten months.

Celestial AI was announced on December 2, 2025, structured as roughly $3.25B upfront with up to $2.25B in earnout [5], and closed on February 2, 2026 [6]. The object of the release headline is scale-up connectivity. Then Polariton on April 22, 2026 [2].

The modulation methods are genuinely different.

Celestial AI’s Photonic Fabric does not use micro-rings. The company-side paper writes down why: narrow optical bandwidth and too much temperature sensitivity to sit next to a hot chip. It uses germanium-silicon electro-absorption modulators instead [7]. Apply a voltage and the material changes how much light it swallows. Simple structure, small footprint, easy to pack densely inside a package. I followed how far that company has actually gotten through two of its papers in Marvell Bought Celestial AI. How Far Along Is Photonic Fabric?.

Polariton’s POH does not swallow light. It rotates phase. Light is confined in a narrow gap between metal walls and that gap is filled with an organic electro-optic material working through the Pockels effect.

So the two platforms run on different physics. Celestial AI has nothing to do with plasmonics. No organic slot, no Pockels effect.

If you’d like to support my independent research and creative journey, please consider a “pledge.” Your support keeps the photons moving.

What overlaps is the target application rather than the technology. Marvell bought Celestial AI for scale-up connectivity, per the release headline itself [5], and the June blog places plasmonics on scale-inside and scale-up CPO as well [3]. Two technologies built on entirely different physics have both become placement candidates for the same scale-up slot.

One caution here. Do not shrink the acquisition down to a modulator. Celestial AI came with an optical fabric architecture, packaging, system IP and a team. Reading this overlap as “one of them gets dropped” is premature, because the system layer and the device layer are separate problems.

Three things are still worth recording.

One. The two technologies get explained separately. The June plasmonics post never mentions Celestial AI or Photonic Fabric [3], and the Photonic Fabric material shown at FMS in August talks about package, server and rack scale [9]. Reach figures differ between company documents, so I will leave those numbers out here. Product blogs sticking to their own subject is ordinary, so the silence itself should not carry much weight. What remains is that the two ranges the company drew for itself meet at scale-up.

Two. The August 27 earnings release contains none of the words optics, photonics, plasmonic, Celestial AI, scale-up, scale-across or coherent [8]. The quarterly report filed the next day reads differently. Listing what drove data center revenue growth, it puts electro-optics first, and it describes the Celestial AI acquisition as a Photonic Fabric technology platform purpose-built for next-generation scale-up interconnect [10]. So the optical business is already inside the consolidated statements. The same document also says this.

“Revenue and earnings of Celestial since the acquisition date were not material”

Revenue and earnings since the acquisition were not at a level worth calling out. The precise reading is that it came into the consolidation but was not large enough to break out. Writing that it does not touch the income statement gets it wrong, and reading a revenue figure out of it gets it wrong too. Polariton and plasmonics, meanwhile, appear nowhere in that same quarterly report [10].

Three. The timing guidance has a different character on each side. For Celestial AI the company guided meaningful revenue contribution to the second half of FY2028. On the plasmonics side there is no such guidance, and instead there is a completed sentence about customer deliveries in 2026 [3]. Since that delivery is not described as samples or production units, and not dated relative to the acquisition, the two cannot be lined up to say which came first. What can be said is that one side has a shipment on record and the other has a revenue date.

Put those three together and a picture of layers divided by design does not emerge. It looks more like something not yet settled, and where it gets settled is not bandwidth.

5. Two blanks left in the material layer

The word the company picked for its next-phase focus was reliability [3]. Start with why that word is heavy.

POH only works with an organic electro-optic material in the slot. A strong field stands the molecules up in one direction, and the Pockels effect holds as long as that alignment does. Time and heat are the problem. Near the glass transition temperature the alignment starts to relax, and the coefficient comes down with it. Add moisture and oxygen and the molecules themselves degrade. That neighborhood is roughly where organic EO polymers kept failing to clear the commercialization bar.

The first blank is the name of that material.

The backbone of the May 1 piece was a single acknowledgment line. The record device published in Optica in 2025, the one reporting 997 GHz at 3 dB, credits Lightwave Logic for supplying Perkinamine chromophore series 3 [11]. Primary material showing that the highest-performing demonstration of the device Marvell bought ran on a listed company’s product.

So I went looking at the same spot in everything published since.

The acknowledgments in the April Nature Communications paper list the Swiss defense procurement agency, three European research projects, NTT Electronics for the photodetector, the SwissChips project and the cleanroom team [1]. No material supplier. The body calls the device an organic-plasmonic Mach-Zehnder modulator and leaves the material unnamed.

The June Marvell blog goes exactly one step further and stops.

“Plasmonic waveguides consist of a metal-dielectric channel filled with an active Pockels material...”

Filled with an active Pockels material. So saying the material is missing entirely would be wrong. What is missing is everything below that. No chemical family, no product name, no supplier. The word polymer does not appear. Neither does chromophore, and neither does any company name. Written by the people who know that material better than anyone.

Open the other side’s books, though, and there is a name.

Page 16 of the investor presentation Lightwave Logic put out in August lays out five Stage 3 prototype customers, and the first slot reads Polariton / Marvell [14]. The other four say only Fortune Global 500. The company put the acquired partner and the acquirer side by side in its own material.

So the asymmetry comes out like this. Marvell discloses down to the material class and stops before the supplier and the product name. LWLG discloses the customer name and stops before the formulation and the bill of materials that actually go into the program. Each side published exactly as far as it is comfortable publishing. Which means the sentence “Marvell uses LWLG material” is still not completed by either side’s documents.

None of that changes the fact that POH is not tied to one material. NLM Photonics in Seattle has its own paper with Polariton and ETH, and that material has been reported with a higher coefficient [12].

The second blank is which layer the reliability data sits in.

Here too the company drew the line itself. Page 13 of the same presentation is titled around reliability performance exceeding the industry standard, and one line under it reads [14]:

“Material Reliability Challenges Addressed / Device Reliability In Progress“

Material level done, device level in progress. The same page walks through four failure modes for EO polymers one at a time. Loss of poling gets a projected lifetime beyond 100 years at 85 degrees C. Chromophore decomposition shows none observed at 85 degrees C and 85 percent relative humidity. This 85/85 condition first appeared in a separate release in July 2025, which went as far as publishing 1,000 hours on encapsulated thin films with an average absorbance loss of 1.6 percent [13]. Photolysis is marked not applicable for applications where the polymer sees no photons at 326 nm or below, and photo-oxidation is described as unlikely with the proprietary encapsulation. A separate slide mentions more than 2,000 hours of damp heat stress testing under AI datacenter temperature and humidity conditions [14].

So saying there is no reliability data would be wrong. There is quite a lot of it. It comes with two qualifiers though. These are the company’s own claims rather than third-party verification, and the test articles are encapsulated thin films and test structures. The last line on that same page says chip and device level reliability testing to validate foundry-specific processes and device-level encapsulation is ongoing [14].

Which matters because the reliability Marvell named as its next-phase problem in June is exactly this back half [3]. An encapsulated thin film holding up in an environmental chamber and the same material holding up after it has been through a foundry process, gone into a slot, been packaged and had light running through it are different tests. And that stretch cannot be answered by the material company alone. Whoever owns the process and the packaging has to be in it.

To put it together: each side published its own half, and reliability has answers only at the material layer.

Sitting at Stage 3 is also not a guarantee of production. The company writes as much in its own quarterly report [15].

“There can be no assurance that any given customer program will result in royalty-bearing production.”

“We do not currently expect significant revenue from volume commercial production of customer products until 2027 at the earliest.”

The 12 to 18 months the presentation gives for Stage 3 is a typical duration, not a conversion rate [14]. The same filing says that revenue recognized during 2026, if any, would relate primarily to material supply, NRE arrangements and prototype and development activities [15].

One more thing to record. The names Polariton and Marvell appear in the presentation but not in the quarterly report. I checked the full document. Investor decks and statutory filings run on different disclosure standards, so this is not strange in itself, but it is worth keeping track of which document a name came from.

6. One sentence from August 11

On the August 11 second quarter call, CEO Yves LeMaitre described the market this way [16].

“The practical limits of building ever larger data centers are forcing operators to connect multiple campuses located a few miles to tens of miles apart into one virtual data center.”

“This requires extensive use of ultra-high-speed optical links, dense wavelength division multiplexing, and coherent modulation.”

He added that while the AI conversation has focused on scale-up and scale-out, scale-across is making a strong comeback. That three-layer split is the same frame PhotonCap used back in July in One Layer Below SemiAnalysis’s Meta Map: Scale-Across Is Coherent, Not CPO. Then came the actual news of the quarter.

“This quarter, a new unnamed Fortune Global 500 customer moved into stage 3 of our design win process.”

“This new customer program is initially focused on scale-across transceivers using coherent modulation.”

That brings Stage 3 programs to five. Foundries are GlobalFoundries, Tower and Silterra plus one undisclosed partner, with two sets of wafers due in August and a third foundry’s wafers in the fourth quarter. One material supply and licensing agreement is signed, and volume production with the furthest-along customer is targeted for the second half of 2027 [16]. Quarterly revenue per the filing is $32,751, against $25,605 a year earlier. That revenue comes from the license agreement, and one customer accounts for more than 10 percent of the total. The company believes its cash carries operations through at least December 2027 [15].

Now do some arithmetic. The entry threshold for the 2026 Fortune Global 500 is $33.2 billion in revenue. Marvell’s most recent quarter came in at $2.739 billion [8], so annualizing puts it around $11 billion. The customer that moved into Stage 3 this quarter is therefore not Marvell.

Hold that threshold up against the optical industry and something more interesting happens. Every company whose main business is coherent falls below the cut. Ranked by most recent fiscal year revenue [17]:

Nokia 2025, EUR 19.9B, roughly $23.8B
Corning 2025, $15.6B
Marvell, annualized from the latest quarter, around $11B
Coherent FY2026, $7.1B
Ciena FY2025, $4.8B
Lumentum FY2026, $3.0B

Nokia leads that list at $23.8B. Even after swallowing Infinera it lands nearly $10 billion short of the cut. So this customer sits outside the coherent pure plays entirely. The condition becomes a company with a foot in coherent and far more revenue coming from somewhere else. A few groups fit.

First, systems companies that own coherent products. Cisco is the obvious one. It is on the Global 500 list, and through Acacia it designs and sells coherent pluggables directly [18]. It has both the reason and the ability to put a new modulator material into ZR and ZR+ modules. On technical fit alone it is the closest.

Second, the hyperscalers. Amazon, Alphabet, Microsoft and Meta all sit high on the list [18]. Scale hardly needs discussing, and the phrasing LeMaitre chose points this way anyway. Operators tying campuses into one virtual datacenter is a description of those four. They also have a long record of driving component roadmaps themselves.

Third, semiconductor and diversified companies with optical parts inside. Broadcom is on the list and carries data converters for coherent transport in its portfolio. Mitsubishi Electric has an optical device business, and as it happens it co-authored a paper with the Leuthold group at this year’s ECOC [19]. That is not directly tied to the material question, but it is a record of existing contact with that lab. Hon Hai and Samsung Electronics, with large optical module and component volumes, and NVIDIA and TSMC, which design or build optical interconnect directly, all clear the revenue condition too.

One name I would put at low odds. Huawei has the scale and a complete coherent stack, but running a multi-year design win program with a company based in Colorado does not look easy under the US export control environment. I have not checked which specific items fall under which rule, so I will not state it as settled.

That is as far as public material narrows things. There is no basis for naming any single company, and the grouping above is a qualitative estimate built from revenue and business mix. LWLG’s own filing describes its customers and prospects as semiconductor foundries, silicon photonics device designers, optical module manufacturers, and system integrators serving AI, cloud, datacenter and telecom markets [15], so the field cannot be pinned to three categories either. The company did not give a name and I do not know it. The direction is clear enough though. LWLG has a second scale-across vector separate from the Marvell one, and the counterparty is not a company whose main business is coherent.

I should also record the silence on that same call. Going through the full transcript, plasmonics, Polariton and Marvell never come up [16]. On the March call cited in the May 1 piece, the CEO called Polariton a longtime customer and partner. Five months later the name is gone from the spoken remarks while it sits on slide 16 of the deck. Whether that reflects caution about naming an acquired company in public or a change in the weight of the relationship is not something current material settles.

7. The ECOC 2026 checklist

ECOC 2026 runs September 20 to 24 at FYCMA in Malaga, Spain. The exhibition runs the 21st through the 23rd.

I went through the conference programme directly [19]. Searching by company name returns zero for Polariton, but the programme only displays the affiliation of the presenting author, so co-author affiliations are never indexed. Searching by person turns one up.

#418 / We3-D2, “Resonant Plasmonic Optical DAC for Low-Power IM/DD Links” Wednesday September 23, 14:15, room C1.1

Wolfgang Heni and Benedikt Baeuerle are on the author list [19]. The presenter is David Moor, affiliation shown as ETH Zurich. The abstract describes optical digital-to-analog converters built on resonantly enhanced plasmonic devices, reaching up to 336 Gb/s per polarization per wavelength.

The grain runs a little differently here. This paper is IM/DD, not coherent. Light intensity modulated across multiple levels and directly detected, with low power and reduced processing complexity as the selling points. That reads differently from the ZR and ZR+ language the June blog attached to the coherent side.

Do not turn that into a placement claim, though. The abstract gives no transmission distance and no target system. IM/DD is used in scale-out and inside the package for CPO alike, so this paper on its own cannot say which layer. What it does say is that the result going to the conference is not on the coherent side.

The exhibition has to be checked separately. ECOC splits the conference and the exhibition across two different sites, so searching only the papers misses half of it. The exhibitor list has Lightwave Logic registered at stand 1461 [20]. Marvell is at 2002, Ciena 1045, Nokia 1100, Lumentum 2050, Coherent 2100. Polariton is not on the list.

Two of those companies also take the Product Focus stage inside the exhibition hall [21]. This is the highest-value item on the checklist.

Monday September 21, 11:55, Marvell, Lian Qin. The talk is titled “Near-Packaged Optics for AI Scale-Up Networks: Comparing with Pluggable Optics and CPO.”

Read the title and it lands right on this article’s question. Scale-up networks, with near-packaged optics compared against pluggables and CPO. The June blog put plasmonics on scale-up CPO in a conditional, and this talk has the company comparing, out loud, which approach takes that slot. It also puts a third option in the title.

Tuesday September 22, 16:05, Lightwave Logic, Robert Blum. Titled “Scaling Silicon Photonics Platforms: Foundry PDKs and Ecosystems for 400G and Beyond.” Foundry PDKs and ecosystems, so not a place where a material name is likely to surface, but it does show how the company talks about progress across its four foundries on a public stage.

The two talks land a day apart on the same stage. And there is precedent: at ECOC in September 2024, LWLG and Polariton jointly demonstrated a packaged 110 GHz device and put out a release together.

Marvell itself is not on the programme as a technical paper author, but it has two more workshop slots, both Lenin Patra [19]. September 20 at 15:06 in a workshop on channel design, and at 16:00 the same day presenting “Next-generation DSP trends for the AI ecosystem,” with Ciena’s Kim Roberts on datacenter coherent immediately after.

So four things to check across the third and fourth weeks of September.

One. Which of near-packaged optics, CPO and pluggables plasmonics gets attached to in Qin’s talk on the 21st. This is the first time the company narrows a slot it opened as a conditional in June. Settle the scale-up placement there and Celestial AI’s position reads alongside it.

Two. Whether a slot material is named in the body and the acknowledgments of We3-D2. A supplier credit like Optica 2025 keeps the May 1 connection alive. A blank like the April paper and the June blog extends the undecided state one more round. A different company name means the argument needs revising. The conference guidance says papers are posted to Optica Proceedings without specifying a release date, so I plan to check again around September 23. And a talk happening does not guarantee a material name appears, so this box may simply stay empty.

Three. Which layer plasmonics gets attached to in Patra’s two workshop talks. The same question as the first item, checked on a different stage.

Four. Whether a joint demonstration or release comes out of stand 1461 and Blum’s talk on the 22nd, and if so, whose name is on it. In 2024 that slot belonged to Polariton.

8. Closing

A look back at an earlier call first.

In the May 1 piece I wrote that what Marvell bought was the slot, and that the material going into it sits outside. Four months on that holds. The April paper, the June company blog and the September conference paper are all about the plasmonic structure, and nothing has surfaced showing Marvell making the material in house.

One thing I could not see then has been filled in. At the time an academic acknowledgment was the only link. In the August material, LWLG has Polariton and Marvell written into its own books as a Stage 3 customer [14]. That is a record of the relationship moving from academic collaboration to a product program. Only one side keeps those books, though, and the buying side still says nothing.

One thing I had wrong needs correcting on the record. In the first draft of this piece I read Marvell as having bought two modulators split by layer. Reading the June blog, I dropped that. The company puts plasmonics both inside the package and between campuses. The layers do not divide. They overlap.

So the current state comes out like this. Marvell holds two optical platforms whose targets overlap at scale-up, and the company’s technical documents do not explain the relationship between them. The plasmonics side has a 2026 customer delivery on record, with no disclosure of whether those were samples or production units, or whether they shipped before or after the acquisition. Reliability is the problem the company named next for that side. The supplier claims the material-stage challenges are addressed while chip and device reliability testing runs on. And the counterparty’s name lives in that supplier’s investor deck rather than in its statutory filing.

This reading changes if a few things surface. Marvell narrowing plasmonics to a single layer would settle the overlap as a transitional phase. Material brought in house would cut the ground out from under exposure through a listed material supplier. Packaged device-level high-temperature reliability results arriving with a supplier name attached would close the checklist and move the discussion to product schedules instead.

The dates run like this. September 21 for Qin’s Product Focus talk, the 20th for two Marvell workshops, the 22nd for Blum’s talk, the 23rd for the paper with Heni on it and the last day of the exhibition. After that week, LWLG’s third quarter results show how the five Stage 3 programs progressed and whether the fourth quarter foundry wafers arrived. The second half of 2027 production target is something to judge after that.

I still like both of these acquisitions. What changed is the reason. Rather than a design that split the layers cleanly, this is one company holding two overlapping options at scale-up. Overlapping does not mean one has to go, since dividing them by system boundary and running both is entirely possible. So what I am watching is placement rather than a winner. Where the company puts what, and whether material and reliability data follow into each slot. That answer is not out yet, which is why this piece ends on a checklist.

References

[1] B. Vukovic et al., “Plasmonic modulator enabling kilometer-range high-throughput sub-THz links for radio access networks”, Nature Communications 17, 5384 (2026-04-18)

[2] Marvell, “Marvell Announces Acquisition of Polariton Technologies”, 2026-04-22

[3] C. Hoessbacher, W. Heni, “Plasmonics: A Path to Higher Bandwidth in Optics in the AI Era”, Marvell blog, 2026-06-17

[4] “Monolithic lithium niobate photonic chip for efficient terahertz-optic modulation and terahertz generation”, Nature Communications, 2025-11-24

[5] Marvell, “Marvell to Acquire Celestial AI”, 2025-12-02

[6] Marvell, “Marvell Completes Acquisition of Celestial AI”, 2026-02-02

[7] D. Oh et al., “Photonic Fabric Interconnect for a Scale-up Network Solution in Accelerated Computing,” 2026 IEEE 76th ECTC, pp. 862-867. DOI 10.1109/ECTC51846.2026.00146

[8] Marvell, “Second Quarter of Fiscal Year 2027 Financial Results”, 2026-08-27

[9] Marvell, “Photonic Fabric Technology: How Optical Connectivity Enables AI Infrastructure”, Marvell blog

[10] Marvell Technology, Form 10-Q, quarter ended 2026-08-01, filed 2026-08-28

[11] Y. Horst et al., “Ultra-wideband MHz to THz plasmonic EO modulator”, Optica 12(3):325-328 (2025)

[12] H. Xu et al., “Ultrahigh Performance Cross-Linkable Organic Electro-Optic Material for Hybrid Modulators”, Chemistry of Materials (2025)

[13] Lightwave Logic, “Perkinamine Polymer Reliability Breakthrough: Passes Telcordia 85/85 Test”, 2025-07-15

[14] Lightwave Logic, Investor Presentation, August 2026. Quoted here from page 13 on reliability and page 16 on Stage 3 customers

[15] Lightwave Logic, Form 10-Q, quarter ended 2026-06-30, filed 2026-08-14

[16] Lightwave Logic second quarter 2026 earnings call transcript, 2026-08-11. Secondary source

[17] Most recent fiscal year revenue per each company’s own results announcements. Marvell is the quarterly figure in [8] annualized

[18] Fortune, Global 500 2026, announced 2026-07-28

[19] ECOC 2026 Conference Guide PDF. We3-D2 on p.122, the Marvell workshops on pp.47 and 48

[20] ECOC Exhibition 2026 exhibitor list, retrieved 2026-08-30

[21] ECOC Exhibition 2026 Product Focus timetable, retrieved 2026-08-31

? The ECOC programme and exhibition material are as retrieved on the dates shown and can change before the show opens. The We3-D2 co-author list does not appear in the Conference Guide and is shown only in the online session browser (retrieved 2026-08-30).

? The acknowledgments and any material designation in We3-D2 cannot be checked at the time of writing, since the full paper is not public before the conference.

? The customer candidate grouping is a qualitative estimate and has not been confirmed by any company.

This article is an independent technical analysis published by PhotonCap, based on an engineering perspective. All content is derived from publicly available information and is intended solely for educational and informational purposes. In other words, nothing in this material should be construed as a recommendation to buy, sell, or hold any specific securities. Please note this carefully.

The author may hold positions in the securities mentioned herein and reserves the right to trade such securities at any time without prior notice. Readers should conduct their own thorough review and research before making any investment decisions.

Oh nooooo!!!!

conventional thinking = ARS & SL... DNC operatives!

CNBC’s Rick Santelli to Retire After More Than 27 Years
The longtime markets correspondent will remain a contributor at the network
thewrap.com

A.J. Katz
September 4, 2026 @ 8:24 AM

Rick Santelli, the longtime CNBC markets correspondent whose fiery reports from the trading floor made him one of the business network’s most recognizable personalities, is retiring from his full-time role after more than 27 years.

His last day on air will be Oct. 2, TheWrap has learned, though he plans to return as a contributor after his retirement. His departure was announced Friday morning on “Squawk Box.”

Santelli officially joined CNBC in 1999 after first appearing as a guest on the network in 1994. He became a fixture of CNBC’s coverage of financial markets, Federal Reserve decisions and major economic reports, reporting primarily from Chicago.

“Rick’s passion and personality have led to countless memorable exchanges,” CNBC editor in chief David Cho wrote in a memo to staff on Thursday. “Whatever the subject, Rick has always approached it with the same energy, authenticity and willingness to challenge conventional thinking that have made him such a distinctive presence on CNBC.”

Before joining CNBC, Santelli spent years in financial markets, “beginning at the Chicago Mercantile Exchange in 1979 and later holding roles in futures, options and institutional trading,” per Cho.

Santelli’s profile extended well beyond business news in 2009, when his impassioned on-air criticism of the Obama administration’s housing rescue plan became associated with the emerging Tea Party movement. Speaking from the floor of the Chicago Mercantile Exchange, Santelli criticized government assistance for struggling homeowners and called for a “Chicago Tea Party.”

He also became known for heated exchanges with his CNBC colleagues. In 2020, Santelli clashed on air with “Squawk Box” co-anchor Andrew Ross Sorkin during a discussion about COVID-19 restrictions.

Santelli’s retirement comes during a period of change at CNBC following parent company Versant’s official separation from Comcast/NBCUniversal earlier this year. The network continues to reshape its newsroom and expand its digital operations, tapping The Information veteran Ken Brown as managing editor of digital and editorial strategy.

CNBC plans to celebrate Santelli’s career on air in the weeks leading up to his departure. His final day will fall on the first Friday of October, a day Santelli traditionally spent covering the federal government’s monthly jobs report.

[#Breeze] Kirk, here's a LRCX Wave Count Update. If that alternate (W)-(X)-(Y) is the coun...

#Breeze — 9/4/2026 4:15:44 PM

Kirk, here's a LRCX Wave Count Update. If that alternate (W)-(X)-(Y) is the count then the correction is over. Breeze

[#Breeze] tocco, here's a LULU Wave Count Update. Not much changed since my last update. B...

#Breeze — 9/4/2026 4:08:32 PM

tocco, here's a LULU Wave Count Update. Not much changed since my last update. Below a 76.8% retracement from the ATH. Hasn't recovered that line. OOCH. Breeze

[#Breeze] UPRO Wave Count Update. Minor Wave 5 of Intermediate Wave (3) is in play with a ...

#Breeze — 9/4/2026 4:04:22 PM

UPRO Wave Count Update. Minor Wave 5 of Intermediate Wave (3) is in play with a target of 205. Breeze

[#Breeze] GEV Wave Count Update. Minute Wave (circle) v of Minor Wave 3 is in play with a ...

#Breeze — 9/4/2026 3:35:18 PM

GEV Wave Count Update. Minute Wave (circle) v of Minor Wave 3 is in play with a target of 1,448.55. Breeze

[Augustus Gloop] I'm not a bear but I'm going to be cautious here. I'm waiting for a hanging fas...

Augustus Gloop — 9/4/2026 10:56:04 AM

I'm not a bear but I'm going to be cautious here.

I'm waiting for a hanging fastball

[Augustus Gloop] ???????

Augustus Gloop — 9/4/2026 10:45:00 AM

Wow, for math geeks.... 30303

LRCX down channel.... eventually a pennant?

bears may still want 61.8 retrace and fill da gap unless...
.

[Augustus Gloop] I went back to the well on SOXL earlier this week. It went down and below the 2...

Augustus Gloop — 9/4/2026 10:40:04 AM

I went back to the well on SOXL earlier this week. It went down and below the 200 and (based on a P&F) did a low pole reversal that appeared to me as if it would require 126 to break.

I'm a long term bull on chips but I didn't want to get caught in a downdraft on a triple cripple. Long story short - I took 117.12 for a minor gain. I'm still showing a bearish PO of 79 (P&F) so I'd rather buy a real breakout than toy around at this time of year.

Interesting that the $SOX itself never got close to its 200-dma.

Message #30300 from Kirk © at 9/4/2026 10:02:38 AM

SOXL looks like it made a double bottom as it tested the 200-dma twice with an appropriate time between incursions (about a month) to shake most out of the tree.

And the $SOX made a higher low...

.

SOXL looks like it made a double bottom as it tested the 200-dma twice with an appropriate time between incursions (about a month) to shake most out of the tree.