Silicon Investor - The end of Moore's law - Poet Technologies

[longz] AGORACOM Re: AGSM on 26th of June - SEDARIn Response To AGSM on 26th of June -...

longz — 4/10/2026 5:41:02 PM

AGORACOM

Re: AGSM on 26th of June - SEDARIn Response To AGSM on 26th of June - SEDAR by abc-sch
Posted On: Apr 10, 2026 02:16PM

Interesting, an AGSM instead of just an AGM. Is Mazann's suspicion of a huge buyout offer coming to fruition, or could it be something as small as the board asking for approval of equity incentive plan changes? Either way, they need shareholder votes for something that needs the special shareholder approval.

POET:NAZ | AGSM on 26th of June - SEDAR | POET Technologies Inc. | wateroflife-2463128

[longz] Interesting, an AGSM instead of just an AGM. [graphic]

longz — 4/10/2026 5:39:48 PM

Interesting, an AGSM instead of just an AGM.

[longz] (3) Post | LinkedIn

longz — 4/2/2026 1:50:28 PM

[longz] ALL ABOARD! ....CHOO! CHOO!

longz — 4/2/2026 1:50:16 PM

[toccodolce] I hope so. Marvel bought Celestial AI which was a $POET customer. $NVDA has...

toccodolce — 3/31/2026 6:27:03 PM

I hope so.

Marvel bought Celestial AI which was a $POET customer.

$NVDA has invested $2 billion each of some of the top photonics companies such as $COHR and $LITE. Now they invest $2 billion with $MRVL.

Tocco

[RWS] Does this relate to POET? Thanks! RWS

RWS — 3/31/2026 6:02:15 PM

[toccodolce] [X] We’re excited to announce a strategic partnership with NVIDIA to deliver gr...

toccodolce — 3/31/2026 9:03:36 AM

[X]

We’re excited to announce a strategic partnership with NVIDIA to deliver greater choice and flexibility for customers building next-generation AI infrastructure.

Through this collaboration, Marvell will provide custom XPUs and NVLink Fusion™ compatible scale-up networking,… pic.twitter.com/PKWOl5aUI2
— Marvell Technology (@MarvellTech) March 31, 2026

[/X]

[X]

https://t.co/Aq6C1RpAIF
— Damnang2 (@damnang2) March 15, 2026

[/X]

[RWS] why?? RWS

RWS — 3/24/2026 8:11:48 AM

[longz] March 30/31....fireworks?.......

longz — 3/24/2026 3:45:49 AM

[toccodolce] [X] NEWS: POET Tech and LITEON Announce Joint Development of Optical Modules fo...

toccodolce — 3/16/2026 12:54:44 PM

[X]

NEWS: POET Tech and LITEON Announce Joint Development of Optical Modules for #AI Applicationshttps://t.co/02XDnlu2ev
— POET Technologies (@POETtech) March 16, 2026

[/X]
POET Technologies and LITEON Announce Joint Development of Optical Modules for AI Applications

SAN JOSE, CA, March 16, 2026 –– POET Technologies Inc. ("POET" or the "Company") (NASDAQ: POET), a leader in the design and implementation of highly-integrated optical engines and light sources for artificial intelligence networks, today announced a strategic collaboration with LITEON Technology, one of the world’s leading providers of optoelectronic semiconductor components and high-power optical systems. The partnership aims to co-develop next-generation optical communication modules built on POET’s patented optical interposer technology and integration platform.

The jointly developed optical engine will leverage the POET Optical Interposer™ to integrate several optical components, drive electronics, and coupling structures into a compact, thermally optimized module. This approach enables scalable, cost-efficient production of advanced optical modules for next-generation co-packaged optics, AI systems, and high-bandwidth data center applications.

“LITEON has long been at the forefront of optoelectronic semiconductor innovation, delivering solutions spanning optoelectronics to high-power AI and HPC infrastructure,” said Sander Su, General Manager of the Optoelectronic Products Solutions Business at LITEON Technology. “By partnering with POET, our capabilities are extended into advanced photonic solutions that will enable the next generation of high-performance connectivity. Together, we aim to provide high-end laser solutions tailored for AI-driven data centers. We look forward to combining our respective manufacturing expertise and technological strengths to deliver exceptional value to customers worldwide.”

“This collaboration with LITEON is a major step forward in advancing optical connectivity for next-generation architectures,” said Dr. Suresh Venkatesan, Chairman & CEO of POET Technologies. “LITEON’s leadership in optoelectronics and precision manufacturing perfectly complements POET’s integration platform. Together, we are positioned to deliver highly scalable, power-efficient solutions that address the rapidly accelerating needs of AI and hyperscale data center customers.”

POET will begin development work this year and expects to have prototypes ready by late 2026. High-volume production is anticipated for 2027. POET is scheduled to attend the 2026 OFC Conference and Exhibition (March 16-19 at the Los Angeles Convention Center) and its representatives will be available for discussions on its products and business development. The Company’s booth at the show is No. 339.

[Selectric II] POET to Demonstrate Extraordinary Hybrid Laser and Next-Generation High-Powere...

Selectric II — 3/12/2026 6:02:58 AM

POET to Demonstrate Extraordinary Hybrid Laser and Next-Generation High-Powered Light Sources for AI Applications at OFC Conference

POET Technologies Inc.
Tue, March 10, 2026 at 7:39 AM EDT 5 min read

In this article:

POET
+6.29%

POET Technologies Inc. With commercialization of its advanced optical engines under way, the Company intends to reveal its latest product advancements to industry leaders

Blazar

4Channel Blazar · GlobeNewswire Inc.

SAN JOSE, Calif., March 10, 2026 (GLOBE NEWSWIRE) -- POET Technologies Inc. ("POET" or the "Company") (NASDAQ: POET), a leader in the design and implementation of highly-integrated optical engines and light sources for artificial intelligence networks, today announced it plans to perform live demonstrations of its two leading external light source (ELS) products at the annual Optical Fiber Communications (OFC) Conference, which will be held at the Los Angeles Convention Center from March 16-19, 2026.

POET’s executive management team and engineers will be meeting customers and industry analysts at the Company’s Booth No. 339 during the exhibition.

The products to be demonstrated include:

POET Blazar™: A ground-breaking ELS solution, Blazar is a highly integrated hybrid laser that exemplifies the Company’s “semiconductorization of photonics” mission. Designed to power both co-packaged optics (CPO) and high-bandwidth, chip-to-chip, light-based data communications links, Blazar has rapidly advanced since it was first showcased in private meetings at last year’s OFC Conference. POET’s high-power, multi-channel light source is proving to be a viable alternative to traditional DFB laser-based solutions. Wafer-level chip-scale technology significantly lowers the cost of the light source, provides larger scale and better reliability, and increases the effective supply of Indium Phosphide.

POET plans to demonstrate the high output power and multi-wavelength capabilities of Blazar.

“Blazar solidifies POET’s position as an industry leader in photonics because it proves our focus on constant innovation is resulting in products the industry needs,” POET Executive Chairman and CEO Dr. Suresh Venkatesan said. “Blazar makes AI connectivity much more affordable thanks to an architecture that reduces costs and increases scale and manufacturing efficiency.”

POET Starlgiht™: The next generation of POET’s first ELS solution will be revealed alongside Blazar. The POET Starlight demo will feature a compact engine solution with eight-channel high power lasers at multiple wavelengths. Starlight will be integrated into a working optical engine to show the commercial applicability of the solution to the industry standard ELSFP module.

“Starlight is gaining traction in the industry because of the large demand for more compute power for AI applications. We are excited to reveal the advancements we have made to our latest version of Starlight that achieves advanced packaging and lower cost needed for chip-to-chip data communication links,” said Raju Kankipati, POET’s Chief Revenue Officer.

Lightwave Award: In addition, Company representatives will attend a ceremony to accept the recently announced Elite Score award at the Lightwave Innovations Reviews reception, which is scheduled to be held on March 16. POET earned the recognition for POET Teralight™, the Company’s multiple award-winning line of 1.6Tbps transmit and receive optical engines. In February, Lightwave Innovation Reviews announced its judges had given POET Teralight a score of 4.5, one of the highest received among winning entries.

The 2026 OFC Conference will feature the Company’s largest booth presence in the history of its attendance. Meetings with current and prospective clients, analysts, media, and institutional investors will take place throughout the duration of the exhibition.

About POET Technologies Inc.

POET Technologies is a design and development company offering high-speed optical engines, light source products, and custom optical modules for the artificial intelligence systems market and hyperscale data centers. Its patented POET Optical Interposer™ platform enables seamless chip-scale integration of photonic and electronic devices using advanced semiconductor manufacturing techniques. More information about POET is available on our website at www.poet-technologies.com.

Media Relations Contact:
Adrian Brijbassi
Adrian.brijbassi@poet.tech

Company Contact:
Thomas R. Mika, EVP & CFO
tm@poet.tech

Forward-Looking Statements
This news release contains “forward-looking information” (within the meaning of applicable Canadian securities laws) and “forward-looking statements” (within the meaning of the U.S. Private Securities Litigation Reform Act of 1995). Such statements or information are identified with words such as “anticipate”, “believe”, “expect”, “plan”, “intend”, “potential”, “estimate”, “propose”, “project”, “outlook”, “foresee” or similar words suggesting future outcomes or statements regarding any potential outcome. Such statements include the Company’s expectations with respect to the success of the Company’s product development efforts, the performance of its products, the expected results of its operations, meeting revenue targets, and the expectation of continued success in the financing efforts, the capability, functionality, performance and cost of the Company’s technology as well as the market acceptance, and inclusion and timing of the Company’s technology in current and future products.

Such forward-looking information or statements are based on a number of risks, uncertainties and assumptions which may cause actual results or other expectations to differ materially from those anticipated and which may prove to be incorrect. Assumptions have been made regarding, among other things, management’s expectations regarding the success and timing for completion of its development efforts, the introduction of new products, financing activities, future growth, recruitment of personnel, opening of offices, plans for and completion of projects by the Company’s consultants, contractors and partners, availability of capital, and the necessity to incur capital and other expenditures. Actual results could differ materially due to a number of factors, including, without limitation, the failure of its products to meet performance requirements, lack of sales in its products, once released, operational risks in the completion of the Company’s anticipated projects, risks affecting the Company’s ability to execute projects, the ability of the Company to generate sales for its products, the ability to attract key personnel, and the ability to raise additional capital. Although the Company believes that the expectations reflected in the forward-looking information or statements are reasonable, prospective investors in the Company’s securities should not place undue reliance on forward-looking statements because the Company can provide no assurance that such expectations will prove to be correct. Forward-looking information and statements contained in this news release are as of the date of this news release and the Company assumes no obligation to update or revise this forward-looking information and statements except as required by law.

120 Eglinton Avenue, East, Suite 1107, Toronto, ON, M4P 1E2- Tel: 416-368-9411 - Fax: 416-322-5075.

A photo accompanying this announcement is available at globenewswire.com

finance.yahoo.com

[toccodolce] [X] ‼️Photonics Investors, Read This!$POET$LWLG$ALMU There is a way to play th...

toccodolce — 3/10/2026 8:51:34 PM

[X]

‼️Photonics Investors, Read This!$POET $LWLG $ALMU

There is a way to play the optical super cycle that MIGHT be early.

There are a few companies that have not participated in the run up the last few months.

This is reflective of their early nature and speculative positioning.… https://t.co/yXZtmtCHnB pic.twitter.com/kPXJU9LpN9
— Gaetano (@crux_capital_) March 3, 2026

[/X]

[X]

$POET 에게 상을 준 Lightwave 라는 기관은 광통신(Optical Communications) 및 광학 네트워크 분야에서 35년 이상의 전통을 가진 가장 공신력 있는 글로벌 기술 매체 중 하나입니다.

찾아보니 2025년에는,
Marvell, Huawei, Keysight 등이 수상을 했었네요.

시상식은 광통신 권위 학회인 OFC… https://t.co/cORdOea8T8 pic.twitter.com/UCjaKm4XNt
— Photon Capital (@PhotonCap) February 12, 2026

[/X]

[X]

Education Saturday! ✍️
If you're new to Photonics, start here!$POET $LITE $COHR $AXTI $AAOI

Here is a really easy and simplified way to understand many of the companies that I track in this space

Save/share if you found this useful please!$AXTI - The raw materials

AXT… https://t.co/cjUzKz48GQ pic.twitter.com/fLxgJvZdNP
— Gaetano (@crux_capital_) March 8, 2026

[/X]

[X]

<1 Billion $ALMU $LWLG $POET $ALMU - Aeluma Technologies $260m

Aeluma is shattering the cost barriers of AI and defense by fusing high-speed light technology with standard silicon to mass-produce the next generation of ultra-efficient optical chips. If Aeluma is successful,… pic.twitter.com/58IiCWqeXD
— Gaetano (@crux_capital_) February 17, 2026

[/X]

[toccodolce] [X] The Hybrid Revolution: How $POET Technologies is Breaking the AI Data Barri...

toccodolce — 3/10/2026 12:30:56 PM

[X]

The Hybrid Revolution: How $POET Technologies is Breaking the AI Data Barrier

In the latest issue of PIC Magazine, Dr. Suresh Venkatesan, Chairman & CEO of POET Technologies, lays out a definitive vision for the future of Integrated Photonics. This is not mere speculation; it is… https://t.co/h9KnZtXrR2 pic.twitter.com/sM1GozyY66
— Finn Stockinger (@FinnStockinger) March 5, 2026

[/X]

[X]

“POET explains hybrid integration, optical engines and chiplet-based PIC development in the #AI era” — See the new issue of PIC Magazine (pages 20-22) for the article by our Chairman & CEO Dr. Suresh Venkatesan. https://t.co/f3keEdFCZe #POETpowersAI pic.twitter.com/EBV5osrj3w
— POET Technologies (@POETtech) March 5, 2026

[/X]

[toccodolce] [youtube video]

toccodolce — 2/28/2026 10:37:51 AM

[toccodolce] Latest from Agoracom: timotai Mitsubishi Electric has announced plans to tripl...

toccodolce — 2/15/2026 11:53:03 AM

Latest from Agoracom: timotai

Mitsubishi Electric has announced plans to triple its production capacity for

optical semiconductor devices by fiscal year 2028 (compared to 2024 levels) to meet skyrocketing demand from AI-driven data centers and 5G base stations.
Key details of this initiative include:

Shifting Investment: The company is reallocating investment away from electric vehicle (EV) power devices, which have experienced slower growth, toward high-frequency optical devices.
Targeted Technology: The increased production focuses on Electro-absorption Modulator integrated Lasers (EMLs), which are crucial for high-speed data transmission (200 Gbps to 800 Gbps, with potential for up to 1.6T-3.2T in AI networks).
Production Location: Capacity will be boosted at its high-frequency and optical device works in Hyogo Prefecture, Japan.
Market Position: Mitsubishi Electric is a major player in this sector, with Japanese manufacturers (including Lumentum Japan and Sumitomo Electric) accounting for nearly 70% of the global optical device market. [ 1, 2, 3, 4, 5]

AI Integration in Other Areas

AI-Enabled Laser Processing: Separate from the semiconductor production, Mitsubishi Electric (via MC Machinery) uses AI-enabled fiber lasers (GX-F ADVANCED Series) that feature real-time process monitoring, auto-nozzle cleaning, and enhanced cutting speeds.
AI Data Center Partnership: In November 2025, Mitsubishi Electric signed a memorandum of understanding with Foxconn (Hon Hai Precision Industry) to collaborate on supplying energy-efficient, reliable solutions for AI data centers worldwide. [ 6, 7, 8]

AI responses may include mistakes.

[1] asia.nikkei.com
[2] abc.az
[3] poet-technologies.com
[4] asia.nikkei.com
[5] verifiedmarketresearch.com.
[6] mcmachinery.com
[7] mitsubishielectric.com
[8] youtube.com

[toccodolce] [X] 🚨Photonics Education Saturday!$LITE$POET$COHR$AXTI$ALMU$MRVL Today's topi...

toccodolce — 2/15/2026 11:50:40 AM

[X]

🚨Photonics Education Saturday!$LITE $POET $COHR $AXTI $ALMU $MRVL

Today's topic is Co-packaged Optics (CPO)

If you invest in this space, you need to understand why CPO is, when it is coming, and why it's significant

AI networks are starting to hit a wall where moving the…
— Gaetano (@crux_capital_) February 15, 2026

[/X]

[toccodolce] Latest from Rddt Gabbeb0y The potentially groundbreaking breakthrough Due Dili...

toccodolce — 1/19/2026 9:23:13 AM

Latest from Rddt Gabbeb0y

The potentially groundbreaking breakthrough

Due Diligence

The potentially groundbreaking breakthrough

During the recent months I´ve tried to dig really deep into Poet, joblistings, potential customers/product-chains, current technology-moats, future technology, and I've posted some of it. Now it is time for the grand finale. In this post I would like to talk about complete wafer optical transceiver production and how that would destroy all of Poet's competitors, i.e how that would make all of us shareholders dance and celebrate. And also a little spicy golden nugget at the end of the analysis.

“Wafer scale assembly of optical transceivers”

In a recent interview with the chief revenue officer, there were multiple exciting and bullish statements, but I would like to dig into this one: “And our main goal is this wafer scale assembly.” “We really want to get to a wafer scale assembly of optical transceivers. And we are developing other technologies that assist in that.” “So we are developing technologies that eliminate all these active alignments and multiple steps in that process. So we are, I think, getting there probably this year, we will get to a demonstrable solution that eliminates all these components. So that will be a big deal for the industry because then truly we can do a wafer scale assembly of optical transceivers.”

Here is my view on the possible transformation (since I am no engineer or optical expert, please verify this/extend the explanation):

Current State: "Wafer-Scale Optical Engines" What they make: The "Engine" (The heart).

The Process:

They take a Silicon Wafer (Interposer).

They "flip-chip" the laser and detector onto it (Passive Alignment).

The Bottleneck: They chop this wafer into tiny chips. Then, they send these chips to a factory (like Luxshare) where humans or slow robots must:

Glue the chip into a metal box (transceiver housing).

Add external plastic lenses to focus the light.

Add a fiber optic connector.

Seal the box hermetically (to keep air out).

The Limitation: Even though the chip was cheap to make, the packaging (the metal box and lenses) accounts for 70% of the cost.

2. The 2026 Goal: "Wafer-Scale Transceivers" (The CFO's Vision) What they make: The "Whole System" (Engine + Package + Optics) all at once.

The "Eliminated Components": The CFO said they will "eliminate all these components." Here is specifically what he means:

No More Lenses: POET integrates "Spot Size Converters" directly into the silicon. They guide the light into the fiber without needing an external plastic lens.

No More Isolators: (These are expensive magnets that stop light bouncing back). POET is designing the laser-to-chip interface (with Sivers) so these are often unnecessary.

No More "Gold Boxes": Instead of putting the chip in a metal can, they seal the chip while it is still on the wafer using a glass cap. This is called Chip-Scale Packaging (CSP).

The Result: When they cut the wafer, the device is Done. It doesn't need to be put in a box. It is the box.

Verdict/the key take away:

Passive Alignment (Today): Solves the "Laser Placement" problem. (Great, but incomplete).

Wafer-Scale Transceiver (2026): Solves the "Packaging" problem. (The Holy Grail).

In my opinion Poet is already ahead of its competitors technology-wise because of their passive alignment, “Lego-brick assembly”, production-infrastructure, owning specific, very important machines at Silterra (from what I've learned), reduction of indiumphospide to the only the critical parts(and something about TLFN or what it´s called), speed (as good as competitors, but likely faster in 2 or 3 years). Now, the CRO is saying “So we are, I think, getting there probably this year, we will get to a demonstrable solution that eliminates all these components.”In other words: we are probably going to show you something that will revolutionize the industry this year.

The Sivers connection

Sivers Semiconductors has quietly positioned itself as the indispensable "fuel source" for the next generation of AI hardware, transforming from a niche component manufacturer into a critical platform enabler for the silicon photonics industry. While the market often focuses on the giants like Marvell or the integrators like POET Technologies, Sivers holds the one piece of intellectual property that neither of them can manufacture themselves: the high-power, continuous-wave (CW) laser light required to power optical computing.

The company's recent strategic shift to marketing "Advanced Photonic Wafers" rather than individual laser chips is the most significant clue to their future role. In the traditional optical supply chain, lasers are grown, chopped into thousands of tiny squares, and sold as individual units that must be actively aligned—turned on and moved around by robots until the light is focused—inside a gold box. This process is slow, expensive, and completely unscalable for the millions of connections needed by AI clusters. Sivers has solved this bottleneck by developing the InP100 Platform, a manufacturing process that allows them to produce Indium Phosphide lasers on 4-inch wafers that are designed specifically for "flip-chip" bonding. These lasers effectively have "guide rails" etched into them, allowing them to be dropped onto a partner’s silicon wafer (like POET’s optical interposer) with sub-micron precision, without ever needing to be turned on during assembly. This capability is the definition of "wafer-scale" manufacturing, moving optics from a craftsman-like assembly process to a mass-production printing process.

This technology is the direct enabler for the "2026" production targets cited by partners. The high-power lasers Sivers is demonstrating—capable of delivering up to 150 mW per channel—are not designed for standard pluggable transceivers, but for External Light Source (ELS) modules. These are the "batteries" that will power the optical fabrics of companies like Celestial AI (now under the Marvell umbrella). Because the actual AI processors generate too much heat for a laser to survive next to them, the lasers must be housed remotely in these ELS modules and piped in via fiber. Sivers is one of the few foundries in the world capable of producing these high-power, high-reliability lasers at the volume required for this new architecture.

Financially, Sivers sits in a unique and arguably temporary position. After halting the spinoff of its Photonics division in early 2025 due to poor market conditions for SPACs, the unit remains a highly undervalued asset trapped within a broader semiconductor holding company. This "for sale" status, combined with their technological lock-in with POET Technologies, makes them a prime acquisition target. They have already secured the supply chain role; they provide the raw photonic wafers that POET’s interposers need to function. As the industry moves toward the commercial launch of the Celestial AI photonic fabric in 2026, Sivers has effectively become the exclusive gas station for the new AI superhighway—making them either a massively profitable standalone vendor or the next logical purchase for a partner looking to secure their supply chain.

The final verdict

Talking about groundbreaking technology I'd like to bring up something truly incredible. Looking forward, into the 2030s this “thing” could be massive. I am talking about Ferroelectric photonic memory. It's distant, but would be incredibly, immensely amazing. Bringing all of the bullish factors together, recent job hirings, celestial/marvell connection, statements, the interview of the CRO, technology advancements, industry maturing, the CRO hinting about POs during the interview(indirectly or directly depending on how you view it). These are astonishingly interesting times and I wonder how it all plays out. There are of course risks and so on, but the current outlook is incredibly bullish, IMO. Now I would like some help. Please help me research ferroelectric photonic memory and what that would mean for the market. Please post this on Agoracom if you´d like. Here is a little explanation about Ferroelectric photonic memory, but it´s not much ( I know it´s AI you don´t have to tell me in the comments, but it´s refined):

What is "Ferroelectric Photonic Memory"? Remember how I said, "You can't store a photon because it always moves"? This technology cheats that rule. instead of trapping the photon, it uses a special material (Ferroelectric Lithium Niobate) that "remembers" a state.

How it works: You hit the material with a voltage pulse. The material's crystal structure physically flips (like a switch) and stays that way without power (Non-Volatile).

The "Memory": When you send light through it later, the light changes behavior based on whether the crystal is flipped "Up" or "Down."

The result: You have effectively created a Photonic Hard Drive. You can store 0s and 1s on an optical chip without needing electricity to keep them there.

2. Why is POET's CEO involved? Suresh Venkatesan isn't just a business CEO; he is a deep-tech scientist (ex-GlobalFoundries CTO).

The Collaboration: The paper was a collaboration between the National University of Singapore (NUS) and POET Technologies.

The Material: The research uses Lithium Niobate, which is the exact same "super material" POET is currently integrating into its Optical Interposer for high-speed modulators.

The POET Advantage: This research suggests POET is not just building "dumb pipes" for light; they are building the IP to eventually put memory and logic directly onto the interposer.

3. Does this solve the "Electricity Problem"? Yes. This bridges the gap I mentioned.

Old Problem: We needed electronics to store data because light couldn't "stop."

New Solution: We use Ferroelectric materials to store the data optically (or rather, in a material state readable by light).

Energy Savings: This memory requires Femto-Joules (almost zero energy) to switch and requires Zero Energy to hold the data (Non-volatile).

4. Is this a Risk or an Opportunity? It is a massive long-term opportunity (Blue Sky Catalyst), not a risk.

If POET owns this IP: It means they hold the keys to the "All-Optical Computer" of the 2030s.

The Timeline: This is still in the "Research Lab" phase. It works in a Nature paper, but it is not ready for mass production in 2026/27.

The Bridge:

2025–2028: POET sells the "Interposer" (Communication).

2029–2035: POET introduces "Photonic Memory" (Storage).

Correction to my previous statement: I told you "Photonic Memory is impossible." Suresh Venkatesan just proved me wrong. With this technology, POET could eventually replace even more of the electronic parts of a computer than we thought. You were right to bring this up—it is the wildest "Bull Case" for the stock.

[macnai] The Total Addressable Market for 800G Optical Interconnects: Market Size, Value ...

macnai — 1/19/2026 9:13:49 AM

The Total Addressable Market for 800G Optical Interconnects: Market Size, Value Distribution, and Implications for POET Technologies | by The Optical Edge | Jan, 2026 | Medium

The Total Addressable Market for 800G Optical Interconnects: Market Size, Value Distribution, and Implications for POET Technologies

The Optical Edge

Follow

4 min read
·
23 hours ago

Introduction

The rapid expansion of artificial intelligence (AI) workloads has fundamentally altered the structure of data centre networking and, in turn, the economics of optical interconnects. Whereas traditional enterprise and cloud applications were dominated by north – south traffic patterns, AI training and inference workloads generate sustained east – west traffic between accelerators, memory, and storage nodes. This shift has increased the importance of bandwidth density, power efficiency, and latency within data centre networks.

In response, 800 gigabit per second (800G) optical interconnects have emerged not as an incremental upgrade, but as a necessary architectural response to changing workload characteristics. This paper examines the total addressable market (TAM) for 800G optical interconnects, quantifies its scale in monetary terms, and analyses how value is distributed within an 800G optical module. Particular attention is given to the positioning of POET Technologies within the optical interconnect value chain and the portion of market value that is realistically addressable by its Optical Interposer architecture.

Market Drivers and the Necessity of 800G

AI workloads scale by increasing the number of interconnected accelerators. As cluster sizes grow into the tens or hundreds of thousands of GPUs, network bandwidth becomes a critical determinant of overall system efficiency. Industry studies indicate that insufficient interconnect bandwidth can result in meaningful reductions in effective GPU utilisation, undermining the economic return on high-cost accelerator infrastructure (Cisco, 2023; NVIDIA, 2024).

At 400G, data centre operators are already encountering constraints related to port density, power consumption, and cabling complexity. The move to 800G enables operators to double bandwidth per port, reducing the number of required optical modules, switch ports, and fibre connections for a given level of throughput. Consequently, adoption of 800G is increasingly driven by total cost of ownership considerations rather than peak performance metrics alone.

Quantifying the 800G Total Addressable Market

Optical interconnects typically represent approximately 30 – 40% of total data centre networking capital expenditure. Annual global spending on data centre optical interconnects reached approximately $10 – 15 billion in the early 2020s and is expected to expand materially as AI-related infrastructure investment accelerates (Dell’Oro Group, 2024). As hyperscale operators continue to build AI-centric data centres, total optical interconnect spending is forecast to exceed $20 – 25 billion per year by the latter half of the decade.

Within this expanding market, 800G optical modules are expected to account for a growing proportion of high-performance ports. Conservative scenarios suggest that 800G optics could represent 25 – 30% of total optical interconnect revenue at maturity, implying an annual TAM of approximately $5 – 7 billion. More aggressive scenarios place 800G penetration at 40 – 60% of high-end ports, yielding an annual TAM of $10 – 15 billion (LightCounting, 2024).

Average Selling Prices and Module Economics

Early-generation 800G optical modules have reportedly sold at average selling prices in the range of $1,500 – 3,000 per unit. This price reflects low initial volumes, technical complexity, and yield challenges. Importantly, this ASP represents the fully assembled optical module, incorporating lasers, modulators, photonic integrated circuits, digital signal processors, packaging, housing, and testing.

Distribution of Value Within an 800G Optical Module

Get The Optical Edge’s stories in your inbox
Join Medium for free to get updates from this writer.

Subscribe

At advanced speeds such as 800G, packaging, assembly, and integration-related activities account for a growing share of total module cost. Industry analyses estimate that 30 – 50% of the total cost of an 800G module may be attributable to packaging, alignment, assembly, and integration processes (Yole Group, 2023). This shift elevates the importance of manufacturability, yield, and thermal management as sources of economic value.

POET Technologies’ Addressable Share of the 800G ASP

POET Technologies does not seek to capture the full $1,500 – 3,000 module ASP. Its Optical Interposer architecture addresses the optical engine and integration layer of the value chain. Under conservative assumptions, POET’s addressable revenue may represent approximately 10 – 20% of module ASP, equivalent to roughly $150 – 600 per 800G module. Under more optimistic assumptions, which one would assume given the advancements in their technology, this share could approach 20 – 30%, or approximately $300 – 900 per module.

Implications for POET’s Serviceable Addressable Market

Applying these assumptions to a total 800G module TAM of $5 – 15 billion annually implies a POET serviceable addressable market in the range of approximately $1.5 – 4.5 billion per year. This reflects POET’s exposure to the integration layer rather than the full module market.

Conclusion

The transition to 800G optical interconnects reflects a structural change in data centre networking driven by AI workloads. While the total 800G optical module market is estimated at $5 – 15 billion annually, POET Technologies’ realistic opportunity lies in the integration and packaging layer of this market. Based on estimated value capture of 10 – 30% per module, POET’s serviceable addressable market can be reasonably framed at approximately $1.5 – 4.5 billion annually. As data rates progress toward 1.6T and beyond, the economic importance of scalable integration architectures is likely to increase further, reinforcing the long-term relevance of POET’s positioning.

References

Cisco Systems (2023). Global Cloud Index and Data Center Networking Trends.

Dell’Oro Group (2024). Data Center Switch and Optical Components Forecast.

LightCounting (2024). High-Speed Ethernet Optics Market Report.

NVIDIA (2024). Scaling AI Infrastructure: Networking Considerations.

Yole Group (2023). Photonic Integration and Advanced Packaging Economics.

[macnai] Gaetano on X: "POET CRO Interview - Deep Dive" / X On Friday, the CRO of POET T...

macnai — 1/19/2026 9:10:45 AM

Gaetano on X: "POET CRO Interview - Deep Dive" / X

On Friday, the CRO of POET Technologies, Raju Kankipati, was interviewed on Picmagazine and he dropped some really important crumbs that are worth dissecting. In this post I will call out to what I believe are the most significant, dissect them, and create a ‘pulse’ check on the company going into 2026.

2026 is a show me year for POET, and they are starting off swinging.

Let’s jump right in.

Here are the 5 quotes that I want do discuss:

“This year is our year to productize, meaning to get into production and start shipping volume of these engines.”
“The poet solution enables wafer scale assembly of EMLs on our silicon platform, so without any active alignments.”
“One of the huge benefit of this point terralight family is the same engines can be used for two different module types... it saves a lot of engineering resources and makes manufacturing less capex intensive.”
“At that speed [3.2T], our engine will just sit like a BGA package on a module... it eliminates wire bonds throughout the solution... 2026 will be good year for us to demonstrate 400 gig per lane.”
“We’ve been working with celestial AI, which recently got acquired by Marvel... we’re also actively working on engaging a few other large customers.”

“This year is our year to productize, meaning to get into production and start shipping volume of these engines.”

If you have been following my content you know how often I speak about the significance of 2026. This is the year for POET to show that all the technological foundation and partnerships that have been laid can actually be 1) productized and 2) commercialized. They are claiming to be really busy across almost a dozen active customers. While I am patiently awaiting the $5m production order to be fulfilled, I am also believing that the company is building a repeatable, high-yield manufacturing hub that can service the entire hyerpscaler market. With such a strong cash position, focus is on execution.

“The poet solution enables wafer scale assembly of EMLs on our silicon platform, so without any active alignments.”

This is the lego concept/analogy that I and others speak of often. It is the core of the investment thesis for POET. In traditional optics, manufacturing is a challenge. You power on every laser using a high-precision robot to wiggle it by microns until the light hits the fiber. Traditionally relatively slow, capital-intensive and could result in low throughput.

POET’s solution is that they use microscopic guide rails printed directly onto the silicon interposer. They drop the EML lasers into these slots and they just place in there accurately. No powering on the laser. I don’t need to go to in depth here as I have covered it in depth before and if you want to understand it deeper it would be best to look at their explanations of it with a video.

Management explicitly confirmed that they are doing this at wafer scale. This means they can process hundreds of engines at once on a single wafer. That is how you hit the margins needed to compete with the giants. If they can maintain the yields on this process during the H2 26 ramp, the cost advantages over traditional SiPh could be massive.

“One of the huge benefit of this point terralight family is the same engines can be used for two different module types... it saves a lot of engineering resources and makes manufacturing less capex intensive.”

Here we are talking about efficiency and stickiness. Usually if a customer wants a 500m reach (DR8) and a 2km (2xFR4) then they have to design two entirely different modules. This is expensive, potentially twice the engineering talent, and it significantly increases capex. POET is basically saying “use our one Teralight engine for both”. Think about the implications of this for a customer like Luxshare or Foxconn. They can design on moterhboard, one manufacturing process, and one testing sweet. Then they could just plug in the POET engine to give their hyperscaler customers the reach they need.

This could save months of R&D. The vision is that once a customer builds their massive roadmap around POET’s common engine, it would become a massive hurdle for them to switch. This makes POET sticky and potentially the standard heart of the transceiver market.

“At that speed [3.2T], our engine will just sit like a BGA package on a module... it eliminates wire bonds throughout the solution... 2026 will be good year for us to demonstrate 400 gig per lane.”

Ok fair enough, I could see some people looking at this statement and glossing over. So much technicality. So let’s break it down.

800G refers to the single fiber connection moving 800 Gigabits of data per second. This is the current widely used state. The industry is moving to 1.6T (1,600 Gigabits) and they are already focusing in on 3.2T (3,200 Gigabits). I asked ChatGPT to put this into perspective that people could conceptualize and it said that 3.2T is roughly enough bandwidth to stream 120,000 4K movies at the exact same time through a cable the size of your thumb. I thought that was a fun and mind-blowing fact. Please don’t shoot the messenger if it’s not exactly 120,000, just understand that the amount of data being moved, and the speed at which it is moving, is immense.

To hit these speeds the goal is to increase the lane speed. Think of a highway. You can either add more lanes (which makes the chip larger and more expensive) or you can make the cars go faster. The industry wants to make the data travel faster. So moving from 100G to 200G to 400G. There are some physics problems that get introduced here. In a traditional chip there is communication through tiny gold wires called wire bonds. At lower speeds, these wires are fine. But at 400G per lane, electricity is vibrating so fast that those wires leak energy, pick up noise from other parts of the chip, and create massive heat. This inductance essentially crashes the signal.

POET’s goal is to kill the wire bond entirely. They are moving to a semicon package style that looks more like a high-end processor than a traditional optical part. Through TSV (through-silicon vias), signals travel vertically through the chip rather than around the edge, which is a significantly shorter path and results in way less signal loss. And with BGA packaging (ball grid array), the entire optical engine sits flat on the circuit board. It connects using tiny solder bumps underneath which is the same way a NVIDIA GPU or an Apple M3 chip connects to a motherboard. Management will be demonstrating this 400G-per-lane performance in 2026. This would be a massive upgrade in physics. If wire bonds become a bottleneck, POET could prove to be the socket that is the solution.

“We’ve been working with celestial AI, which recently got acquired by Marvel... we’re also actively working on engaging a few other large customers.”

Ah, is this the validation we have been waiting for? Ever since the Marvell deal, we have been operating under the assumption that the POET and Celestial AI partnership is still ongoing. While we obviously don’t quite know what Marvell will do with all the components of the supply chain long term, I do believe this is a massive signal.

I believe it is safe to assume that Marvell didn’t spend $3.25 billion on Celestial AI without a deep audit. And this audit would show that the POET Starlight is a light source for that architecture. IMO ou don’t buy the photonic fabric if you aren’t comfortable with the the engine. In fact, Celestial’s CEO David Lazovsky (who was POET’s former Chairman) has been very vocal about the “co-designed” nature of this stack.

The ultimate validation will be any specifics tied to this, whether its official press releases, design-ins, or the big one: volume production orders. If Marvell is aiming for $500M+ in revenue from this fabric by 2028, they need a light source that is already qualified and manufacturable today. POET is currently the only one currently sitting in that seat. Technically speaking, Marvell COULD shake up the supply chain, but I don’t see why they would want to. But that is why it will be a big deal to see any further confirmation.

Also, if we get more updates or orders from these other ‘large customers’ that could be significant upside for the stock. We’re already seeing hints of this with the NTT 100G bidi chip and the Luxshare 800G ramp. I would like to see some updates on these engagements throughout 2026 to prove that POET is becoming the quote “Switzerland of Photonics” for more than just one giant player.

Closing thoughts This interview was just one of many exciting bits that have come out of POET the last 6 months. Some new bits and some confirmation. The tech is very promising. I am bought in on that.

But we really need to see the first 5m order fulfilled and get all the key information that let’s us know if its economous, scalable, high yield etc.

Then things could really get shaking.

As always, while I get really excited about the technical aspects of these companies, I never want to understate the risks. We are talking about a company that is valued at roughly 1 billion dollars that has little meaningful revenue. The market is pricing in significant growth. If there are concerns about that growth or it proves to not be commercially viable, there will be significant compression in the share price. As an investor, I obviously have faith in their ability to be successful, but I also understand the risk that I take.

Please always do you due diligence and don’t blindly follow my investments as this is not a recommendation to purchase the stock.

[Claude Cormier] AI Celestial & Optical Interposer are mentioned in this video. But it is not abo...

Claude Cormier — 1/4/2026 5:42:12 PM

AI Celestial & Optical Interposer are mentioned in this video. But it is not about POET. Fascinating but I am sure I dont' understand it all. Comments welcome.

[longz] Check your inbox. mail.......

longz — 12/24/2025 12:44:07 AM

[Claude Cormier] Those laser links...would have to be photons based? Is there any other alternat...

Claude Cormier — 12/22/2025 11:25:44 AM

Those laser links...would have to be photons based? Is there any other alternatives?

[w0z] SeekingAlpha: POET Technologies Is Lighting Up The AI Data Center Bottleneck ...

w0z — 12/22/2025 10:30:34 AM

SeekingAlpha: POET Technologies Is Lighting Up The AI Data Center Bottleneck

Summary

Poet Technologies sits at the center of the AI-driven data center transformation with its patented optical interposer platform.
POET's technology enables faster, cheaper, and more energy-efficient data transfer, directly addressing hyperscaler needs as AI workloads surge.
With $150M in cash, no debt, and major manufacturing partners, POET is well-capitalized to scale operations and pursue growth opportunities.
Strategic partnerships with Foxconn, Luxshare, and Mitsubishi Electric validate POET's technology and open direct channels to hyperscale data centers.

Eoneren/E+ via Getty Images

Investment Thesis
First and foremost, Poet Technologies ( POET) sits in the sweet spot of once in a generation tech boom. The company's patented optical interposer platform addresses a critical bottleneck in AI data centers. And this is because it's all about moving massive amounts of data at high speed and low power using light instead of traditional copper wires. I have previously rated POET a Strong Buy and the stock rewarded that conviction by soaring roughly 63% after my initial call and another 2.4% after my follow up. And this is before taking a recent breather. Even as the Quant Factor grades swung from a strong Buy to a Strong Sell during the pullback, I still firmly believe POET's story is just getting started. And that's why I am maintaining my Strong Buy rating. I will be discussing the reasons for my reiteration below.

Seeking Alpha

POET has a unique technology solving a big problem. POET's optical interposer enables the seamless integration of electronic and photonic devices on once chip. And in simple terms, this means that it lets data centers replace short distance copper connections with tiny optical links that are cheaper, faster and more power efficient. That is exactly what hyperscalers like Meta ( META) and Microsoft ( MSFT) need as AI workloads explode. By essentially turning crowded electrical lanes into light paths, POET's chips dramatically boost bandwidth while slashing energy use and cost. This is a revolutionary solution to the heat and speed constraints plaguing modern AI supercomputers.
Second there is a huge demand for AI compute capacity that is outstripping current data center infrastructure. Both MSFT and Amazon ( AMZN) have openly said that they can't keep up with the AI demand. Yet big tech plans to spend hundreds of billions on AI infrastructure. Optical transceiver sales are already projected to grow 56.5% annually reaching 31.9Mn high speed units in 2025. Every one of those units needs optical engines and light sources. And this happens to be POET's specialty. This means that the AI revolution is forcing a shift to optical connectivity inside data centers and POET provides the driving forces for that shift.
Thirdly, the company is well funded for Growth. 2025 marked POET's pivot from R&D mode to commercial mode. The company reported its first modest revenues as customers began paying for initial products. And more importantly, POET has secured major manufacturing partners and raised a war chest of cash. As of Q4, POET had roughly $150Mn in cash and zero debt. This includes a recent $75Mn private placement that was completed in October. It is important to note that the $75Mn cheque was the largest in POET's history and came from a single institutional investor. This shows a strong vote of confidence. This capital gives POET the runway to scale R&D, expand production and even pursue strategic acquisitions without financial strain. This means that the balance sheet is no longer a concern. And management can focus 100% on execution.
Lastly, POET isn't toiling in a vacuum. It has tier 1 tech manufacturers on board. For example Foxconn's ( FXCOF) network arm and Luxshare Tech which are both heavyweights module integrators are co developing 800G and 1.6T optical transceiver modules using POET's engine. POET has also partnered with Mitsubishi Electric ( MIELY) to integrate lasers into its designs. And just recently, it added Lessengers as a new customer for an 800G project. Clearly, these relationships validate POET's technology and give it direct channels into the largest hyperscale data centers via well connected suppliers. And I believe that when industry leaders line up to work with a tiny company, then it is a sign that something big is brewing.
Now that I have elaborated my investment thesis, it is prudent to have a look at the catalysts that will propel POET's value higher.
The Catalysts
The biggest catalyst for POET is the unprecedented buildout of AI data centers happening globally. We are literally in an AI arms race. And one bottleneck that's emerged is connecting all those advanced chips and servers. It is not just enough to have the fastest GPU if you can't feed it data fast enough. And this has created a paradigm shift from electrical to optical connectivity inside data centers. Every major cloud player is literally scrambling to upgrade connectivity. MSFT's CEO admitted that their data centers are constrained by networking limits. And AMZN is investing over $100Bn to alleviate bottlenecks. And the result is that the transceiver unit demand is on a steep upward trajectory. And suppliers are struggling to keep up. Yet, this is exactly the environment POET has been waiting for.
The second catalyst is the Marvell ( MRVL) endorsement. A recent game changing development for POET is the Marvell's blockbuster $3Bn acquisition of Celestial AI. Celestrial AI is an optical computing startup whose revolutionary photonic Fabric tech is built on POET's optical interposer platform. In fact, POET worked closely with Celestial to co-develop the customized interposer that powers the photonic Fabric modules. This means that Marvell Technology, a $50Bn semiconductor company has essentially validated POET's technology as a crucial enabler for next gen AI infrastructure. It also de-risks POET's future prospects. And this is because Celestial is no longer a venture stage startup but a part of a well capitalized industry leader. MRVL expects celestial to reach an annualized $1Bn revenue run rate by 2029. And if POET's content makes up just 10% to 15% of those systems, this implies about $100 to $150 Mn annual revenue for POET from this one customer. For context, $150Mn would dwarf all other current projections for POET's 2029 sales. This potential windfall is not immediate but it provides a clear line of sight to large scale revenues as Celestial's photonic Fabric gains traction.
It is important to mention that Marvell's move has put competitors on notice. Other module manufacturers and chipmakers will want to ensure they are not left behind on optical interconnects. This could pave the way for POET to secure significant orders from tier 1 players who now must compete with MRVL on cost and power efficiency. In my view, MRVL unintentionally did POET a favor. It sounded a wake up call across the industry that integrated photonics is key to AI performance. Additionally, it wouldn't surprise me if POET itself becomes an acquisition target in the long run. Companies like Coherent Corp. ( COHR), a leader in lasers and optics, could eye POET's platform to leapfrog into integrated photonics. POET's tech could help COHR for instance produce 1.6T optical transceivers at much lower cost which would be crucial to compete with MRVL new photonics advantage.
Valuation
By conventional metrics, POET stock looks extremely expensive. But I believe that's the price of admission for a potential payoff. Also, I am not buying POET for what it is today but for what it could become in 2 to 5 years. I'd say that the market is essentially valuing POET like an IP driven, high growth platform rather than a hardware manufacturer with current earnings. For context, let's take a look at established optical component peers that do have revenues. Coherent ( COHR) trades at around 4.1x FWD P/S. Ciena ( CIEN) is at about 4.9X FWD P/S and Lumentum Holdings ( LITE) is at 9x as well. These are single-digit sales multiples which underscore how frothy POET's valuation is on a relative basis. It means that POET is 191x more expensive by FWD P/S right now as shown below.

Seeking Alpha

But again, those peers are growing in single digits while POET's growth is off the charts. If POET achieves even a fraction of its potential, the current share price could end up looking cheap in hindsight.

Seeking Alpha

So how do we gauge that potential? In my previous analysis I estimated POET's TAM for optical engines could reach $17Bn by 2028. And if POET captures a conservative 3% market share, that would be about $500Mn in annual revenue. Assuming a typical semiconductor valuation on that, say 3x EV/Sales, and you'd get an EV of $1.5Bn. After adding net cash and dividing by the share count, then that scenario works out to a stock price in the teens. In fact that methodology would yield between $13 and $25 per share which would be about $9 to $16 in today's value after discounting.
POET's Key Risk
As of now, the biggest risk I see for POET is execution risk. POET must successfully transition from development projects to large scale production and sales. That means scaling up manufacturing without hiccups, meeting stringent customer specs and hitting delivery timelines. Any missteps or delays at the new Malaysia production facility could frustrate early customers and open doors for competing solutions. This is a classic challenge for small companies moving to volume. Because you get one chance to impress a big customer. POET needs to deliver on its initial orders in 2026 so that follow on orders materialize. If there are quality or yield issues, the expected revenue ramp could be pushed out hurting investor confidence.
Final Thoughts

POET is not a blue chip stock and not yet a profitable company. It is a speculative play riding a megatrend. But in my opinion, the pieces are in place for POET to succeed. A unique and validated tech, a massive wave of end market demand, blue-chip partners, a strong cash position and now early revenue traction. The stock is trading at the fraction of what I believe is its 2028 potential and the market still doesn't fully appreciate POET's role in the AI data center ecosystem. Nevertheless, I remain confident that POET has all the ingredients of increasing in value. I believe the road may be quite volatile but the destination could be well worth it. I reiterate my Strong Buy rating.

This article was written by

Kennedy Njagi

[w0z] POET Technologies Turns A Corner Introduction Late 2025 has been full of pivot...

w0z — 12/22/2025 6:31:22 AM

POET Technologies Turns A Corner

Introduction

Late 2025 has been full of pivotal updates for POET technologies. From critical financing to new orders to key partnerships. These last few months are demonstrating that POET is transitioning from an idea to a reality. This report will cover all of the recent relevant updates for POET that I believe have significantly reduced their risk profile. In addition I will also lay down some foundational knowledge on POET for first time readers. Lastly, I will include some one-off bits from user requests that were made on X. If you are invested in POET or you are at least intrigued in the company, you need to read this!

What IS POET Tech?

POET Tech is a semiconductor company that specializes in photonics integration. They build what is called an ‘optical engine’ that helps move data using light (photonics) instead of electricity (copper) inside AI and data-center hardware. The core idea is to put multiple photonics parts like lasers, modulators, and detectors all together in a very compact package called the Optical Interposer (OI). Customers can plug these optical engines into transceivers or other modules to get higher bandwidth, lower power, and smaller size than older designs that use more discrete parts and more complex assembly. The theory is that they can solve the cost and scalability bottlenecks that haunt data centers.

Overview of this quarter

The events over the last few months have moved this thesis from theoretical to empirical. The receipt of an over $5m production order for 800G Infinity engines signifies the official exit from the R&D phase. This signal was fortified with a $150m financing round, which brought pro-forma cash to in excess of ~$300m immediately following closing and materially reduced near-term financing risk. However, what I believe to be the most profound development was Marvell’s $3.25B acquisition of Celestial AI. As a disclosed light-source partner to Celestial (via POET Starlight), POET now has real exposure to a roadmap that is moving inside one of the largest data-infrastructure semiconductor platforms in the world. The presence of David Lazovsky (former POET chairman and current C-AI CEO) at the center of this transaction serves as a powerful validation of the tech lineage and strategic fit of the OI and the industries shift towards optics.

On top of all of this we see POET continuing to keep up pace (or outpace in many instances) the industry standard. The unveiling of the 1.6T transmitter PIC on 12/8 reasserts the company’s lead over many traditional silicon photonics. Further, the strategic collaborations with Sivers Semiconductors and Quantum Computing Inc. expands the TAM into ELS and TFLN modulators for 3.2T applications. We will unpack this more later.

All in all, I believe that not only did POET become way less risky these last few months, but their upside also grew significantly. I personally had concerns about their long term future due to increased competition and tech disruption. This has changed in a very meaningful way due to all the updates I will cover in this report.

I want to drop a quick disclaimer. None of this is financial advice and should not be seen as a recommendation. While I will discuss how all of these updates helped REDUCE risk, there is still plenty of risk to be had with a company that still does not have meaningful revenue and has alot to prove. While I am clearly bullish and am positioned accordingly, I have no crystal ball into what happens next with the price. I think we will see plenty of volatility over the next few months with wild price swings. Use this report as a way to expand your perception of POET rather than a signal to buy.

Foundation

Before diving into all of the recent news I think it’s important to set the stage for why POET exists and the need that it fits in the market. When I think about the need for POET it almost always goes back to the transition away from copper to photonics and the need for something better than traditional active alignment.

For decades data centers have relied on copper interconnects to move data between racks and servers. As GPU speeds dramatically increase, the physical limitations of copper are being seen across many distances.

First off is the signal and power loss. At speeds required for training AI (800G, 1.6T and beyond), electrical signals rapidly degrade. To compensate for this, we must pump more power into the signal creating massive heat waste.

The industry has introduced Active Electrical Cables (AEC) which essentially put boosted chips inside the cable to extend copper’s reach to roughly 3-4 meters. For distances 0-2 meters (inside the rack) passive copper is cheap, reliable and zero power. This distance will remain copper for the foreseeable future. 2-5 meters (rack-to-rack) is where AEC puts up the fight. Copper needs the boost to survive here. While it is heavy and power-hungry it is currently cheaper than optics. 5+ meters (row-to -row and cluster-wide) is where photonics shines. This is POET’s focal point.

Second, is the thermal constraint. Modern AI clusters are power limited and every watt spent on moving data is a watt that won’t be spent on computing data. Photons have far lower loss over these distances and generate significantly less heat than pushing the same bandwidth over copper.

Lastly we see the need for a new style of manufacturing. The traditional method of active alignment is slow, capital intensive and results in low throughput. Basically a laser is powered on, a high-precision robots moves a lens of fiber in microscopic increments to find the point of max light transmission, and then the component is glued in place. Imagine trying to plug a USB cable into a port in a pitch-black room. You have to fumble around, wiggle the connector, and maybe turn on a flashlight to see if it’s lined up before you push it in.

Compare this to POET’s passive alignment strategy. Imaging you are a kid again and you are playing with legos. The bumps and holes on the bricks are so precise that you don’t need to look closely or wiggle them. You just snag them together and they fit perfectly every time. This is what POET has done. When POET makes their silicon base (the wafer) they use microscopic printing to create tiny guide rails and slots, if you will. These are the legos. Because these guide rails are perfect, high-speed robots simply pick up a laser component and drop it into the slot. No need for a laser to be on. No need for wiggling. This might sound small, but this allows chips to be mass produced in a method akin to standard computer parts resulting in a lower bill of materials and lower capex.

Ok, now on to the news.

Financial Inflection

Over the last few months we have witnessed a dramatic restructuring of POET’s financial foundation. The company has moved from a position of financial scarcity to one of abundance that allows them to be aggressive on their expansion strategy.

Let’s first talk about the $5m production order from October. This was POET’s first significant production order and it was for 800G Infinity optical engines. While the dollar amount seems financially insignificant for a company of POET’s size, its structural significance is immense. This order represents real customers and real revenue. POET has been around for a long time. Taking a product like the OI from idea to deployable takes many many years. There are rigorous qualification gates associated with this process. Another interesting piece of this order is the ‘Infinity’ architecture. This design allows customers to ‘daisy-chain’ 400G and 800G engines to achieve 1.6T and 3.2T speed. This provides a modular upgrade path. These orders are scheduled for shipment in the second half of 2026. Once this happens we will get crucial information on yields, reliability, margins etc. and see whether or not POET is as important and cost-effective as I believe it to be.

Next let’s talk about the Balance Sheet. The financial transformation began on October 7, when POET closed a $75 million private placement with a single institutional investor. A critical vote of confidence that set the stage. This was immediately followed by a $150 million financing round. The offering was characterized by ‘all institutional money’, implying that sophisticated investors conducted deep DD on the tech and order book. The absence of warrants from this raise indicated high demand. Post-raise POET had roughly $300m in cash. This gives them funding to aggressively pursue partnerships and fulfill orders.

The Marvell Acquisition of Celestial AI

The most significant industry event of the last few months for POET was the acquisition of Celestial AI for $3.25B. I view that as a meaningful de-risking signal for Celestial’s key enabling components, including the light-source side where POET participates. While the vast majority of details of the deal are mostly irrelevant for POET investors, I will discuss the relevant bits and where we should pay attention. $3.25b upfront ($1B cash, $2.25B stock) and a potential earnout of up to $2.25B. Total value up to $5.5B. The earnout is tied to revenue milestones, with one-third earned if Celestial’s cumulative revenue by the end of Marvell’s fiscal 2029 exceeds $500m, and the full earnout earned if cumulative revenue exceeds $2.0b by that same deadline. This structure implies that Marvell expects C-AI to generate billions in revenue over the next 3-4 years. For suppliers to C-AI (like POET) this signals a massive ramp in component volume.

The critical insight for us as POET investors is the supply chain relationship. Celestial needs external light sources to drive its platform, and POET has disclosed a development/production relationship tied to POET Starlight. POET disclosed an agreement with Celestial in February 2022, and then in April 2023 it introduced POET Starlight and disclosed an agreement for development/production plus an advanced purchase order for initial production units. Celestial needed a light source that was passively aligned, scalable, and capable of providing multiple channels of light in a small form factor. Poet’s platform was the unique enabler for this, allowing 4-channel and 8-channel configuration with flip-chipped lasers.

So by acquiring C-AI, Marvell is implicitly underwriting the stack that makes Celestial work. I view that as a meaningful de-risking signal for Celestial’s key enabling components, including the light-source side where POET participates.

This connection is reinforced by deep leadership ties. David Lazovsky, the Founder and CEO of Celestial AI, served as the Executive Chairman of POET Tech prior to founding Celestial. Lazovsky possess intimate knowledge of POET’s OI capabilities. His decision to select POET as a supplier for Celestial was based on a deep understanding that POET offers the most viable path to cost-effective, wafer-scale light sources. Lazovsky has publicly stated that Marvell is the ‘‘ideal home’ for Celestial. For me it is probable that he has advocated for the robustness of the POET solution within Marvell’s ecosystem.

To finish this section I believe that this relationship with Lazovsky significantly reduces the risk of Celestial ‘‘designing out’ POET. Instead, it seems to me like it would increase the likelihood of deeper integration or even a potential future acquisition of POET by Marvell to vertically integrate the light source tech. Purely speculative, but my brain could connect those dots.

The 1.6T Inflection

On December 8, 2025, POET announced its hybrid-integrated 1.6T transmitter PIC. This is POET showing it can build the kind of high-speed optical core that next-gen AI clusters will need.

They took four laser arrays and attached them directly onto the Optical Interposer, then placed the driver chips right next to them. That matters because it reduces the amount of delicate, manual-style assembly that usually makes these products expensive and hard to scale. They also pulled the wavelength-combining step into the same package using integrated multiplexing, instead of relying on extra standalone parts.

The takeaway is scalability. Laser arrays at these speeds are usually a manufacturing nightmare because everything has to line up perfectly across multiple lanes. POET’s whole platform is built to make that placement repeatable, fast, and high-yield. This is why I view the 1.6T update as a real proof point, and why it keeps the 3.2T roadmap credible as bandwidth steps up.

Ecosystem and Partnerships

POET has transitioned from a vertically integrated R&D shop to the hub of a diverse ecosystem. By partnering with best-in-class material and component suppliers, POET creates complete solutions that are greater than the sum of their own parts. Let’s look at Sivers Semiconductors, Quantum Computing Inc, Foxxconn and Luxshare

On September 29, 2025, POET solidified a partnership with Sivers Semiconductors to co-develop External Light Sources (ELS) for Co-Packaged Optics (CPO). Without getting into all the highly technical details, Sivers provides the ‘bulb’ (laser chip) and POET provides the ‘socket’ (the interposer packaging). Together they create a compact, pluggable light engine that aims to solve the thermal management issues of CPO. The ELS market is projected to be a $1B annual opportunity within the broader AI connectivity market. Through this partnership they expect to demonstrate prototypes in the first half of 2026, with production readiness by the end of 2026. This partnership demonstrates one major way that POET intends to capture market share in the CPO world.

Next up we have the announcement with Quantum Computing Inc. on November 11th. This partnership aims at developing 3.2T optical engines using a material called Thin-Film Lithium Niobate (TFLN). TFLN allows us to switch light on and off at incredible speeds with very little power, which is exactly what you need as data rates hit 200G and 400G per lane. The problem is that TFLN is notoriously difficult to handle; it is brittle and hard to package. This partnership aims to solve that. QCi brings the TFLN modulator capability, and POET’s interposer acts as the ‘socket’ to package it safely and integrate it into an engine that can scale. By bonding the TFLN chip to the POET interposer, they create a hybrid engine that combines the raw speed of Lithium Niobate with the manufacturability of silicon. This puts POET at the bleeding edge of physics. POET is laying the groundwork for the 3.2T era (post-2027). This partnership was significant for me because it’s a sign that no matter where the market goes - from pluggables to CPO - POET is going to adapt and stay relevant in a meaningful way beyond just providing lasers.

The partnership with Foxconn (FIT) is about co-developing 800G and 1.6T pluggable optical transceiver modules that use POET’s optical engines. At the same time, POET has been assembling Infinity engines at its high-volume production facility in Malaysia, which strengthens the narrative around scale and a resilient supply chain.

Then we have Mitsubishi Electric, a heavyweight in laser technology. This collaboration is massive because they are co-developing 3.2T optical engines for the AI market. Mitsubishi brings their top-tier 400G EMLs (Electro-absorption Modulated Lasers) to the table. These are components that are notoriously difficult to package due to heat and signal integrity issues. POET integrates these lasers using the Optical Interposer to create a unified 3.2T chipset. POET and Mitsubishi originally said they aimed to complete the 1.6T and 3.2T optical engine chipsets in early 2025 and jointly support major-customer demonstrations, and since then POET has publicly showcased 1.6T engines partnered with Mitsubishi’s 400G EML lasers as a tangible step on that roadmap.

Finally, there’s Luxshare. As a major global manufacturer of optical modules, Luxshare’s involvement is a strong commercial signal. This partnership has expanded to include both POET’s transmit and receive optical engines for Luxshare’s 400G and 800G pluggable transceivers. Luxshare is using these engines to build 800G transceivers for data centers. This partnership is all about volume and design-in status where Luxshare takes POET’s engine and packages it into a finished pluggable module that they then sell directly to the world’s largest cloud providers.

Conclusion

For years, the bear case on POET was that it was a lab experiment that would never achieve manufacturing scale. I believe that all the updates spoken through on this report help to challenge that bear case.

Furthermore, the strategic landscape has shifted in POET’s favor. The acquisition of Celestial AI by Marvell is a $5.5 billion endorsement of the exact architecture POET enables. As a disclosed light-source partner to Celestial, POET is no longer just a science project narrative; it is tied into a high-value architecture that is now being scaled inside Marvell.

Technologically, the Optical Interposer remains the only viable path to passive alignment at wafer scale. As the industry moves to 1.6T and 3.2T, the mechanical precision required renders traditional active alignment economically unviable.

We still need to see if POET’s tech is commercially viable. We will know way more about this at the end of 2026. But the more the news stacks up, the more partnerships announced, and the bigger the TAM, the more upside we see if POET is successful.

[longz] Amazing man! Warren Buffett is a legendary American investor, philanthropist, an...

longz — 12/11/2025 12:23:35 AM

Amazing man!
Warren Buffett is a legendary American investor, philanthropist, and business magnate, known as the "Oracle of Omaha," who built his fortune as the Chairman and CEO of Berkshire Hathaway, a conglomerate that owns companies like Geico, Dairy Queen, and Duracell, and holds significant stakes in others like Coca-Cola. He's famous for his value investing philosophy, long-term approach, and unparalleled track record of beating the S&P 500, making him one of the world's wealthiest people and a major figure in finance.

[longz] check ur e mail... [graphic]

longz — 12/9/2025 9:26:15 PM

check ur e mail...

[longz] add a few 000's

longz — 12/9/2025 9:12:10 PM

[da_cheif™] i got 4000 shares

da_cheif™ — 12/9/2025 9:11:12 PM

[longz] Top Canadian stocks to buy immediately with just $1,000

longz — 12/9/2025 8:12:57 PM

[VApeanut] POET post by Ahmed Abdelazim Click on above for Ahmed's POET post. VApeanut .

VApeanut — 12/9/2025 2:45:27 PM

[toccodolce] [X] $POET is basically saying they built a smarter, cleaner way to send huge am...

toccodolce — 12/9/2025 11:56:14 AM

[X]

$POET is basically saying they built a smarter, cleaner way to send huge amounts of data using light. Traditional optical modules are messy. They need lots of tiny parts lined up perfectly, which makes them expensive, fragile, and hard to scale. POET’s new 1.6-terabit transmitter… https://t.co/kHUKQCBJZG
— mon (@moninvestor) December 9, 2025

[/X]

[X]

pic.twitter.com/DL6pe8b0DY
— Henry (@HenryKengie) December 9, 2025

[/X]

[toccodolce] Latest from RDDT: Recent advances in optical technologies for data centers: a r...

toccodolce — 12/9/2025 11:33:14 AM

Latest from RDDT: Recent advances in optical technologies for data centers: a review

Recent advances in optical technologies for data centers: a review

Qixiang Cheng, Meisam Bahadori, Madeleine Glick, Sébastien Rumley, and Keren Bergman

Abstract

Modern data centers increasingly rely on interconnects for delivering critical communications connectivity among numerous servers, memory, and computation resources. Data center interconnects turned to optical communications almost a decade ago, and the recent acceleration in data center requirements is expected to further drive photonic interconnect technologies deeper into the systems architecture. This review paper analyzes optical technologies that will enable next-generation data center optical interconnects. Recent progress addressing the challenges of terabit/s links and networks at the laser, modulator, photodiode, and switch levels is reported and summarized.

This is a research paper -- TLDR.

[toccodolce] Latest from RDDT POET's Hybrid-Integrated 1.6T 2xFR4 Transmitter VS Competitors...

toccodolce — 12/9/2025 11:21:57 AM

Latest from RDDT

POET's Hybrid-Integrated 1.6T 2xFR4 Transmitter VS Competitors

1. How big of a step up is this?

Short answer:

It is a meaningful architecture-level advancement, not a small incremental upgrade. The key step-up is manufacturability at 1.6T, not raw optical performance.

Why this matters:

In 1.6T optical engines, the bottleneck is no longer the laser speed or modulator speed, the bottleneck is packaging complexity, EML shortages, yield, thermal density, and alignment cost.

POET’s transmitter directly attacks these bottlenecks:

What’s new / important here:

Hybrid integration of 4× 200G EML arrays on a single interposer
- No one else has shown multi-lane EML arrays bonded onto a passive platform with built-in MUX in a wafer-level flow.
- Competitors rely on manual active-alignment + micro-optics.
No wire-bonds at 100+ GHz signals
- Industry wires up 1.6T engines using long RF paths ? more crosstalk, lower yield, higher power.
- POET’s flip-chip RF layout is materially better for 100–150 GHz era optics.
EML capacity improvement
- EML shortages are real.
- POET aggregates four EML arrays effectively, reducing required individual chip count and making better use of scarce InP.
Wafer-scale, passive assembly of a full 1.6T PIC
- Today the competitors assemble engines piece by piece (slow, expensive).
- POET is pushing “optical semiconductorization” — the first real demonstration of a mass-production-ready 1.6T engine.

Bottom line:
This is not “faster optics.”
This is faster manufacturing of high-speed optics, which is the actual industry bottleneck.

2. Where is the rest of the industry right now?

Here is the competitive situation for 1.6T FR4 / 8×200G transmitters:Where POET sits:

The industry is nowhere close to wafer-scale 1.6T hybrid EML engines with built-in AWG MUX.POET is ahead in one very specific area:
High-volume manufacturable 1.6T optical engines.

Not necessarily ahead in raw photonics design, but ahead in packaging, integration, yield, and scalable assembly, which actually determines who ships volume.

3. Is POET ahead of competitors?

Yes, in packaging architecture for 1.6T FR4.

No one has published a wafer-scale hybrid-integrated 1.6T with:

EML arrays
integrated AWG MUX
flip-chip drivers
passive alignment only
no lenses
no wirebonds
<0.5 dB coupling loss

This is unique in the market right now.

What competitors do better:

Coherent and Lumentum have stronger raw photonic device performance (best-in-class EMLs).
Broadcom and Nvidia dominate DSP and link-level architectures.

What POET does better:

Manufacturing architecture
Cost, yield, scalability
Compact RF layout
Ability to mass-produce engines cheaply

For the next 5 years, manufacturability is more important than bleeding-edge photonics speeds.
So this advantage is commercially meaningful.

4. Is the market ready for this product NOW?

Yes.

The timing aligns with hyperscaler roadmaps:

800G ramps 2024–2025
1.6T starts sampling in 2025
1.6T volume begins 2026–2027
Nvidia’s GB200 requires enormous bandwidth; next-gen architectures require 1.6T pluggables and 1.6T LPO.

Hyperscalers (AWS, Meta, Google, Nvidia) are already:

tendering 1.6T transceivers
evaluating engines
selecting manufacturing ecosystems
preparing for 1.6T testbeds

The biggest bottleneck for hyperscalers today:

They cannot manufacture enough 1.6T modules cheaply.

POET’s transmitter directly solves:

* yield * alignment cost * EML supply constraints * thermals * form factorYes, the market wants this.

5. Realistic assessment of competitive advantage

POET is NOT:

The only way to build a 1.6T transmitter.
Guaranteed to win sockets.
Technologically superior in every photonics dimension.

POET IS:

The only company showing mass-producible hybrid 1.6T architecture.
Offering the lowest-complexity, highest-yield 1.6T manufacturing path.
Directly aligned with hyperscaler cost and volume requirements.

That matters more than tiny optical-performance differences.

6. Summary (hard, direct)

How big is the step-up?
- Large. It shifts 1.6T from lab prototypes to manufacturable engines.
How much better than competitors?
- Better in packaging, cost, yield, EML scaling, and simplicity. Competitors still lead in raw device performance.
Is POET ahead?
- Yes, in the part of the stack that matters for volume and cost.
Is the market ready for it?
- Yes. Hyperscalers are preparing 1.6T ramps now and need manufacturable solutions.
Commercial significance:
- If POET wins even a single major 1.6T design-in, it becomes a multi-billion-dollar business, because hyperscalers will buy millions of 1.6T engines per year.

[longz] I wonder whats under that circle blanked out area?...................... :))...

longz — 12/9/2025 11:17:56 AM

I wonder whats under that circle blanked out area?......................

:)))

[toccodolce] [X] NEW: POET Redefines Optical Integration with Its Hybrid-Integrated 1.6T 2xF...

toccodolce — 12/9/2025 11:02:22 AM

[X]

NEW: POET Redefines Optical Integration with Its Hybrid-Integrated 1.6T 2xFR4 Transmitter PIC

"The size, ease of adoption, functionality, value, and performance makes POET’s Hybrid-Integrated 1.6T 2xFR4 transmitter PIC a breakthrough solution for customers. The industry can…
— POET Technologies (@POETtech) December 8, 2025

[/X]

POET Technologies Redefines Optical Integration with Its Hybrid-Integrated 1.6T 2xFR4 Transmitter PIC

Whitepaper by Senior VP, Product Development Dr. Mo Jinyu and Chief Revenue Officer Raju Kankipati (based on documentation presented at ECOC 2025)

POET Technologies has introduced its Hybrid-Integrated 1.6T 2xFR4 Transmitter PIC that represents a leap forward in optical engine integration. The product represents three primary achievements that are notable for the industry:

It validates the POET Optical Interposer™ as a practical manufacturing platform for the world’s fastest and most efficient optical transceivers.
It unites photonic and electronic components within a single chip-scale package — delivering exceptional bandwidth, simplicity of manufacturing, and unmatched scalability for next-generation pluggable and co-packaged optics (CPO).
It efficiently utilizes an EML array (an industry first), which results in a dramatic capacity increase of Indium Phospide-based EML lasers at a time when the industry is facing shortages. With monolithically integrated optical multiplexers (MUX), the 1.6T transmitter PIC provides superior performance, scalability, and manufacturing efficiency that customers demand for next-generation artificial intelligence and cloud infrastructure.

Why the Industry Needs a Hybrid-Integrated Solution

The accelerating growth of AI, cloud computing, and hyperscale data centers is driving an insatiable demand for higher-speed, lower-power optical interconnects. Finding applications that meet this need is of urgent importance to the industry because end users desire more and faster connectivity to fulfill their corporate visions.

Change is required. Not only in device performance but also in integration efficiency. New processes, novel use of materials, and a more elegant approach to design and architecture are called for in this era of rapid evolution.

As network operators transition from 400G to 800G and 1.6T architectures, traditional assembly techniques are reaching physical and economic limits. Wire-bonded packages and micro-optical alignments add cost, magnify RF crosstalk and thermal challenges, and restrict yield.

Profit margins become thinner when using current approaches. A better way of moving data from one point to another has been sought for years. The POET Optical Interposer and the products built from it — including the Hybrid-Integrated 1.6T 2xFR4 Transmitter PIC — are perhaps the best solution for an industry in need. POET eliminates active alignment and wire bonds, providing a high-throughput, CMOS-compatible path to scale.

Product Overview: Hybrid-Integrated 1.6T 2xFR4 Transmitter PIC

Compact, integrated, and ready for volume production, the 1.6T 2xFR4 transmitter PIC integrates the below components:

Four 2x200G EML arrays flip-chip bonded on POET’s optical interposer.
Eight high-speed driver ICs, symmetrically positioned for optimal signal integrity.
Two Arrayed Waveguide Gratings (AWGs) used as Multiplexers (MUX) for efficient wavelength multiplexing.
Spot-Size Converters (SSCs) mode-matched to SMF with <0.5 dB coupling loss.

Key Specifications:
Total capacity: 1.6Tbps (8x106 Gbaud) PAM4 lanes
Chip size (WxL): 7.975mm x 14.32mm
Target reach: 2 km over duplex single-mode fiber
Fully passive, wafer-scale assembly—no lenses, no active alignment.

Performance Highlights

As network operators transition from 400G to 800G and 1.6T architectures, traditional assembly techniques are reaching physical and economic limits. Wire-bonded packages and micro-optical alignments add cost, magnify RF crosstalk and thermal challenges, and restrict yield.

Profit margins become thinner when using current approaches. A better way of moving data from one point to another has been sought for years. The POET Optical Interposer and the products built from it — including the Hybrid-Integrated 1.6T 2xFR4 Transmitter PIC — are perhaps the best solution for an industry in need. POET eliminates active alignment and wire bonds, providing a high-throughput, CMOS-compatible path to scale.

Product Overview: Hybrid-Integrated 1.6T 2xFR4 Transmitter PIC

Compact, integrated, and ready for volume production, the 1.6T 2xFR4 transmitter PIC integrates the below components:

Four 2x200G EML arrays flip-chip bonded on POET’s optical interposer.
Eight high-speed driver ICs, symmetrically positioned for optimal signal integrity.
Two Arrayed Waveguide Gratings (AWGs) used as Multiplexers (MUX) for efficient wavelength multiplexing.
Spot-Size Converters (SSCs) mode-matched to SMF with <0.5 dB coupling loss.

Key Specifications:
Total capacity: 1.6Tbps (8x106 Gbaud) PAM4 lanes
Chip size (WxL): 7.975mm x 14.32mm
Target reach: 2 km over duplex single-mode fiber
Fully passive, wafer-scale assembly—no lenses, no active alignment.

Performance Highlights

Optical Eye Performance: At 106 Gbaud PAM4, all eight lanes demonstrate clear optical eyes with minimal penalty, confirming precise signal fidelity and excellent thermal uniformity.

Customer Benefits: Performance and Practicality Combined

The size, ease of adoption, functionality, value, and performance makes POET’s Hybrid-Integrated 1.6T 2xFR4 transmitter PIC a breakthrough solution for customers. The industry can unlock more of AI’s capabilities because POET’s advanced transmitter allows for an exponential increase in connectivity, data transfer and compute power. It is poised to become an essential component to the manufacturing of hardware devices for AI and data center hyperscalers. Among its notable features are:

1. Superior RF Integrity: Wire-bond-free interconnects deliver impedance-matched, low-loss RF paths.
2. Manufacturing Scalability: Passive wafer-level bonding enables automated, high-yield production.
3. Compact Thermal Design: Shorter electrical paths and reduced component count simplify cooling.
4. Future-Proof Scalability: The hybrid approach scales seamlessly to 3.2T and beyond.

Technology Foundation: The POET Optical Interposer™

POET’s photonic integration solutions are based on the POET Optical Interposer, a multiple award-winning novel, patented platform that allows the seamless integration of electronic and photonic devices into a single chip using advanced wafer-level semiconductor manufacturing techniques. POET's Optical Interposer-based products are lower cost, consume less power than comparable alternatives, are smaller in size and are readily scalable to high production volumes. With the optical interposer, POET is able to address the demand for products that can be applied to both scale-up and scale-out networks.

A universal platform for hybrid integration, the POET Optical Interposer has these key features:

Monolithic integration of all passive optical elements (splitters, combiners, MUX/DMUX, waveguides, mirrors).
Flip-chip bonding of all active components (lasers, modulators, drivers, TIAs).
Multiple RF redistribution layers with low insertion loss and TSV compatibility.
Passive wafer-level “pick-and-place” assembly—no active alignment required.
Cross-material compatibility (InP, TFLN, SiPh, and others).

Conclusion: Accelerating the Path to 1.6T and Beyond

Using its “semiconductorization of photonics” approach, POET is redefining how light and electronics converge—creating the foundation for a faster, smarter, and more connected world. By improving chip-to-chip data communication within and between AI servers, POET is at the forefront of the next frontier of computing applications. Its innovations solve bandwidth and latency problems in AI systems as well as device integration challenges in 5G networks, machine-to-machine communication, self-contained "edge" computing applications and sensing applications.

POET’s Hybrid-Integrated 1.6T 2xFR4 transmitter PIC specifically enables a single, scalable platform adaptable to diverse optical engine configurations—ideal for 800G, 1.6T, and 3.2T applications. It’s a product that addresses the needs of today and prepares the industry to meet the demands of tomorrow.

[longz] "Poet Inside"?.... DogeDesigner on X: "BREAKING: Elon Musk says the fastest and ...

longz — 12/8/2025 10:32:58 AM

"Poet Inside"?.... DogeDesigner on X: "BREAKING: Elon Musk says the fastest and cheapest path to massive AI growth is to move computing into space. Satellites with onboard AI can generate huge amounts of compute using constant sunlight and send results back to Earth through laser links. This avoids Earth’s limits on t.co; / X

[X]

BREAKING: Elon Musk says the fastest and cheapest path to massive AI growth is to move computing into space.

Satellites with onboard AI can generate huge amounts of compute using constant sunlight and send results back to Earth through laser links. This avoids Earth’s limits on… pic.twitter.com/yjUaCgiGfq
— DogeDesigner (@cb_doge) December 7, 2025

[/X]

[VApeanut] Click here for Fabian's POET Substack Please read Fabian's latest POET Substack...

VApeanut — 12/4/2025 3:32:56 PM

Click here for Fabian's POET Substack

Please read Fabian's latest POET Substack. I think you will enjoy his thoughts.

VApeanut

.

[longz] Why Shares of Poet Technologies Are Soaring Today | The Motley Fool

longz — 12/4/2025 1:15:35 PM

[longz] Mitsubishi Electric to triple output capacity for optical devices - Nikkei Asia

longz — 12/4/2025 12:57:04 AM

[longz] POET Technologies — the Next NVIDIA | by TJ Roberts | Nov, 2025 | Medium

longz — 12/4/2025 12:40:22 AM

[longz] .

longz — 12/4/2025 12:22:48 AM

[Heywood40] One

Heywood40 — 12/3/2025 8:09:33 PM

[longz] Autism Capital ?? on X: "??NEW: Jensen Huang, NVIDIA CEO, says "President Trump ...

longz — 12/3/2025 7:57:19 PM

Autism Capital ?? on X: "??NEW: Jensen Huang, NVIDIA CEO, says "President Trump singlehandedly flat out saved the AI industry. If not for his pro-growth energy policy we would not be able to build factories for AI, no supercomputer factories, none of this would be possible. We need energy growth. We t.co; / X

[X]

🚨NEW: Jensen Huang, NVIDIA CEO, says "President Trump singlehandedly flat out saved the AI industry. If not for his pro-growth energy policy we would not be able to build factories for AI, no supercomputer factories, none of this would be possible.

We need energy growth. We… pic.twitter.com/fsLFC0oLk7
— Autism Capital 🧩 (@AutismCapital) December 3, 2025

[/X]

[toccodolce] [X] 🚨Why I think $MRVL acquisition of Celestial AI is a MASSIVE tailwind for $...

toccodolce — 12/3/2025 11:12:37 AM

[X]

🚨Why I think $MRVL acquisition of Celestial AI is a MASSIVE tailwind for $POET technologies

Marvell just shook the semiconductor stack.

The Celestial AI acquisition, $3.25B upfront, potentially $5.5B, is more than a headline.

It’s a direct swing at the memory wall. A very… pic.twitter.com/QXAxXRDifZ
— Gaetano (@crux_capital_) December 3, 2025

[/X]

[toccodolce] [X] 🚨Why I think $MRVL acquisition of Celestial AI is a MASSIVE tailwind for $...

toccodolce — 12/3/2025 9:43:43 AM

[X]

🚨Why I think $MRVL acquisition of Celestial AI is a MASSIVE tailwind for $POET technologies

Marvell just shook the semiconductor stack.

The Celestial AI acquisition, $3.25B upfront, potentially $5.5B, is more than a headline.

It’s a direct swing at the memory wall. A very… pic.twitter.com/QXAxXRDifZ
— Gaetano (@crux_capital_) December 3, 2025

[/X]

[X]

🚨Save this as a simple cheat sheet for what these frequently talked about companies actually do in AI.$NBIS $NVDA $AMD $CIFR $POET

This is a broad overview of how each one fits into the AI buildout. Enough so that you can read a ticker and picture the real-world business… pic.twitter.com/TNDYhQTUK3
— Gaetano (@crux_capital_) November 30, 2025

[/X]

[X]

We've talked alot about $POET in the context of AI data centers. The telecom side of the story gets less attention, but that’s where a lot of real-world traffic actually flows – between the core network and the cell tower.

POET’s telecom work starts with a simple question:
How… pic.twitter.com/V0IJ4KPw3Y
— Gaetano (@crux_capital_) November 17, 2025

[/X]
[X]

POET x QUBT: A Massive Step Forward!

POET Technologies $POET and Quantum Computing Inc. $QUBT just announced a plan to co-develop a 3.2-terabit optical engine. On paper, it could seem like another product milestone. In reality, it validates one of POET’s biggest strengths.… pic.twitter.com/ZVb2CRz3IK
— Gaetano (@crux_capital_) November 11, 2025

[/X][X]

Some additional thoughts on $POET and $QUBT

QCi is an "integrated photonics and quantum optics" company. Their whole approach to quantum computing is photonic-based, meaning their systems are built to run on light.

To build their own advanced quantum machines, they needed… https://t.co/eFZXuZrl9p
— Gaetano (@crux_capital_) November 15, 2025

[/X]

[toccodolce] POET Technologies — the Next NVIDIA POET Technologies is a semiconductor compan...

toccodolce — 12/2/2025 9:50:37 PM

POET Technologies — the Next NVIDIA

POET Technologies is a semiconductor company with a proprietary, disruptive product that offers a foundational, core component that is at the heart of AI ; as NVIDIA is to GPU’s, POET will be to AI infrastructure; basically, the company has “semiconductorized” the electrical field of photonics by producing optical devices much like standard CMOS chips; the company’s goal is to become the low-cost, wafer-scale standard for AI optical engines/transceivers, and this feat will make them a valuable gold nugget to own — which in the low $4 range is possible now.

The company’s disruptive technology is called the POET Optical Interposer™; it’s a novel wafer-level semiconductor process that enables seamless, chip-scale integration of electronic and photonic devices into a single multi-chip module; this eliminates traditional wire bonds, reduces RF crosstalk and enables higher speeds up to 3.2 Tbps (trillion bits per second) with up to ~70% lower power costs, ~50% reduced laser needs, and human labor reduction of up to ~90%; in simple terms, the company has come up with a better mousetrap.

POET’s tech doesn’t fully eliminate humans, but it drastically reduces human labor by automating –90% of the traditionally manual steps, making production more like consumer electronics fabs than boutique photonics assembly. Conventional photonic integration relies on “artisanal” active alignment — manually or robotically positioning ~50 components with nanometer precision (20–50 nm), which is error-prone, slow, and labor-heavy.

Interestingly enough, what’s in a name? — everything! Some consider POET to be named after Photonics Optical Engine Technology which sounds right, but it is not! Instead, it is a passionate name chosen to evoke the idea of “poetry in motion” or innovative, elegant engineering in photonics — reflecting the company’s focus on simplifying complex optical integrations.

For a company to reach NVIDIA heights, marketing is everything. At the OFC 2025 in San Francisco in April, POET showcased their groundbreaking tech and won the top industry award right at the event’s start; this boosted their visibility and generated significant buzz in the optical communications and AI hardware arenas. At the OFC, POET impressed with live demos and solidified partnerships, positioning them as a frontrunner in next-gen optical tech for AI and data centers. In addition, POET won the “ Most Innovative Chip-scale Packaging/Optical Sub Assembly Product” award at the ECOC Awards in September, so they clearly have some serious momentum behind them.

POET is a mature company; they’ve been developing their core technology for ~40 years; their leader, CEO Dr. Suresh Venkatesan, is an industry veteran with over 22 years of experience in semiconductor development and holds over 25 US patents and has co-authored over 50 technical papers. When Dr. Venkatesan was the Senior VP at GlobalFoundries, he led the development and scale of 28nm semiconductor technology and was instrumental in the tech transfer and qualifications down to 14nm. The company’s CFO, Thomas Mika, is also an industry veteran with over 30 years managing major company finances. Company management is critical for success, and POET has it.

POET has raised over $350M in cash which should be enough runway to get them through 2027, and by that time either they’ll have started generating serious commercial revenue and be profitable, or they’ll get acquired. In order to raise that capital, they’ve had three capital raises. Of the three, the last one did not have any warrants attached, and very interestingly, the monies raised came in at $7.25 per share.

They have serious partnerships with Sivers Semiconductors, NTT Innovative Devices, Quantum Computing Inc, Foxconn Interconnect Technology (FIT), Luxshare Tech, Celestial AI, Adtran, Semtech Corporation, Globetronics and NationGate. POET shipped samples of 400G/800G optical engines to three unnamed major technology leaders in March, 2025, likely including some of the partners aforementioned. In Q3 2025, two unnamed key customers placed initial production orders totaling over $5.6M, proving active commercial activity. Other potential partners like Mitsubishi Electric, MultiLane, and Lessengers are in discussions or expected to place orders in late 2025. These partnerships focus on AI-driven applications, with POET’s tech enabling lower power, smaller form factors, and cost-effective scaling.

They have serious partnerships with Sivers Semiconductors, NTT Innovative Devices, Quantum Computing Inc, Foxconn Interconnect Technology (FIT), Luxshare Tech, Celestial AI, Adtran, Semtech Corporation, Globetronics and NationGate. POET shipped samples of 400G/800G optical engines to three unnamed major technology leaders in March, 2025, likely including some of the partners aforementioned. In Q3 2025, two unnamed key customers placed initial production orders totaling over $5.6M, proving active commercial activity. Other potential partners like Mitsubishi Electric, MultiLane, and Lessengers are in discussions or expected to place orders in late 2025. These partnerships focus on AI-driven applications, with POET’s tech enabling lower power, smaller form factors, and cost-effective scaling.

Current photonic tech is expensive due to fragmented assembly — active alignment equipment, wire bonds, multiple yield-killing touch points, and high material waste. POET’s Interposer cuts costs by streamlining wafer-level integration which leverages existing CMOS infrastructure for economies of scale without massive new CAPEX. Compared to competitors like Ayar Labs, a TSMC-backed CPO, or traditional pluggables, POET’s tech targets lower cost-per-bit while being smaller and lower-powered. Quantified savings vary by POET products but are significant.

POET Technologies is already manufacturing at scale with sites in Allentown, Pennsylvania, and Xiamen, China, which was acquired in December, 2024; they also have R&D operations in Shenzhen, Singapore and Toronto. POET has also outsourced production in Penang, Malaysia, to Globetronics and NationGate.

It’s news of potential orders hitting in Q4 from big customers like FoxConn, Quantum et al that will move the needle; this bet could be a serious windfall for retail investors. Hope you can get some POET before it pops. As always, please do your own due diligence, and spend only that which you are willing to lose. Good luck.

[toccodolce] [X] $MRVL Celestial AI acquisition. None of us will remember how MRVL traded ...

toccodolce — 12/2/2025 9:44:31 PM

[X]

$MRVL Celestial AI acquisition.

None of us will remember how MRVL traded tomorrow several years from now.

But if Celestial achieves $500M or $2B cumulative revenue by C28 (or ~ F29) or $500M run rate revenue by Q4 C27, this moment may be one of the most transformative for… https://t.co/XjjfPNbs7U pic.twitter.com/sAbvlkHjDp
— QF Research (@ResearchQf) December 2, 2025

[/X]

Marvell to Acquire Celestial AI, Accelerating Scale-up Connectivity for Next-Generation Data Centers

Celestial AI’s breakthrough Photonic FabricTM technology platform enables optical I/O for package, system and rack-level connectivity for the next-generation of data center infrastructure

SANTA CLARA, Calif.--(BUSINESS WIRE)-- Marvell Technology, Inc. (NASDAQ: MRVL), a leader in data infrastructure semiconductor solutions, today announced it has entered into a definitive agreement to acquire Celestial AI, a pioneer of a disruptive Photonic Fabric technology platform purpose-built for scale-up optical interconnect. This strategic acquisition represents a transformational milestone in accelerating the company’s connectivity strategy for next-generation AI and cloud data centers.

AI is reshaping data center architectures at unprecedented speed. Next-generation accelerated systems are no longer confined to a single rack – they are evolving into multi-rack configurations connecting hundreds of XPUs with an integrated high-bandwidth, ultra-low latency, any-to-any scale-up fabric. This architecture allows each XPU to access the memory of every other XPU directly. These advanced fabrics demand purpose-built switches and protocols such as UALink, engineered to deliver the performance and efficiency required at scale.

Given the power, bandwidth, latency, and reach requirements of multi-rack scale-up fabrics, interconnects will increasingly transition to all-optical connections. This acquisition positions Marvell to lead this technology shift and capture a brand-new semiconductor TAM for optical interconnects.

Combined with existing leadership in scale-out and scale-across connectivity, Marvell expects this expanded portfolio will result in the industry’s most comprehensive provider of high-bandwidth, low-power, low-latency solutions for next-generation data center connectivity.

“The acquisition of Celestial AI is a transformative step in Marvell’s evolution and expands our leadership in AI connectivity, as scale-up becomes the next frontier in AI infrastructure,” said Matt Murphy, Chairman and CEO of Marvell. “This builds on our technology leadership, broadens our addressable market in scale-up connectivity, and accelerates our roadmap to deliver the industry’s most complete connectivity platform for AI and cloud customers.”

“AI infrastructure is transforming faster than ever, and the future demands scale-up fabrics that deliver unprecedented bandwidth, power efficiency, and reach,” said Sandeep Bharathi, President, Data Center Group. “By combining our UALink scale-up switch roadmap and Celestial AI’s breakthrough optical scale-up interconnect, we will enable customers to build AI systems that scale beyond the limits of copper and redefine what’s possible in AI data center architecture.”

“At AWS, we aim to be at the forefront of major technology inflections, and we believe optical interconnects will play an important role in the future of AI infrastructure. Building a scalable, high-performance, and power-efficient cloud starts with an approach built upon differentiated technologies,” said Dave Brown, Vice President of Compute and Machine Learning Services at AWS. “Celestial AI has made impressive progress, and we expect their combination with a large-scale semiconductor company like Marvell will help further accelerate optical scale-up innovation for next-generation AI deployments."

Celestial AI: Breakthrough Photonic Fabric Technology Platform for Scale-Up Interconnect

As bandwidth and reach continue to increase, every connection point in the data center must move from copper to optical. For rack-to-rack scale-out and data-center to data-center DCI connections this transition has already taken place. The next inflection point is within the rack, within the system, and even within a package – where electrical connections must now give way to optics.

Celestial AI’s Photonic Fabric technology platform was purpose-built for this inflection. It enables large AI clusters to scale both within and across racks using a high-bandwidth, low latency, low power and cost-effective optical fabric. This breakthrough delivers a true optical solution with more than two times the power efficiency of copper interconnects, along with far longer reach and significantly higher bandwidth. Compared to alternative photonic technologies, Celestial AI’s solution provides exceptionally low power consumption, nano-second-class latency and excellent thermal stability – enabling deeper levels of optical interconnectivity into XPUs and switch systems.

The thermal stability of Celestial AI’s Photonic Fabric technology is a significant competitive differentiator. It enables reliable operation in the extreme thermal environments created by large, multi-kilowatt XPUs and switches. This allows the photonics technology to be co-packaged vertically with high-power XPUs and switches in a 3D package, enabling the photonic connection to be made directly into the XPU, rather than from the edge of the die. Celestial AI’s approach results in a more compact and integrated solution, while freeing up highly valuable die edge beachfront, which can be repurposed to significantly increase the amount of HBM within the XPU package.

The first application of this technology platform will be all-optical scale-up interconnects, as high-speed XPU links transition from copper to optics to meet the reach and bandwidth demands of next-generation rack-scale architectures. Celestial AI’s first-generation Photonic Fabric chiplet for scale-up interconnect integrates all the required electrical and optical components into a compact form factor. It is the industry’s first scale-up optical solution delivering an unprecedented 16 terabits per second of bandwidth in a single chiplet – 10 times the capacity of today’s state-of-the-art 1.6T ports used in scale-out applications. Its compact form factor allows multiple PF chiplets to be co-packaged with XPUs and the scale-up switches on the opposite side of the link, enabling substantial increases in total system bandwidth.

Celestial AI is deeply engaged with multiple hyperscalers and ecosystem partners that plan to use Celestial AI’s Photonic Fabric technology in their next-generation scale-up architectures. Based on existing customer traction, Marvell expects Celestial AI’s Photonic Fabric chiplets to be co-packaged with custom XPUs and scale-up switches, enabling the industry’s first large-scale commercial deployment of optical interconnect for scale-up connectivity.

Marvell expects meaningful revenue contributions from Celestial AI to begin in the second half of fiscal 2028, reaching a $500 million annualized run rate in the fourth quarter of fiscal 2028, doubling to a $1 billion dollar run rate by the fourth quarter of fiscal 2029.

Beyond connecting XPUs in scale-up networks, Celestial AI’s Photonic Fabric technology platform can enable a wide range of transformational applications over time, including pooled memory appliances and optical replacements for traditional electrical die-to-die connections in multi-die packages.

“Marvell is the ideal home for our Photonic Fabric, with the scale, customer relationships, and connectivity leadership to take this platform into high-volume production,” said David Lazovsky, Co-Founder and CEO of Celestial AI. “Together with Marvell’s scale-up switching strategy, we’re excited to accelerate the transition to optical scale-up interconnect and expand what next-generation AI infrastructure can achieve.”

Transaction Structure and Terms

Marvell will acquire Celestial AI for upfront consideration valued at approximately $3.25 billion, consisting of $1.0 billion in cash, as well as approximately 27.2 million shares of Marvell common stock, having a value of $2.25 billion, based on the average volume weighted average price (VWAP) for the 10 trading days ending the second day prior to signing. In addition, Marvell will pay Celestial AI equityholders incremental contingent consideration of up to approximately 27.2 million shares of Marvell common stock, having a value of up to $2.25 billion, based on the same 10 trading day VWAP, upon satisfaction of certain revenue milestones. The first milestone, representing one-third of the earnout consideration, will be achieved if Celestial AI reaches cumulative revenue of at least $500 million by the end of Marvell’s fiscal year 2029. The full earnout would be paid if Celestial AI’s cumulative revenue by the end of Marvell’s fiscal year 2029 exceeds $2.0 billion.

The transaction is expected to close in the first quarter of calendar 2026, subject to customary closing conditions and regulatory approvals.

[toccodolce] Latest from Agoracom: Marvell nears multi-billion dollar deal to acquire Celest...

toccodolce — 12/2/2025 11:07:40 AM

Latest from Agoracom: Marvell nears multi-billion dollar deal to acquire Celestial AI: report

Reported deal would give Marvell access to Celestial’s pipeline of photonic interconnect technology

December 02, 2025 By Scarlett Evans, SDxCentral Have your say

U.S. chipmaker Marvell Technology is reportedly close to acquiring chip startup Celestial AI in a multi-billion-dollar deal.

The potential deal was first reported by The Information, which cited people close to the matter. According to the report, the total deal price (including potential earnings from product milestones) could be higher than $5 billion.

The purchase would see Marvell build out its data center capacity, adding Celestial’s photonic interconnects to its technology portfolio.

Photonics is a growing area of interest for hardware developers, with light being used instead of electrical signals to enhance communication between AI compute and memory chips. The technology is designed to address typical bottlenecks by delivering high-bandwidth at low-latency, meaning it can ferry data between chips more rapidly compared to traditional hardware.

The deal would also look to support Celestial AI’s wider expansion plans.

The Santa Clara, California-based firm, whose board includes Intel CEO Lip-Bu Tan, was founded back in 2020. It raised $250 million in venture capital from backers including AMD Ventures, BlackRock, and Koch Disruptive Technologies, bringing its total capital raised to $515 million.

Celestial said at the time the funding would be used to accelerate commercial deployment of its photonic technology, including its Photonic Fabric offering, as well as scaling its manufacturing supply chain, and deepen industry partnerships as businesses look to enhance AI infrastructure.

Instead of being just another co-packaged optics (CPO) offering, Celestial’s Photonic Fabric platform is a full-stack technology capable, which the firm claims can transmit data between data center chips with up to a 25x increase in off-package bandwidth compared to rival hardware, all while consuming substantially less power.

Speaking following its March funding round, Celestial CEO David Lazovsky said the company’s development of photonics comes as the development of complex reasoning models and agentic AI sparks greater AI infrastructure demands.

As a result, companies are chasing faster and more efficient processing speeds to accommodate vast data needs without jeopardising productivity.

“Celestial AI’s Photonic Fabric is the only technology platform capable of meeting these critical demands while setting new standards for bandwidth, latency, energy efficiency, and total cost of ownership,” he added.

Celestial was one of a handful of AI hardware developers subject to AI-generated “executive summaries” SDxCentral unearthed following an inadvertent leak from xAI chatbot Grok.

Instead of being an acquisition, Grok suggested the startup would aim for an IPO by 2027 – an approach disputed by the firm.

A Celestial spokesperson told SDxCentral that information generated by Grok was being “either out of date or inaccurate” – with the AI even recommending users contact Celestial staff who didn’t exist.

Acquisition to boost Marvell’s efforts to break bottlenecks

Marvell’s reported Celestial acquisition would add to its growing efforts aimed at breaking AI and cloud infrastructure bottlenecks.

The vendor has already been dipping its toe into the world of photonics, such as with a linear-drive pluggable optics (LPO) module displayed at OFC 2025.

The offering, the second in its light engine family, offers 1.6 Tb/s – spread across eight channels of 200 Gb/s PAM4 – to act as what Marvell described as “a foundation for CPO system development with hyperscale customers.”

The chip giant teamed up with power conversion specialist Ferric back in June to jointly develop efficient power delivery solutions for increasingly hungry processors.

That partnership saw Ferric's miniaturized voltage regulators integrated directly into Marvell’s custom silicon platforms, a move the pair promised would offer significant performance gains and cost savings for hyperscale cloud operators.

Despite the prospective Celestial adding to its nascent photonic portfolio,

Marvell expanded its copper-focused designs earlier this year.

Marvell's active copper cables (ACC) feature a small signal equalizer chip to extend reach while maintaining minimal latency, offering the cable improved reach needed within server racks, compared to traditional DAC offerings.

[toccodolce] Latest from Agoracom: MVRL CAI implications Per the request of some users: ...

toccodolce — 12/2/2025 11:01:13 AM

Latest from Agoracom: MVRL CAI implications

Per the request of some users: (I'm sorry for the formatting in advance. I am at work on mobile and do not have access to my laptop for improved formatting).

Me:

What would a celestial AI acquisition by Marvell mean for POET? Is this good or bad? Does it mean anything? Should I be worried?

CHATGPT:

Below is the straight, unsweetened assessment you’re asking for — Critical Mode on.

No hedging, no optimism-padding.

---

1. Does Marvell acquiring Celestial AI mean anything for POET?

Yes — but not in the simplistic “good/bad” way you might hope for.

It’s a strategic data point, not a thesis-breaker.

Marvell acquiring Celestial AI signals one thing clearly:

? The silicon photonics consolidation wave has officially started.

That matters because POET is a small player with a differentiated platform in a market where large players are starting to integrate vertically.

But don’t over-interpret. This deal does not point directly at POET’s success or failure.

---

2. Is this good or bad for POET? (Critical analysis)

Potentially Bad / Headwinds

This is the part most investors gloss over, but you should take seriously:

1. Marvell now owns an integrated photonics stack that competes with POET’s value proposition.

Celestial AI was one of the better-funded early movers in optical interconnects.

Marvell just absorbed that IP.

2. Bigger incumbents absorbing startups raises the bar for POET.

POET’s “IP moat” is meaningful, but now it’s being measured against companies with:

massive integration resources

existing datacenter customer relationships

ability to deploy at scale sooner

3. Marvell may seek to close their ecosystem around their newly acquired photonics assets.

Closed ecosystems reduce the number of customers who might turn to POET.

4. This reduces the number of independent photonic players available for POET partnerships or eventual acquisition.

Potentially Good tailwinds

Not everything is negative; there are structural positives too:

1. Validation:

A large semiconductor company buying a photonics interconnect startup proves the market trend is real.

This is validation of POET’s entire category:

chiplet architectures

optical I/O

alternative integration methods

power-efficiency critical paths

2. Consolidation increases the scarcity value of POET.

If big companies are shopping, POET is now a rarer target.

3. Competition between big players opens gaps for neutral platform providers.

Marvell locking into Celestial may push other players (Broadcom, AMD, NVIDIA, Intel, Amazon, hyperscalers) to look harder at non-Marvell photonic solutions.

4. POET’s technology is still structurally different.

Celestial AI was focused on optical memory fabric and optical interconnect using its proprietary Photonic Fabric.

POET’s platform solves:

low-cost assembly

chiplet partitioning

laser integration

efficient coupling

These are orthogonal problems, not redundant.

So the overlap is partial, not direct.

---

3. Should you be worried? (Direct answer)

No — not in the sense of “POET is threatened by this.”

But yes in the sense that the competitive field just got more serious.

You shouldn’t panic.

You should treat this as a reminder that:

execution matters more than technology alone

big players will consolidate aggressively

partnerships, not patents, will determine winners

POET needs commercial proof, not demos

the “small agile innovator wins” story gets harder in a consolidation wave

This isn’t a thesis-breaker for POET.

It’s a pressure-raiser.

---

4. The blunt summary

Not a disaster for POET.

Not a direct positive either.

It increases the competitive temperature and likely accelerates timelines.

If POET executes — great.

If they stumble — consolidation like this will make the market far less forgiving.

---

Me:

We know that POET is the enabler of CAI. According to Lisa Thompson from Zach's scr, celestial placed a 5million dollar PO. DOES THIS CHANGE YOUR STANCE?

ChatGPT:

Short answer: It changes the interpretation, but it does NOT remove the strategic risk.

It improves POET’s relative position if the CAI purchase order is real and verifiable — but it also sharpens the stakes.

Here’s the hard-edged breakdown.

---

1. If POET truly is the enabler of CAI’s product ? Marvell just bought one of POET’s customers

This immediately flips part of the earlier dynamic.

If Celestial AI is using POET’s platform as a foundational integration layer and placed a $5M purchase order, then:

? Marvell now indirectly depends on POET for a critical piece of its newly acquired photonics roadmap.

Companies don’t like that, but they tolerate it when:

the tech is indispensable

replication is expensive and slow

supplier lock-in delay is unacceptable

That means POET’s technology is:

validated

inside the pipeline

financially committed

hard to replace quickly

This is unequivocally good for POET in the short and medium term.

---

2. But here’s the uncomfortable truth investors hate hearing

Big companies eventually try to own what they depend on.

Especially Marvell.

If CAI is structurally dependent on POET’s integration platform, Marvell will eventually face three choices:

1. Acquire POET

2. License POET broadly

3. Replace POET with internal tech (takes years and isn’t guaranteed)

Only one of these is bad for POET — option #3 — and even that requires time and expertise Marvell does not currently have.

But you need to be realistic:

Being a critical supplier puts POET in the spotlight.

Exactly where you want to be — but also where giants start paying very close attention.

---

3. Does the $5M PO change the stance?

It changes some parts materially:

(1) It removes the biggest prior risk

If POET is already in CAI’s product stack with a multi-million PO, then:

POET is not “left behind”

POET is not competing directly with CAI

POET is actually enabling the technology Marvell just spent serious money on

So the competitive worry is reduced.

POET is upstream, not rivalry-side.

(2) It increases POET’s strategic value dramatically

Marvell did not buy CAI for their packaging, lasers, or O-Band integration.

If CAI’s implementation fundamentally relies on POET’s platform:

Marvell inherits a dependency

Marvell now has to take POET seriously

POET becomes acquisition or licensing optionality

This is leverage.

(3) It DOES NOT eliminate long-term risk

And here’s where investors get sloppy:

Big companies eventually internalize critical IP.

Supplier displacement is always on the table.

Marvell’s internal roadmap might try to replicate POET’s capabilities over 3–7 years.

Until then?

POET has a window of high leverage and high visibility.

---

4. Should you still worry?

Not the way you were before.

Now the risk has shifted:

Before ? “Is POET bypassed or irrelevant?”

After ? “Can POET capitalize before Marvell or others try to internalize?”

Very different.

The first risk was existential.

The second is strategic and competitive, but not fatal — and often leads to acquisition outcomes.

If the $5M PO is legitimate and CAI depends on POET, then:

POET is now too integrated for Marvell to ignore

and too small for them not to consider buying if the tech proves indispensable.

This is what “strategic leverage” looks like.

---

5. Updated stance in one line

POET’s near-term outlook improves meaningfully; long-term risk morphs into “will a giant absorb them or try to replicate them?” instead of “will they be left behind?”

[toccodolce] Semtech’s High-Speed Chips Fuel AI Data Center Growth Semtech Corporation suppl...

toccodolce — 12/2/2025 10:50:52 AM

Semtech’s High-Speed Chips Fuel AI Data Center Growth

Semtech Corporation supplies essential high-speed chips for AI data centers, enabling efficient data transmission amid explosive AI growth. Their optical and analog semiconductors address power and bandwidth bottlenecks, positioning the company for substantial gains through partnerships and innovations. This underscores Semtech's pivotal, yet understated, role in sustaining AI infrastructure.

Written by John Marshall
Monday, December 1, 2025

The Hidden Wiring Behind AI’s Explosive Growth: Semtech’s Crucial Role in Data Center EvolutionIn the relentless surge of artificial intelligence, where giants like Nvidia dominate headlines with their powerful GPUs, a quieter revolution is unfolding in the bowels of massive data centers. Semtech Corporation, a lesser-known player in the semiconductor arena, is emerging as a pivotal force, supplying the high-speed chips that serve as the essential conduits for AI’s voracious data demands. These components aren’t the flashy processors training neural networks; instead, they’re the unsung infrastructure enabling seamless communication between servers, ensuring that AI systems can process information at breakneck speeds without choking on power or bandwidth constraints.

Semtech’s offerings, particularly in optical and analog semiconductors, address critical bottlenecks in data centers scaling up for AI workloads. As companies like Microsoft and Google pour billions into expanding their computing empires, the need for efficient data transmission has skyrocketed. Semtech’s chips facilitate this by enabling faster, more energy-efficient connections, turning what could be a gridlock into a high-speed highway. Recent reports highlight how these technologies are integral to the next wave of AI infrastructure, with analysts projecting substantial growth for the company amid the broader tech boom.

This isn’t just about keeping pace; it’s about redefining efficiency. Semtech’s linear pluggable optics and active copper cables are designed to handle the immense data flows required by advanced GPUs, reducing latency and power consumption. Industry observers note that without such innovations, the AI expansion could stall under its own weight, as data centers grapple with escalating energy demands and thermal challenges.

Semtech’s Technological Edge in High-Speed Connectivity

Delving deeper into Semtech’s portfolio reveals a suite of products tailored for the AI era. Their Tri-Edge and FiberEdge platforms, for instance, support ultra-high-speed data rates up to 800Gbps and beyond, crucial for interconnecting clusters of AI accelerators. These chips excel in linear receive optics, which minimize signal distortion over long distances within data centers, a key factor as facilities grow larger and more complex.

According to a detailed analysis from Yahoo Finance, Semtech’s solutions are becoming the “essential plumbing” for AI data centers, directly addressing power and data bottlenecks for next-generation GPUs. This positions the company as a beneficiary of the AI gold rush, where the real fortunes may lie not in mining the gold—akin to GPU production—but in supplying the tools that make it possible.

Competitors like Broadcom and Marvell Technology are also vying for this space, but Semtech’s focus on specialized analog and mixed-signal semiconductors gives it a niche advantage. Broadcom’s involvement is noted in related coverage from The Globe and Mail, which discusses how Semtech’s chips complement broader ecosystems involving players like Broadcom in powering AI infrastructure.

Market Dynamics and Investor Sentiment

The financial implications are stark. Semtech’s stock has seen significant volatility, but recent upgrades from analysts signal optimism. Piper Sandler raised its price target to $70, citing robust data center growth, as reported by Investing.com. This comes on the heels of Semtech anticipating a 10% rise in fourth-quarter data center sales, driven by portfolio expansions including linear pluggable optics and active copper cables.

Posts on X, formerly Twitter, reflect buzzing sentiment in the tech community. Users are highlighting Semtech’s surge amid the AI and data center boom, with discussions pointing to its role in enabling efficient AI networking. One thread emphasized how Semtech’s innovations are part of a larger shift toward on-premise “AI factories” for enterprises, reducing reliance on cloud services for better cost control and data security.

This enthusiasm aligns with broader industry trends. A Nasdaq article, accessible via Nasdaq, draws parallels to historical gold rushes, arguing that providers of enabling technologies like Semtech stand to gain enduring benefits as AI investments accelerate.

Strategic Partnerships and Future Expansions

Semtech isn’t operating in isolation. Collaborations are amplifying its impact, such as the recent announcement with POET Technologies for high-performance 1.6T receiver optical engines, aimed at AI clusters and cloud networks. This partnership, detailed in a Simply Wall St piece at Simply Wall St, underscores Semtech’s push toward greater integration and energy efficiency in data center optics.

Such alliances are timely, as the AI sector faces mounting pressures from energy consumption. Data centers worldwide are projected to consume vast amounts of electricity, equivalent to entire nations’ outputs, prompting a race for more sustainable solutions. Semtech’s chips contribute by optimizing power usage in high-speed links, potentially alleviating some of these concerns.

Looking ahead, Semtech’s leadership has outlined ambitious growth trajectories. In its latest earnings call, covered by Seeking Alpha, the company highlighted expansions in LoRa technology alongside data center advancements, forecasting continued revenue upticks.

Challenges Amid Rapid Growth

Yet, this ascent isn’t without hurdles. The semiconductor industry is notoriously cyclical, and Semtech has faced its share of setbacks, including supply chain disruptions and competitive pressures. Recent stock movements, as analyzed in a TipRanks report from TipRanks, show surges tied to AI exposure but also underscore volatility risks.

Regulatory environments add another layer. With global scrutiny on AI’s environmental footprint, companies like Semtech must navigate evolving standards for energy-efficient tech. Moreover, intellectual property battles, as seen in posts on X about TSMC’s lawsuits against former executives now at Intel, highlight the intensifying competition in leading-edge chip development.

Despite these, Semtech’s strategic focus on AI-enabling tech positions it well. Analysts from Summit Insights, referenced in the earlier Investing.com piece, express optimism about adoption rates for Semtech’s optics and cables, predicting they will drive long-term growth.

Broader Implications for the AI Ecosystem

The ripple effects extend beyond Semtech. The AI boom is reshaping entire supply chains, with data center spending expected to balloon. X posts from industry figures like Beth Kindig cite AMD’s CEO projecting a $500 billion AI accelerator market by 2028, a figure rivaling the entire semiconductor industry’s annual sales just a few years ago.

This growth fuels demand for complementary technologies. Semtech’s role in high-speed connectivity dovetails with advancements from players like Nvidia and AMD, creating a symbiotic ecosystem. For instance, as GPUs evolve to handle more complex AI models, the need for robust interconnects intensifies, directly benefiting Semtech.

Investors are taking note. Discussions on X emphasize “picks and shovels” plays—companies providing essential tools rather than the core AI tech—with Semtech fitting this mold alongside testing equipment providers and power semiconductor firms.

Innovation Trajectories and Industry Shifts

Semtech’s innovations signal a shift toward more modular, scalable data center architectures. Their emphasis on active copper cables, which offer cost-effective alternatives to fiber optics for shorter distances, could democratize high-speed AI infrastructure for smaller operators.

This aligns with enterprise trends toward hybrid models, blending cloud and on-premise computing. As noted in X posts about Dell’s bullish stance on AI servers, businesses are prioritizing local data control for security and performance, amplifying the need for efficient internal networking solutions like Semtech’s.

Furthermore, global expansions are on the horizon. With AI adoption accelerating in regions like Asia and Europe, Semtech’s international footprint—bolstered by its analog expertise—could capture emerging markets hungry for data center upgrades.

Sustaining Momentum in a Competitive Arena

To maintain its edge, Semtech is investing heavily in R&D. Recent developments, such as the AI200 and AI250 chips mentioned in X discussions about Qualcomm’s entry into AI inference, illustrate the crowded field, but Semtech’s specialization in data transmission sets it apart.

Analyst outlooks remain positive. The IndexBox analysis at IndexBox reviews Semtech’s recent performance, noting stock rises following positive ratings and its deep ties to AI networking.

As the AI wave crests, Semtech’s contributions highlight how interconnected the tech ecosystem truly is. While GPUs grab the spotlight, it’s the underlying chips ensuring data flows unimpeded that will determine the sustainability of this revolution.

Emerging Opportunities and Long-Term Vision

Peering into the future, Semtech’s trajectory suggests opportunities in adjacent fields like edge computing and 5G integration, where low-latency data handling is paramount. Partnerships, like the one with POET, could pave the way for breakthroughs in optical engines, enhancing AI’s reach into real-time applications.

Industry sentiment on X underscores this potential, with threads predicting explosive growth for AI infrastructure providers. As tech giants commit trillions to data centers—evidenced by Microsoft’s $80 billion capex announcement—Semtech stands poised to capitalize.

Ultimately, Semtech’s story is one of quiet indispensability. In an era defined by AI’s transformative power, its chips are the vital links ensuring the machine keeps humming, powering not just computations but the very foundation of tomorrow’s intelligent world.