To: longz who wrote (843) | 5/15/2024 8:05:09 PM | From: macnai | | | Management Comments “Our team embarked on a dynamic first quarter preparing for the Optical Fiber Conference (OFC) held in San Diego in mid-March,” stated Chairman & CEO, Dr. Suresh Venkatesan. “We could not have guessed at the time that those efforts would lead to a collaboration with Foxconn, one of the world’s leading electronics manufacturing companies to include our optical engines in their 800G and 1.6T optical module products. We are proud to have been selected for their advanced module designs and we look forward to supplying FIT and its global customer base, which includes the largest consumers of optical modules in the AI market. At OFC, we showcased four new products, each of which garnered serious attention from the industry, with the strongest interest in our leading-edge optical engine technology that powers optical modules for AI processing clusters, and in our light source products that facilitate chip-to-chip light-based data communications and high-speed computing. In addition to Foxconn, we are expanding our previously announced relationship with Luxshare, to include additional optical module products, and collaborating with MultiLane, a key supplier of high-speed test equipment in the industry to offer 800G and 1.6T optical modules. The enthusiastic reception from OFC helped us to secure multiple successful capital raises from institutional investors committed to our company's success. The additional capital fortifies our financial foundation as we navigate the next phase of our growth.” |
| The end of Moore's law - Poet Technologies | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last Read |
|
From: toccodolce | 5/16/2024 9:13:42 AM | | | | More insight from Agoracom
On the contrary, The company cannot believe how fast they are in the position they find themselves in, being chosen to have their engines driving Foxconn's modules who happen to be the supplier to the "Tier 1's" that utilize these transceivers.
Line ''em up boys, this party is just getting started.
As for the Tier 1's Foxconn doesn't supply, how about this gem hot off today's press at 530pm:
Management Comments
"In addition to Foxconn, we are expanding our previously announced relationship with Luxshare, to include additional optical module products, and collaborating with MultiLane, a key supplier of high-speed test equipment in the industry to offer 800G and 1.6T optical modules."
I think it's the 200gb per lane engines we've just embarked on. They must be a hot product and must have turned heads at the OFC as it looks like the big boys have put their mandates in to all suppliers, we want higher speed, using less power while utilizing the same amount of Fiber. Guess who can deliver that? Ya, Poet Technologies. (The guys that also drive Celestial AI's entire photonics fabric with their light engines as well.)
In other words, we just got everyone covered, Luxshare can supply the other Tier 1's like Meta and Apple while Multilane will fast track the devices as they are one of only 2 Companies that can even test the speeds we are talking about (200gbps).
I don't think Suresh can even believe how fast the Interposer is set up to dominate 800g-3.2T (using the same form factor) which can satiate the appetite's of all the big boys who need these speeds to dominate in the AI space.
Basically, they can purchase the transceivers from their provider of choice, which happen to be 2 of the biggest contract manufacturers in the world. Both now aligned with Poet Inside* |
| The end of Moore's law - Poet Technologies | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last Read |
|
From: toccodolce | 5/16/2024 2:35:30 PM | | | | Surging demand in the rapidly growing optical transceiver market.
How Many Optical Transceivers Does ChatGPT Require?
FiberMall has extrapolated the AI infrastructure including optical transceivers that ChatGPT brings to the table.
The difference from a traditional data center is that with the InfiniBand fat tree structure common to AI, more switches are used and the number of ports upstream and downstream at each node is identical.
One of the basic units corresponding to the AI clustering model used by NVIDIA is the SuperPOD.
A standard SuperPOD is built with 140 DGX A100 GPU servers, HDR InfiniBand 200G NICs, and 170 NVIDIA Quantum QM8790 switches with 200G and 40 ports each.
Based on the NVIDIA solution, a SuperPOD with 170 switches, each switch has 40 ports, and the simplest way is to interconnect 70 servers each, and the corresponding cable requirement is 40×170/2=3400, considering the actual deployment situation up to 4000 cables. Among them, the ratio of copper cable: AOC: optical module = 4:4:2, corresponding to the number of optical transceivers required = 4000 * 0.2 * 2 = 1600, that is, for a SuperPod, the ratio of server: switch: optical module usage = 140: 170: 1600 = 1: 1.2: 11.4
A requirement similar to GPT4.0 entry-level requirements requires approximately 3750 NVIDIA DGX A100 servers. The requirements of optical transceivers under this condition are listed in the following table.
According to IDC, the global AI server market is $15.6 billion in 2021 and is expected to reach $35.5 billion by 2026. The market size of China’s AI server industry is $6.4 billion in 2021. According to IDC data, 200/400G port shipments are expected to increase rapidly in data center scenarios, with a compound growth rate of 62% from 22-26 years. Global switch port shipments are expected to exceed 870 million in 2026, with a market size of over $44 billion.
FiberMall extrapolates the demand for servers, switches, and optical transceivers from AI data center architecture. In this extrapolation process, FiberMall uses the ratio of 4:4:2. The use of optical modules in the data center is ultimately directly related to traffic demand. This ratio is likely to exist only at full capacity, and it is still worthwhile to study in depth how the service traffic within the AI data center is now. |
| The end of Moore's law - Poet Technologies | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last Read |
|
| |