UEC is the full-communications stack architecture to meet the growing network demands for high performance AI and HPC networking.  High-performance computing (HPC) is technology that uses clusters of powerful processors that work in parallel to process massive multi-dimensional data sets, also known as big data, and solve complex problems at extremely high speeds.  The consortium of companies set the standards. 

(We currently have Celestial AI, but the others remain to be seen.  IMO, we'll make a fortune just providing the light for CAI's optical interconnects.)

Here is where POET provides the light for the interconnects:

TORONTO, April 25, 2023 (GLOBE NEWSWIRE) -- POET Technologies Inc. (“POET” or the “Company”) (TSX Venture: PTK; NASDAQ: POET), the designer and developer of the POET Optical Interposer™ and Photonic Integrated Circuits (PICs) for the data center, telecommunication and artificial intelligence markets, today announced “POET Starlight™”, a packaged light source solution for AI applications. POET has entered into an agreement with Celestial AI, creators of the Photonic Fabric™, for development and production of POET Starlight packaged light sources and has received an advanced purchase order for initial production units.

“POET’s Starlight chips and its technology roadmap are complementary to our advanced silicon photonics and system-in-package technologies,” said Preet Virk, Co-founder and COO of Celestial AI.  “We are pleased to continue our collaboration with POET Technologies, building on its optical interposer technology to enable us to meet our performance, scale, and cost targets.”

The POET Starlight products are packaged light engines based on POET's LightBar™ platform, which recently earned industry praise when it was demonstrated at OFC. Chip developers are working to redesign the core of computer architecture to incorporate light for communication among chips or to perform complex computations, all in an effort to eliminate the bandwidth and power-efficiency bottlenecks inherent in electronic devices. Each such chip requires one or more light sources able to provide several frequencies of light integrated into the fabric of the device.

POET’s Starlight products integrate active components like lasers and certain other passive optical and electrical components into the optical interposer to provide a complete light source solution that can be integrated on a host board and connected to chips and ASICs with built-in silicon photonics. POET achieves its low-cost solution through the use of features and benefits of the POET Optical Interposer platform and the “semiconductorization” of photonics fabrication processes.

“External light sources for Artificial Intelligence applications has been one of the key verticals for POET Technologies. We are delighted to continue to work with Celestial AI to provide a high-volume manufacturable solution for their optical interconnect technology platform,” said Dr. Suresh Venkatesan, Chairman and CEO of POET Technologies.  “The ability for POET to passively attach high power CW lasers on its optical interposer at wafer scale enables the use of single known-good laser chips and eliminates the need for laser arrays which can be cost prohibitive for high-volume applications. The POET team has worked closely with Celestial AI to define a low-cost packaging solution that is up to 75% lower in cost than competing solutions and is highly scalable for the volumes that Celestial AI is projecting.”

POET’s Starlight C-Band and O-Band LightBar products incorporate monolithically integrated passive components like multiplexers, demultiplexers, splitters, combiners, and waveguides. Known-good high power CW lasers and monitor photodiodes are flip chipped and passively attached to the optical interposer platform. POET expects to complete design verification testing and reliability qualification of the Starlight engines by the end of 2023 and start production in the second half of 2024. The size of the AI chip market is projected by Allied Market Research to reach $263 billion by 2031.