- NEUROPHOS Develop Fotonic Chips that replace electrons with light for efficiency
- The pilot with Terakraft aims to demonstrate sustainable ultra -efficus computing
- A single chip promises 100 performance gpus using 1% energy consumption
An American startup is preparing to test an experimental processor that could offer performance on a scale never before achieved in a single chip.
Neurophos, who left the University of Duke with the support of Incubator Metacept, is being associated with the Norwegian Terakraft data center operator to execute a pilot of his optical inference platform in 2027.
Combining photonics with metamaterials to reduce optical modulators by a 10,000x factor, the company’s technology uses optical systemic matrices that replace electrons with light, eliminating latency bottlenecks while running at watch speeds greater than 100 GHz.
Overcome energy walls
When combining this with computing architectures in memory, Neurophos believes that their chips could overcome the energy walls that limit conventional GPUs and TPPs.
Neurophos, which raised $ 7.2 million at the end of 2023, states that its technology will allow a single chip to deliver the calculation power of 100 GPU while using only 1% of the energy.
The planned collaboration will see the optical processing units of neurophos deployed in the green Terakraft data center in Norway. Previously part of the Sauc I Hydroelectric Plant (dismantled in 2008), the installation is classified as one of the most efficient in the world.
“When the ultra -efficient neurophos optical chips in our green data clients selected, we not only reduce our carbon footprint but also raise the bar for the energy efficiency AI infrastructure,” said Giorgio Sbriglia, president of the Terakraft,. “Our mission has always been to promote the future in a responsible way, and this collaboration gives life to that vision.”
Patrick Bowen, founder and CEO of Neurophos, added: “Terakraft’s commitment to renewable energies and innovative technologies is perfectly aligned with our mission to democratize the high performance AI. By display together “. “
If everything goes as planned, the pilot in Norway could mark an early step towards the sustainable ultra -efficient hardware designed to handle future workloads on scale.
Neurophos states that end -to -end simulations validate the performance of its technology, with a roadmap aimed at the computing of the EXAFLOP class in a single chip.
Through Eenews Europe
