Optical Computing Using Meta Materials And Silicon Photonics

Filed 54m ago · Via Forbes · The Buffoon Desk

THIS STORY IS SCORED

As I Figure

Kevin MacLeod · incompetech.com · CC BY 4.0

Photo: Gazouya-japan · CC BY-SA 4.0 · via Wikimedia Commons

Neurophos, based in Austin, builds optical compute units using silicon photonics and meta-materials that aim to perform AI calculations in the optical domain rather than converting light to electrical bits and back, a lossy process that drives up energy and water use in data centers. The company raised $110 million in a Series A led by Gates Frontier, with Microsoft's M12, Aramco Ventures, Bosch Ventures and others participating, and says it has quadrupled its workforce in 2026.

The piece traces optical computing's lineage back to 1960s Fourier-transform optics through 1980s vector-matrix multiplication research that never left the lab, arguing the field's moment has finally arrived alongside hyperscale data centers that now eat 100 megawatts to a gigawatt apiece. The company's CEO is quoted claiming the technology breaks free of Moore's Law limits by packing optical parallelism onto a single chip.

The full dispatch is available from the source below.

✒ FROM THE EDITORIAL DESK

A company spending someone else's hundred and ten million dollars will always say the breakthrough is bigger than physics itself, same as a fisherman's catch grows every time he tells the story at the hardware store. The data center power problem is real enough, nobody's arguing that. But replacing a trillion-dollar GPU market is a claim best measured in shipped chips, not in a founder's quote about breaking free from the laws of scaling.

Source: Read the original at Forbes → Scored: As I Figure · Kevin MacLeod · CC BY 4.0