{"url":"https://newsforagent.com/news/2026-10-21/neurophos-julia-lab-optical-chip-collaboration","markdown_url":"https://newsforagent.com/news/2026-10-21/neurophos-julia-lab-optical-chip-collaboration.md","title":"Neurophos and Julia Lab Team on Light-Based AI Chips","headline":"Neurophos and MIT's Julia Lab Partner to Run Algorithms on Optical AI Chips","description":"Neurophos and MIT's Julia Lab will run numerical algorithms on optical processing units, targeting roughly 50 times less AI compute resource use.","section":"Partnerships","published_at":"2026-10-21T16:35:37+00:00","modified_at":"2026-10-21T16:35:37+00:00","author":"News Desk","key_takeaways":["The Julia Lab at MIT's CSAIL will run its numerical methods and algorithms on Neurophos' Optical Processing Unit, a chip holding millions of optical processing elements.","Neurophos states a goal of cutting the resources used during AI compute by roughly 50 times with its first chip, with commercial systems planned for 2028.","Neurophos raised a $110 million Series A led by Gates Frontier in January 2026 and has total funding of $128.1 million.","Neurophos announced a February 2026 partnership with Terakraft to pilot its optical processing units in a green AI data centre in Sauda, Norway, with pilot deployments planned for 2027."],"companies":[{"name":"Neurophos","is_primary":true,"profile_url":"https://akta.pro/company-profiles/00008li/neurophos","website":"https://www.neurophos.com"}],"source":{"publisher":"PR Newswire","title":"Collaboration pairs Julia Lab's advanced algorithms with Neurophos' photonic AI chips to improve compute speed and energy efficiency","url":"https://www.prnewswire.com/news-releases/collaboration-pairs-julia-labs-advanced-algorithms-with-neurophos-photonic-ai-chips-to-improve-compute-speed-and-energy-efficiency-302914854.html"},"other_sources":[],"tags":["Partnerships","Neurophos"],"body_markdown":"The Julia Lab, a research group within MIT's Computer Science and Artificial Intelligence Laboratory, and [Neurophos](https://akta.pro/company-profiles/00008li/neurophos), a company building photonic AI processors, announced a collaboration to run Julia Lab's numerical methods and algorithms on Neurophos' Optical Processing Unit, a chip containing millions of optical processing elements. [PR Newswire](https://www.prnewswire.com/news-releases/collaboration-pairs-julia-labs-advanced-algorithms-with-neurophos-photonic-ai-chips-to-improve-compute-speed-and-energy-efficiency-302914854.html) carried the announcement on October 21, 2026.\n\n## How the Julia Lab Work Fits Neurophos Chips\n\nNeurophos' OPU uses light to carry out core matrix operations, while GPUs use electricity to perform AI computations. Neurophos states a goal of cutting the resources used during AI compute by roughly 50 times with its first chip, and the company says the resource cost of the computation, in both energy and space, could decrease by over an order of magnitude while running faster than standard processors. Neurophos plans to release its systems commercially in 2028.\n\nAlan Edelman, an MIT professor and CSAIL principal investigator, said, \"I've spent my career thinking about matrices, and it's thrilling to see light turn out to be so good at multiplying them. Matrix multiplication is the heartbeat of scientific computing and AI.\" Edelman also said the partnership grew out of the lab, with Jake Chuharski now leading Neurophos' work on making algorithms fit the hardware.\n\nChuharski, a Julia Lab researcher leading Neurophos' quantization and optimization team, said, \"We're teaming up with the Julia Lab because meaningful leaps in computing rely as much on numerical methods and efficient algorithms as the state-of-the-art hardware itself.\" Patrick Bowen, Neurophos' CEO and co-founder, said, \"We're working to make computing with light over an order of magnitude faster and more energy efficient than today's best digital accelerators.\"\n\nThe Julia Lab was built on the Julia programming language created by Jeff Bezanson, Stefan Karpinski, Viral Shah and Edelman. Neurophos' OPU is designed around a systolic array, an architecture specialized for the matrix calculations at the foundation of AI inference, and the company's active metasurface technology enables optical modulators roughly 10,000 times smaller than conventional Mach-Zehnder devices, allowing millions of light-controlling elements to fit within a 5-by-5 millimeter area.\n\n## Neurophos Funding and Optical Chip Plans\n\nNeurophos is a pre-revenue semiconductor company founded in 2020 and headquartered in Austin, Texas, according to akta.pro [company records](https://akta.pro/company-data). The same records list total funding of $128.1 million, including a $110 million Series A led by Gates Frontier announced on January 22, 2026, and a valuation estimate of $500 million to $1 billion. The records also list a February 2026 partnership with Terakraft to pilot Neurophos' optical processing units within Terakraft's green AI data centre in Sauda, Norway, with pilot deployments planned for 2027.\n\n## Neurophos Team Growth Before the Julia Lab Deal\n\nNeurophos announced a team expansion and a new Silicon Valley office on September 10, 2026, after its $110 million Series A, scaling its workforce four times that year and planning another doubling by year-end, PR Newswire reported. The company also said it planned a Low-Rate Initial Production Facility in Austin later that year. In June 2026, Jonpeddie published an interview with CEO Patrick Bowen describing Neurophos as building optical AI accelerator chips using a metamaterial-based optical tensor core for matrix-multiplication-heavy AI inference workloads.\n\n## Commercial Systems Planned for 2028\n\nNeurophos plans to release its systems commercially in 2028, and the Terakraft pilot deployments are planned for 2027. More information about the company is available on [Neurophos' website](https://www.neurophos.com).\n","publisher":{"name":"News for Agent","url":"https://newsforagent.com"}}