Neurophos is tackling one of AI’s biggest bottlenecks: power and scalability. As AI inference workloads grow, data centres are hitting hard limits on energy use and operating costs. Traditional silicon GPUs are struggling to keep up, prompting companies to seek new approaches to compute.
The company has raised $110 million in a Series A round as it works to bring photonic computing into the heart of AI data centres. The round was oversubscribed, bringing its total funding to $118 million.
The investment was led by Gates Frontier, with participation from M12, Carbon Direct Capital, Aramco Ventures, Bosch Ventures, Tectonic Ventures, and Space Capital.
Additional investors include DNX Ventures, Geometry, Alumni Ventures, Wonderstone Ventures, MetaVC Partners, Morgan Creek Capital, Silicon Catalyst Ventures, Mana Ventures, Gaingels, and others. Cooley LLP serves as legal counsel.
Developing an optical processing unit
Founded by Dr Patrick Bowen and Dr Andrew Traverso, the team includes industry veterans from NVIDIA, Apple, Samsung, Intel, AMD, Meta, ARM, Micron, Mellanox, Lightmatter, and more. Neurophos is developing an optical processing unit (OPU) that uses light rather than electricity to perform AI calculations.
“Moore’s Law is slowing, but AI can’t afford to wait. Our breakthrough in photonics unlocks an entirely new dimension of scaling by packing massive optical parallelism on a single chip. This physics-level shift means both efficiency and raw speed improve as we scale up, breaking free from the power walls that constrain traditional GPUs,” says Bowen.
The chip integrates more than one million microscopic optical processing elements on a single device. According to the company, this can deliver up to 100 times better performance and energy efficiency than today’s leading chips, while fitting into existing data centre setups as a drop-in alternative to GPUs.
Neurophos has made a significant breakthrough by developing tiny optical modulators that are 10,000 times smaller than older photonic devices.
This enables large-scale photonic computing for the first time. As a result, we now have a new type of AI accelerator that is very fast, energy-efficient, and can easily adapt to future AI needs.
What’s next?
With the new funding, Neurophos plans to accelerate work on its first integrated photonic compute system. It includes data centre-ready OPU modules, a full software stack, and early-access hardware for developers.
Additionally, the company is expanding its headquarters in Austin and opening a new engineering site in San Francisco to support early customer demand.
“As the AI industry grapples with a surge in demand that tests our ability to satisfy with compute and power, disruptive approaches to compute may open routes to sustained or accelerated systems scaling that will be needed before the end of the decade. With their approach to hyper-efficient optical computation, the Neurophos team have advanced swiftly from a working proof of concept towards a realistic plan to deliver products on a timeline we can underwrite and believe in,” said Michael Stewart, Managing Partner at M12, Microsoft’s Venture Fund.
“From the start, we backed Neurophos because we believed the future of AI was bound by physics, not by algorithms,” said Chris Alliegro, Managing Partner at MetaVC Partners. “Neurophos is addressing the only problem that really matters for the future of AI: the limits imposed by silicon. Their optical architecture provides the foundation for the next generation of machine intelligence.”
