Combined with Neurophos, a spinoff of Duke University and Metacept, the two rounds have totaled $110 million in a Series A funding round. This successful round was led by Gates Frontier. The company, which is based in Austin, Texas, focuses solely on developing commercial applications for metamaterials. Related to that are the challenges they face at the cutting edge with scaling computing power while maintaining low power consumption. This non-dilutive funding will help accelerate the commercialization of Neurophos’ novel optical processing technology. It vows to deliver orders of magnitude better performance compared to the state-of-the-art silicon GPUs powering today’s AI data centers.
This unique funding round attracted a diverse group of investors. These were joined by Microsoft’s M12, Carbon Direct, Aramco Ventures, Bosch Ventures, and Tectonic Ventures – to name a few. This financial backing is a testament to the confidence these investors have in Neurophos’ innovative approach to optical processing.
According to Neurophos, they’ve created a metasurface that is about “10,000 times” smaller than traditional optical transistors. The company’s optical processing unit (OPU) exceeds the performance of silicon GPUs by extraordinary margins. Beyond performance on training workloads, it has a strong efficiency story for inferencing workloads.
Dr. Patrick Bowen, the leading innovator behind Neurophos and the company’s Chief Scientific Officer, discussed the technological benefits of their innovation.
“When you shrink the optical transistor, you can do way more math in the optics domain before you have to do that conversion back to the electronics domain.” – Dr. Patrick Bowen
Neurophos’ chips are such that they can be reliably manufactured with silicon foundry materials and tooling. This gives manufacturers the grit needed to seamlessly integrate into already established manufacturing frameworks. This remarkable compatibility no doubt serves the company exceedingly well as one might expect in the always tumultuous semiconductor environment.
The unannounced startup plans to open an engineering office in San Francisco, while growing its initial offices in Austin. This economic growth strategy will help them continue to attract top talent and will allow them to continue commercializing their optical processing technology.
TPII’s Dr. Bowen further emphasized the fact that today’s AI inference requires an extraordinary level of energy and compute power.
“If you want to go fast, you have to solve the energy efficiency problem first. Because if you’re going to take a chip and make it 100 times faster, it burns 100 times more power.” – Dr. Patrick Bowen
In addition to its overall development agenda, Neurophos has committed to launching its first chips in mid-2028. Since then, the company has known multiple geographic and vertical customers for its products, though again specific names are still under wraps. Microsoft is currently doing a deep dive on all of Neurophos’ products. It does indicate that the big players in the tech industry are very much committed on this front.
Dr. Bowen elaborated on the competitive advantage that Neurophos hopes to achieve once their drug hits the market.
“Even if we chart out Nvidia’s improvement in architecture over the years, by the time we come out in 2028, we still have massive advantages over everyone else in the market because we’re starting with a 50x over Blackwell in both energy efficiency and raw speed.” – Dr. Patrick Bowen
The future impact of Neurophos’ technology has the potential to change the trajectory of AI computing power. The optical processing unit’s performance is expected to increase exponentially compared to Nvidia’s B200 AI GPU, which rules over 90% of the current AI accelerator market.
