Neurophos Raises $110 Million Series A to Accelerate Photonic AI Data Center Chips

New updates have been reported about Neurophos.

Neurophos has raised $110 million in an oversubscribed Series A round, bringing total funding to $118 million and positioning the Austin-based company as a potential disruptor in data center AI acceleration. The round was led by Gates Frontier with participation from Microsoft’s M12, Carbon Direct Capital, Aramco Ventures, Bosch Ventures, Tectonic Ventures, Space Capital, and others, signaling strong strategic and financial backing for Neurophos’ photonic AI chip roadmap. Neurophos is developing an optical processing unit (OPU) that integrates more than one million micron-scale optical processing elements on a single chip, targeting up to 100x performance and energy-efficiency gains over leading GPUs while remaining a practical, drop-in replacement for existing data center infrastructure. Microsoft executive Dr. Marc Tremblay framed the opportunity as a needed “breakthrough in compute” to match recent advances in AI models, underscoring enterprise demand for high-performance, lower-power inference solutions.

The company’s core innovation is a new class of micron-scale metamaterial optical modulators—miniaturized by a factor of 10,000 versus prior photonic elements—enabling dense, manufacturable photonic compute at scale and creating an AI accelerator architecture designed for both speed and power efficiency. CEO and co-founder Dr. Patrick Bowen describes this as a physics-level shift that allows performance and efficiency to improve as systems scale, addressing the power and cost constraints that increasingly limit GPU-based data centers. The fresh capital will be used to deliver Neurophos’ first integrated photonic compute systems, including datacenter-ready OPU modules, a complete software stack, and early-access developer hardware, while expanding its Austin headquarters and establishing a San Francisco engineering office to support early customers. Investors such as M12’s Michael Stewart highlighted that Neurophos has progressed from proof of concept to a product timeline that institutions are willing to underwrite, suggesting a path toward commercial deployment before the end of the decade as AI infrastructure providers seek more sustainable compute and power profiles.