Viqus Logo Viqus Logo
Home
Categories
Language Models Generative Imagery Hardware & Chips Business & Funding Ethics & Society Science & Robotics
Resources
AI Glossary Academy CLI Tool Labs
About Contact

Photonic Chip Startup Neurophos Raises $110M to Challenge Nvidia in AI Inference

AI Photonics Metamaterials Hardware Neurophos Funding Tech Startup
January 22, 2026
Viqus Verdict Logo Viqus Verdict Logo 7
Strategic Shift
Media Hype 6/10
Real Impact 7/10

Article Summary

Austin-based Neurophos is developing optical processing units (OPUs) using metasurface modulators, a technology initially explored for invisibility cloaks. The company’s approach seeks to address the escalating power consumption challenges associated with scaling AI workloads, particularly in hyperscalers' data centers. By fitting thousands of these modulators onto a single chip, Neurophos claims its OPUs can achieve significantly faster performance than Nvidia’s B200 GPU, while consuming substantially less power—675 watts versus the B200’s 1,000 watts—and delivering 235 Peta Operations per Second. The $110 million Series A funding round, led by Bill Gates’ venture firm Gates Frontier, underscores the potential of this technology. While photonic chips have long been theorized as offering advantages over silicon, challenges such as larger component sizes and conversion complexities have hindered widespread adoption. Neurophos’s claim that its metasurface addresses these issues in a single solution is key to their strategy. Microsoft’s corporate vice president, Marc Tremblay, has highlighted the need for a revolutionary compute solution in AI, suggesting that Neurophos's technology aligns with this demand. The startup anticipates launching its first chips by mid-2028, and is already attracting interest from major players like Microsoft. Despite facing competition from Nvidia, the company aims to leverage its unique technological approach and a strong team to establish itself as a viable alternative.

Key Points

  • Neurophos is developing optical processing units (OPUs) using metasurface modulators to address the energy efficiency challenges of AI inference.
  • The company has secured $110 million in Series A funding led by Bill Gates’ venture firm, demonstrating investor confidence in the technology.
  • Neurophos’s OPUs aim to outperform Nvidia’s B200 GPU in terms of speed, efficiency, and power consumption, potentially disrupting the AI hardware market.

Why It Matters

This news is significant because it represents a potential challenge to Nvidia’s dominance in the AI hardware market. The escalating power consumption of AI models is a major bottleneck, and if Neurophos’s optical processing units can deliver on their performance promises, it could dramatically reduce energy costs and accelerate the deployment of AI across various industries. This development is important for anyone involved in AI development, data center operations, or investment in emerging technologies. It represents a shift in thinking about hardware acceleration and a renewed focus on energy efficiency, a crucial consideration as AI continues to grow in scale and complexity.

You might also be interested in