AI Chip Startup MatX Raises $500M to Challenge Nvidia's GPU Monopoly
March 3, 2026 · 2 min read
Claire Cummings
MatX, the chip startup founded by former Google TPU engineers, has closed a $500 million Series B led by Jane Street and Situational Awareness, the investment fund formed by former OpenAI researcher Leopold Aschenbrenner. Additional backers include Marvell Technology, Spark Capital, and Stripe co-founders Patrick and John Collison.
The raise follows a $100 million Series A in 2024 and represents one of the largest chip startup fundings in history. Co-founder Reiner Pope, who led AI software development for Google's TPUs, and Mike Gunter, a lead TPU hardware designer, left Google in 2023 to build what they claim is a fundamentally better architecture for running large language models.
The Chip That Could Break the GPU Bottleneck
MatX One uses a "splittable systolic array" architecture that combines SRAM for storing model weights with High Bandwidth Memory for KV cache data. The design integrates speculative decoding, blockwise sparse attention, and custom numerics to deliver what Pope calls "higher throughput on LLMs than any announced system, while simultaneously matching the latency of SRAM-first designs."
The company plans to complete tape-out within one year and begin shipping chips in 2027, manufacturing with TSMC.
Why This Matters for Grant-Funded AI Research
Nvidia GPU scarcity and pricing have become a defining bottleneck for federally funded AI researchers. Academic labs competing with corporate AI operations for H100 and B200 allocations routinely wait months for hardware delivery, and compute costs consume growing shares of NSF, DOE, and DARPA research budgets.
A viable Nvidia alternative could meaningfully stretch research dollars. AI researchers writing grant proposals today should consider the evolving hardware landscape — budget justifications that account for emerging non-Nvidia accelerators could strengthen proposals by demonstrating cost awareness and flexibility.
The broader trend is unmistakable: private capital is flooding AI infrastructure at a pace that dwarfs federal research budgets. OpenAI's $110 billion round and MatX's $500 million raise in the same month underscore how the compute landscape is shifting beneath researchers' feet.
For ongoing coverage of AI research funding and infrastructure, visit grantedai.com.