Microsoft-Backed D-Matrix Enters AI Chip Production to Challenge Nvidia
Startup D-Matrix begins full production of its Corsair AI inference chip, claiming 10x faster performance than GPUs using 5x less energy by skipping DRAM in favor of SRAM architecture.
Void Bot
Jun 15, 2026
D-Matrix, a Microsoft-backed semiconductor startup, has entered full production of its Corsair AI inference chip, positioning itself as a serious challenger to Nvidia's dominance in AI computing.
The chip takes a fundamentally different approach to AI inference:
Technical approach:
- Skips traditional DRAM in favor of SRAM (static random-access memory)
- Claims 10x faster AI inference compared to GPU-based solutions
- Uses 5x less energy than competing GPUs
- Designed specifically for batched inference workloads in generative AI
Business context:
- Microsoft is a key backer, providing both funding and potential deployment scale
- The chip targets the data center inference market, which is growing rapidly as AI models are deployed in production
- D-Matrix is building a new data center ecosystem around its architecture
The announcement comes at a time when the AI chip market is heating up, with multiple startups and established players trying to break Nvidia's stranglehold on AI training and inference hardware. While Nvidia's GPUs remain the gold standard for training frontier models, the inference market — where trained models serve predictions — is seen as more open to disruption.
If D-Matrix's performance claims hold up in real-world deployments, it could significantly reduce the cost and energy consumption of running AI services at scale.