Observed Signal · Apr 9, 2026 · Technical Release · Source: Chipstrat · Impact: 3/5 · Sentiment: Positive
MatX CEO Reiner Pope on LLM Chip Design
Reiner Pope, co-founder and CEO of MatX, describes the startup's effort to build LLM-optimized datacenter chips and systems. Pope and co-founder Mike Gunter left Google one week before ChatGPT's launch to pursue a hardware architecture they say is optimal for large language models. MatX has raised about $600 million to date and is building the MatX One system featuring a hybrid SRAM-HBM memory architecture, high matrix-multiply throughput, extensive interconnect, and a software stack for frontier labs. The company targets large model training, prefill and decode workloads, uses AI internally to accelerate chip design, publishes select research (attention work), and cites manufacturing/production ramp as a major near-term challenge. MatX employs just over 100 people across hardware, software and ML disciplines.
MatX describes a novel LLM-focused chip and system architecture (hybrid SRAM-HBM, high matrix throughput and interconnect) that could affect compute economics for frontier labs and increase competition to incumbent GPU vendors, but the company is a startup facing manufacturing and scale-up challenges.
Track Google Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- MatX has raised approximately $600 million to date.
- Co-founders Reiner Pope and Mike Gunter left Google one week before ChatGPT launched.
- MatX One uses a hybrid SRAM-HBM memory architecture combining weights-in-SRAM with HBM throughput.
- Target customers are frontier AI labs focused on high compute spend and long-term co-design.
- MatX has a cross-disciplinary team of just over 100 employees and is building both chip and datacenter system-level designs.
Connected Companies & Entities
9 Entities mappedOntology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
MatX Raises $500M to Challenge Nvidia's AI Dominance
MatX, an AI chip startup founded in 2023 by former Google TPU engineers Reiner Pope and Mike Gunter, raised $500 million in a Series B round led by Jane Street and Situational Awareness (an investment fund formed by former OpenAI researcher Leopold Aschenbrenner). Other investors include Marvell Technology, NFDG, Spark Capital, and Stripe co‑founders Patrick and John Collison. MatX says its processors aim to be roughly 10x better than Nvidia GPUs for training and serving large language models. The company previously raised about $100 million in a 2024 Series A. MatX plans to manufacture its chips with TSMC and start shipping in 2027.
Microsoft‑backed D‑Matrix Begins Shipping Corsair AI Chip
D‑Matrix, a Microsoft‑backed startup founded in 2019, has begun shipping its Corsair AI inference chip in June 2026. The company says Corsair — an SRAM‑centric, tightly integrated memory-and-compute chip manufactured on TSMC's 6nm node — can run small-workload inference up to 10x faster and use up to 5x less energy than a standalone GPU when paired with an Nvidia Blackwell GPU, per cited Gimlet Labs research. D‑Matrix has raised about $500 million (≈ $2 billion valuation), will ship primarily to U.S. hyperscalers and frontier AI labs, and is partnering with Arista, Broadcom and Super Micro to deliver a rack-scale system called SquadRack. The startup plans a next-generation Raptor chip on TSMC 4nm next year.
Ricursive Intelligence Secures $335M, Aims to Revolutionize Chip Design
Ricursive Intelligence, founded by Anna Goldie (CEO) and Azalia Mirhoseini (CTO), raised a combined $335 million within months of launching: a $35 million seed round led by Sequoia followed by a $300 million Series A at a $4 billion valuation led by Lightspeed. The founders, former Google Brain and Anthropic engineers, previously built the Alpha Chip — an AI tool that produced high-quality chip layouts in hours and helped design multiple generations of Google’s Tensor Processing Units. Ricursive is developing AI-driven software that automates chip design (component placement through verification) and leverages learning across designs and LLMs. The company positions chipmakers such as Nvidia, AMD and Intel as customers (Nvidia is also an investor). The founders say their platform aims to accelerate chip development, improve hardware efficiency, and enable tighter co-evolution of models and chips.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.