Xcelsa Labs Launches: Using LLMs and Formal Verification to Accelerate Chip Design
IgorCarron · x · 2026-07-29
Startup Xcelsa Labs has officially launched, aiming to use AI systems to solve the compute bottlenecks and talent shortages in modern chip design.
Their core approach combines Large Language Models with formal verification:
- Propose: LLMs are used to generate and rewrite chip design source code, restructuring datapaths and pipelines to find superior alternatives.
- Prove: Instead of relying on the model's judgment, every candidate design is strictly verified by a formal equivalence engine to guarantee logical correctness.
- Measure: Surviving designs are scored on power, performance, and area (PPA) using trusted industrial synthesis tools, ensuring the results are tapeout-ready.
Founded by three NYU professors, the company is backed by leading VCs and semiconductor industry veterans.
More from Infra
- 26B-Parameter Gemma 4 Runs on Mac with 2GB RAM via SSD Streaming — petrusenko_max · 2026-07-30
- Cloudflare Containers Introduces .exec(): Stream Request Body Directly to Container Process — craigsdennis · 2026-07-30
- TSMC Roadmap: N3 Volume Production Slated for 2027, N2 Expansion Detailed — Beth_Kindig · 2026-07-30
- NYT: A Deluge of AI Computing Power Is About to Come Online, Fueling Major Leaps — coolbern · 2026-07-30
- Micron Trades at Forward P/E of ~5: Analyst Bets on AI Compute Demand — JOBhakdi · 2026-07-30
- Gavin Baker says compute rental prices may keep rising as AI demand stays hot — GavinSBaker · 2026-07-30