AI designs chip from spec to hardware in 2 weeks
rohanpaul_ai · x · 2026-08-30
A new paper demonstrates an AI system converting chip specifications into working hardware, marking a shift in chip design speed.
- Case Study (Redwood): A small, low-power inference chip. 2 human architects wrote the spec; AI generated the design, tests, firmware, and kernels.
- Efficiency: Spec-to-hardware loop took under 48 hours. The first FPGA design had 0 bugs with 95% test coverage.
- Performance: Runs Qwen3-0.6B at 12.1 t/s on AMD FPGA. Projected 49 t/s @ 1.335W on Samsung 8nm, offering 3.4x better tokens-per-watt than NVIDIA Jetson Orin Nano.
Hardware development could soon match the pace of model releases.
Related event: Two Engineers Plus AI Deliver a Working Chip in Two Weeks(2 posts)→
More from Infra
- Grok claims 816B MoE model runs locally on dual M5 Ultra Mac Studios via MLX — grok · 2026-10-16
- Same 1,000 Cloudflare Durable Objects: $4,050/month cut to $22 with one trick — threepointone · 2026-10-16
- NVIDIA ships FP4-quantized Qwen3.8-Flash for local use, trending on Hugging Face — nvidia · 2026-10-16
- Baidu's Virtual Head Attention tames the KV cache, FLOPs, quality triangle in long-context LLMs — baidu · 2026-10-16
- DuoDiscover agents auto-build fast eval proxies, scoring 2.68x TPU kernel speedup — berkeley_ai · 2026-10-16
- Google Raises 2026 Capital Spending to Up to $205 Billion — emmanuelvivier · 2026-10-16