Inference-optimization startup Wafer raised a $40M Series A co-led by Marathon and Chemistry, with AMD Ventures, Y Combinator, Vercel’s CEO and others participating. It pitches AI that optimizes AI—continuously learning traffic and constraints to pick better deployments across model, engine, kernels, and hardware—turning one-off manual tuning into an ongoing perf-per-dollar loop, after reportedly growing ARR from near-zero to about $8M in four months.
Key Takeaways
- ✓$40M Series A co-led by Marathon and Chemistry; AMD Ventures, YC, and notable angels join.
- ✓Agents continuously optimize open-model inference across model, engine, kernels, and hardware for perf per dollar.
- ✓Claimed near-$0 to ~$8M ARR in four months, with demand outrunning GPU capacity.