Ex-OpenAI startup Applied Compute hits $100M ARR in 18 months, raises $350M at $3.25B valuation
ypatil125 · x · 2026-11-04
Applied Compute, founded by three ex-OpenAI researchers (Yash Patil, Rhythm Garg, Linden Li, aged 23/24/25), announced:
- $100M annualized revenue run rate, from zero 18 months ago
- $350M raise at a $3.25B valuation led by Elad Gil, with NVIDIA, Blackstone, Kleiner Perkins, and Lux participating
- Inference platform serving over 1T tokens per day just 12 weeks after launch
- 40% of customers use both custom model training and production inference; enterprise customers doubled since the last raise; team tripled since January, 85% researchers/engineers
- Customers include Microsoft, Shopify, Harvey, DoorDash, Cognition, Clay, Decagon
It builds custom models on open-weight models, claiming up to 10x cheaper than frontier models; a custom model for Harvey improved answer and citation quality at under half the inference cost.
More from Venture
- Dylan Patel: CoreWeave sold out for 6 months, Nebius demands full-year prepay — AccBalanced · 2026-11-04
- SemiAnalysis's Dylan Patel: $11T AI capex by 2029, $7T of it debt; CoreWeave sold out — AccBalanced · 2026-11-04
- GPU rental prices jumped from $3 to $13/hour in six months; 2027 winners will match contract duration to compute — AccBalanced · 2026-11-04
- Ex-OpenAI's realchillben confirms he has joined a16z as an investor — venturetwins · 2026-11-04
- Meta raises $15B in bonds for AI buildout, drawing $58B in orders at 3.9x oversubscription — rohanpaul_ai · 2026-11-04
- Meta raises $15B in bonds for AI build-out, drawing $58B in orders — rohanpaul_ai · 2026-11-04