Applied Compute raised $350 million at a $3.25 billion valuation as enterprise customers use its cloud to both customize and run AI models. The startup said in its November 2 announcement that 40% of customers now use both services.
Applied Compute helps company engineering teams turn open-weight AI models into systems tailored to their own data and tasks. Its AC2 cloud links model training with deployment and uses feedback from real workloads for subsequent training.
Elad Gil led the round, joined by investors including Nvidia, Kleiner Perkins, Shopify Ventures and HubSpot Ventures. Applied Compute also reported a $100 million annualized revenue run rate less than 18 months after its founding.
The new financing is more than four times the $80 million round the company announced on April 8, which Kleiner Perkins led. Applied Compute plans to use the latest capital for more computing capacity, hiring and enterprise deployments.
For Clay, which sells sales and marketing software, Applied Compute built a router that chooses which AI model handles each research request. Previously, users picked a model and ran every request through it. In an October 27 case study, Applied Compute said a live A/B test cut costs by 47% against sending every request to the same frontier model, while maintaining performance on its quality safeguards.