Applied Compute
Enterprise AI model training and deployment platform
CORE INFO
Applied Compute sells infrastructure and engineering support to enterprises that want to build custom AI models and agents around their own data and workflows. Its platform connects model training, production deployment, and ongoing improvement while supporting deployment inside customers' security environments.
- Latest financing: $350 million led by Elad Gil at a $3.25 billion valuation, announced November 2, 2026.forbes.comsiliconangle.com
- Annualized revenue run rate: approximately $100 million, reported in connection with the November 2026 financing.siliconangle.com
- Forbes profiled the company's former OpenAI founders, enterprise customers, open-weight model strategy, and use of the new financing.forbes.com
WHY WE WOULD WORK AT APPLIED COMPUTE
Build Specific Intelligence
Help enterprises turn their own data, workflows, and feedback into AI they truly own. Applied Compute's mission is to close the gap between "smart and useful" with continuously improving, company-specific agents.
Work at the Frontier
Build AC2, a unified platform for post-training, reinforcement learning, production inference, model routing, and continuous learning across open-weight models.
Join a Dense Team
Around 85% of the approximately 70-person team works in research or engineering, and many employees are former founders. The culture emphasizes end-to-end ownership, customer empathy, and moving research into production.
Grow With Real Momentum
Applied Compute has more than doubled its customer base since its previous fundraise, surpassed 1,700 AC2 training runs since July 2026, and reached a reported $100M annualized revenue run rate.
Solve High-Stakes Problems
Work directly with organizations including Microsoft, NVIDIA, Harvey, DoorDash, Mercor, Clay, and Decagon across legal, cybersecurity, life sciences, logistics, and enterprise software.
Build With Serious Backing
The company raised $350M at a reported $3.25B valuation, bringing disclosed cumulative funding to approximately $510M. That backing supports ambitious work on enterprise AI infrastructure while the company scales rapidly.
MARKET AND TRACTION
MARKET POSITION
✦ KEY METRICNOTABLE CUSTOMERS
KEY METRICS
✦ KEY METRICCOMPETITIVE ADVANTAGE
GROWTH TACTICS
PRODUCT AND TECH
AC2 Agent Cloud
Applied Compute's unified platform for training, serving, routing, and continuously improving open-weight AI models and agents. It connects model development with production deployment and feedback loops.
Model Post-Training
Supports reinforcement learning, preference optimization, self-distillation, and custom evaluations for tool-using and long-horizon agents. Customers can train models using their own data, workflows, graders, reward functions, and agent environments.
Inference And Routing
Provides production inference, model serving, autoscaling, observability, and checkpoint deployment for models including Qwen, Kimi, GLM, DeepSeek, MiniMax, and NVIDIA Nemotron. Model routing directs requests to the most suitable model based on task, cost, and performance.
Continuous Learning
Captures production traces such as inputs, outputs, tool calls, and token-level data to support online improvement. New model checkpoints can be evaluated and deployed rapidly as real-world usage generates additional training signals.
Embedded AI Engineering
Applied Compute researchers and engineers help enterprises design evaluations, build data pipelines and agent environments, train specialized models, and deploy them within customer security environments. Engagements can be embedded alongside customer teams or managed by Applied Compute.