Deeptune Inc. raises $43M in Series A

Deeptune Inc., an artificial intelligence startup, has secured $43 million in a Series A funding round led by Andreessen Horowitz, aimed at tackling the AI industry's pressing "data exhaustion" issue. The funding round also saw participation from 776, Abstract Ventures, Inspired Capital, and prominent angel investors, including OpenAI researcher Noam Brown and Mercor.io Corp. CEO Brendan Foody.

Based in New York, Deeptune is focused on creating "training gyms" for AI agents. These high-fidelity reinforcement learning environments simulate complex digital workspaces used by professionals like DevOps engineers, accountants, and customer support representatives. By providing AI agents with these virtual environments—akin to flight simulators—the company aims to teach systems to automate complex, multi-step tasks traditionally carried out by humans using software tools such as Salesforce and Slack. Deeptune CEO Tim Lupo emphasizes that hands-on training in realistic scenarios is crucial, as AI needs more than just static data from the internet to advance.

The funding arrives amidst concerns highlighted by industry figures, including Andreessen Horowitz founder Marc Andreessen, who note the dwindling supply of freely available internet data for training AI systems. Deeptune seeks to shift the paradigm from a labor-intensive data collection model to one centered on engineering and computational ingenuity. In these virtual gyms, AI models perform task "rollouts," earning virtual rewards for correct actions, which drives learning through trial and error at scale.

Andreessen Horowitz partner Marco Mascorro has lauded Deeptune's approach, which reportedly improves the performance of "computer-use" agents on industry benchmarks. This method has led to advanced models mirroring human proficiency in operating computers through various interfaces. According to Mascorro, platforms like Opus 4.6 and GPT-5.4 have surpassed certain human baselines on computer use benchmarks, showcasing the potential of Deeptune's training model.

Looking ahead, Deeptune aims to leverage its innovative platform to capture a substantial share of the global reinforcement learning market, projected to grow from $11.6 billion in 2025 to over $90 billion by 2034. The company claims to have developed hundreds of these simulated environments for leading AI labs globally, aspiring to set a new standard for AI model training grounded in enriched environments rather than just expanding datasets.