Mercor, a $10 billion（約1.6兆円） AI startup, acquired Deeptune to combine simulation training environments with its existing network of domain experts who build reinforcement learning tasks. Frontier labs like Anthropic and OpenAI are now customers who need full digital replicas of enterprise software where their AI agents can practice before production deployment. The deal reflects a maturing AI training market where both simulation environments and task-building infrastructure are essential.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Mercor, a $10 billion（約1.6兆円） AI startup, acquired Deeptune, an Andreessen Horowitz-backed company that builds simulation environments where AI agents practice real-world tasks. Founder Brendan Foody had quietly invested in Deeptune's $43 million（約69億円） Series A funding round three months before the acquisition closed.
Why it matters
Frontier AI labs like Anthropic and OpenAI now require digital replicas of enterprise software to train their AI agents before deploying them to production systems. The acquisition combines Mercor's network of over five million domain experts (who build tasks and scoring rubrics) with Deeptune's simulation environments, creating a complete stack for reinforcement learning. Mercor claims all of the Mag Seven except Tesla are customers.
What to watch
Mercor hit $2 billion（約3200億円） in ARR in June, up from $1 billion（約1600億円） last year—a doubling in just four months. The company previously experienced a data breach in March affecting an estimated four terabytes, but Foody stated that every frontier lab has expanded their relationship with Mercor since the breach.
Mercor's acquisition of Deeptune marks a strategic consolidation in the AI training infrastructure market at a moment when frontier labs face a new challenge: their AI agents need to practice on realistic enterprise software before being deployed in production environments. The body describes this stack as requiring both simulation environments (Deeptune's contribution) and the task-building expertise Mercor already provides through its network of domain experts. Foody's advance investment three months before the acquisition suggests the company had already identified this gap as a growth opportunity.
The timing and scale of Mercor's growth underscore the urgency of this consolidation. The company doubled revenue from $1 billion（約1600億円） to $2 billion（約3200億円） in ARR in just four months (hitting $2 billion（約3200億円） in June), and Foody claims this represents the fastest growth trajectory from $1 million（約1.6億円） to $2 billion（約3200億円） in 24 months. This acceleration occurred despite a significant data breach in March in which hackers exploited a vulnerability in LiteLLM and exfiltrated an estimated four terabytes of Mercor's data. That Foody reports every frontier lab expanded their relationship with Mercor after the breach suggests either remarkable customer loyalty or, as the body notes, that frontier labs have limited alternatives at Mercor's scale.
No discussion yet for this article
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
