Labelbox raises over $100m Series E after turning profitable

Labelbox raises over $100m Series E after turning profitable

14 October 2026: Labelbox, the San Francisco company that started out selling data labelling software and now builds training environments for AI labs, says it has raised more than $100 million in Series E financing, made up of equity and a credit facility, at a $1.25 billion pre-money valuation. Infinitum Partners led, with Wipro Ventures, Andreessen Horowitz, B Capital, Tau Ventures and Snowpoint Ventures among the other investors.

Not yet independently verified. The total combines equity and a credit facility, and Labelbox does not say how much is each. The customer, lab and growth figures are the company's own. We will update this when it can be confirmed, and remove this note.

The company announced the round on 14 October in a post by its chief executive, Manu Sharma, and a release the same day. It says it became profitable in 2026 while growing more than fivefold year on year.

From labelling tools to training environments

Sharma describes three chapters. The first was software for capturing expert judgment as training data, which Sharma started with two co-founders in 2018. The second is Horizon, which gives frontier AI labs reinforcement learning environments: simulated multi-step work in coding, research and other tasks where a model attempts a job, a reward says whether it succeeded, and the model improves. Labelbox says it works with 90% of the leading frontier AI labs and hyperscalers.

The third is Recursion Managed Agents, opened to all enterprise customers earlier in October. Agents start on their own from schedules, alerts and tickets, work inside a company's existing applications, are scored by a separate grader against the team's standard, and carry what they learn into the next run. Labelbox says more than 50 such agents already work across its own engineering, research, sales and finance teams.

Who is paying for compute

The company says it runs its own agent harness, sandbox infrastructure running millions of virtual machines, and post-training systems, plus an expert network of more than 4 million people. Sharma names compute as central to the next stage and points to Infinitum's John Yetimoglu, an early investor in Groq who now works on GroqCloud's compute infrastructure.

"Every company will soon run on AI systems that learn from their own work," Sharma said in the release.

Customers and what changes for them

Labelbox says it serves more than 400 customers, from AI labs and hyperscalers to start-ups and Fortune 500 companies. The money goes to research and engineering on Horizon and Recursion and to the teams that sell and support them.

Our machine learning software category includes the labelling tools Labellerr and V7 Darwin, which compete with the work Labelbox started with.

Company profile on TrustList: Labelbox

Sources

• Labelbox: Labelbox raises $100M+ Series E, 14 October 2026
• Labelbox (via PR Newswire): Labelbox raises $100 Million+ in Series E funding to advance reinforcement learning and enterprise AI agents, 14 October 2026