TypeSafe AI Raises $870M Series C+ Led by Andreessen Horowitz

TypeSafe AI, a San Francisco frontier AI lab, has raised $870 million in a Series C+ round led by Andreessen Horowitz. Sequoia Capital also participated in the round, which was announced on October 9, 2026.

What TypeSafe AI Does

TypeSafe AI is building machine-native, composable AI: models designed to work as primitives inside software rather than as conversational assistants for people. The company's stated goal is to make AI something software can call directly, the way it would call any other component.

Its first model, Jev, is a System One model trained with Reinforcement Learning for Calibrated Decisions (RLCD). Jev returns typed decisions with calibrated confidence scores at under 100 milliseconds of latency. That combination — typed outputs, calibrated confidence, and sub-100ms response times — is intended to let software automate economically valuable work while escalating only low-confidence cases to a human or a more capable system.

Founding Team

TypeSafe AI was founded by Diogo Almeida, an ex-OpenAI researcher and co-inventor of RLHF, together with Erik Gafni and Sasha Sheng. The company emerged from stealth in September 2026 with a $40 million seed round led by DCVC.

Funding History

Seed: $40 million, led by DCVC, announced September 2026
Series C+: $870 million, led by Andreessen Horowitz, with Sequoia Capital participating, announced October 9, 2026

The rapid progression from a $40 million seed round to an $870 million Series C+ within roughly a month reflects the scale of capital currently flowing toward frontier AI labs, particularly those with credible technical founders and a differentiated architectural thesis.

Why the Round Matters

The size of the Series C+ places TypeSafe AI among the most heavily capitalized early-stage AI companies. The round's lead, Andreessen Horowitz, is one of the most active investors in frontier AI, and Sequoia Capital's participation adds a second top-tier fund to the cap table.

For acquirers and strategic investors, TypeSafe AI's positioning is notable for what it is not. The company is not building a consumer assistant or a general-purpose chatbot. It is building models intended to be embedded as primitives inside other software — a distribution model that could make the company a supplier to, rather than a competitor of, large parts of the software industry.

The Technical Bet

The RLCD training approach and the emphasis on calibrated confidence are the core of the company's differentiation. Calibrated confidence matters because it determines when software can act autonomously and when it must escalate. A model that is both fast and honest about its own uncertainty is more useful as an automation primitive than one that is merely accurate on average.

Latency is the second pillar. At under 100 milliseconds, Jev's decisions can sit inside interactive software workflows without degrading the user experience — a threshold that rules out many larger, slower models.

What to Watch

Commercial adoption of Jev as an embedded primitive in third-party software
Whether the company expands beyond its first model into a broader family of composable primitives
How the $870 million is deployed relative to the $40 million seed raised only weeks earlier
Competitive response from other labs targeting machine-native rather than human-facing AI

Summary

TypeSafe AI has raised $870 million in a Series C+ round led by Andreessen Horowitz, with Sequoia Capital participating. The San Francisco lab, founded by ex-OpenAI researcher and RLHF co-inventor Diogo Almeida along with Erik Gafni and Sasha Sheng, builds machine-native AI models designed to function as software primitives. Its first model, Jev, returns typed decisions with calibrated confidence scores at under 100 milliseconds of latency.