TypeSafe AI raises $870 million at a $7.5 billion valuation weeks after Jev's launch

TLDR

• TypeSafe AI has raised $870 million at a $7.5 billion valuation, led by Andreessen Horowitz with Sequoia Capital and existing backer DCVC joining.¹
• The round lands less than a month after the September 15 launch of Jev, a model that returns a structured value (a yes/no, a pick from a list, or a score) instead of conversational text.²
• a16z says Jev has already been integrated by roughly a quarter of the Fortune 500 and generated about 1 trillion tokens in its first three days, calling the launch "the biggest narrative violation we've seen this year."²
• TypeSafe was founded in 2024 by former OpenAI researcher Diogo Almeida with Erik Gafni and Sasha Sheng, and had raised only a $40 million DCVC-led seed round before Jev's launch.³

TypeSafe AI, the San Francisco startup behind the decision model Jev, has raised $870 million in a round led by Andreessen Horowitz, with Sequoia Capital and existing investor DCVC also participating, valuing the company at $7.5 billion.¹ TechCrunch reports the raise comes less than a month after Jev's launch, with response times under 700 milliseconds and running, by TypeSafe's own account, up to 200 times faster and 100 times cheaper than some frontier language models.¹ Jev deliberately handles only three kinds of request: a yes/no answer, a choice from a list, or a customizable score, each returned with a confidence number apps can use to catch hallucinated outputs, rather than generating open-ended text.¹

Andreessen Horowitz frames the investment around a thesis it calls "System One models," AI built to plug directly into code rather than to converse with people, positioned as complementary to rather than competing with the large language models that dominate the rest of the firm's AI bets.² The firm points to adoption numbers as the reason for moving fast: about 25% of Fortune 500 companies integrated Jev within its first week, and roughly 1 trillion tokens were generated in the first three days post-launch, a pattern a16z dubbed "the Jev-ons paradox," where cheap, fast intelligence calls drive explosive usage growth once they're easy enough to wire into any codebase.² a16z also said it has "long been fans" of chief executive Diogo Almeida, framing the bet partly as a continuation of an existing relationship with him.²

Almeida spent around four years at OpenAI working on RLHF, InstructGPT, ChatGPT and GPT-4 before leaving in 2024 to found TypeSafe with Erik Gafni and Sasha Sheng, a former Meta research engineer.³ The company reportedly built Jev in stealth for roughly two years, surfacing publicly only on September 15 with Jev's launch and a $40 million seed round led by DCVC, the same firm that stayed on as an investor in this round.³ Jev is the first model in a planned family TypeSafe calls System One, and the company says the new capital will fund expanding that lineup and adding enterprise features.¹

Why it matters: a16z betting hundreds of millions on a model that deliberately does less than a chatbot, on the strength of adoption data alone, signals growing investor appetite for narrow, fast "decision" models as a distinct category from general-purpose LLMs rather than a smaller version of the same race.