Together AI tops $8 billion valuation as open-source AI demand surges

Together AI has crossed into the $8 billion tier after a 2026 Series C that valued the San Francisco company at about $8.5 billion, putting one of the clearest bets on open-source AI at the center of the market’s cost fight. Founded in June 2022 by Vipul Ved Prakash, Ce Zhang, Chris Re and Percy Liang, the company has built its business around running, fine-tuning and training open-source foundation models at high performance and lower cost.

The financing follows a rapid climb. Together AI raised a $102.5 million Series A in November 2023, then a $305 million Series B in February 2025 at a $3.3 billion valuation. By 2026, reports put its latest round at about $1 billion, with investors including Kleiner Perkins, General Catalyst, Prosperity7, NVIDIA, Lux Capital, Emergence Capital and Salesforce Ventures. The company’s public platform now spans inference, fine-tuning and GPU clusters, and it says more than 200 models are available through a unified API.

Revenue estimates show both momentum and the limits of private-company disclosure. Market trackers put Together AI’s annualized revenue near $1 billion by early 2026, while CB Insights pegged 2025 revenue at $100 million and listed 2024 as its most recent revenue data point at $130 million. That spread reflects how quickly the company has scaled and how little outside investors can verify until a public filing or sale forces the numbers into the open.

The bigger story is the market shift behind the valuation. McKinsey said in April 2025 that more organizations deploying generative AI were turning to open-source AI solutions, while CISA warned in July 2024 that open-source AI can accelerate innovation but also create security risks. Stanford HAI’s 2026 AI Index added another layer of pressure, saying AI capability is not plateauing and that the U.S.-China frontier-model gap has effectively closed.

That backdrop helps explain why Together AI matters beyond one funding round. NVIDIA said in February 2026 that Together AI was among inference providers cutting cost per token by up to 10x on its Blackwell platform, a concrete signal that infrastructure efficiency is becoming as important as model scale. For enterprises weighing OpenAI, Anthropic, Google or Meta against open-source alternatives, the question is no longer only which model is best, but which stack can deliver usable AI at a price that survives production deployment.