Baseten secures $1.5 billion, valuing the company at up to $13 billion for AI inference.

      Baseten is nearing the completion of a $1.5 billion funding round that will value the company at up to $13 billion. The structure of this funding is almost as significant as its size. It consists of two tiers, with some investors entering at an $11 billion valuation while others are at $13 billion, as noted by the company in the Wall Street Journal. This method is being employed by several AI startups recently to boost their headline figures. The financing round is co-led by Altimeter Capital, Conviction, Spark Capital, Sands Capital, and Wellington Management.

      Baseten focuses on a less glamorous aspect of the AI stack. Founded in 2019 and based in San Francisco, the company offers the software and computing power that businesses require to perform inference, which is the process where a trained model responds to a query. It leases capacity from around 20 cloud providers and builds its own inference software on top, allowing customers to deploy and adjust models using their own data.

      Much of this operation utilizes lower-cost, open-source models rather than the advanced systems from OpenAI or Anthropic. This is part of their overall strategy. “Open-source models are improving rapidly,” CEO Tuhin Srivastava told the WSJ, noting that customers are increasingly combining open-source and proprietary models based on the difficulty of the tasks.

      A rapid re-evaluation is taking place as investors are quickly supportive of that idea. Baseten’s valuation was $5 billion only in January when it raised a $300 million round with backing from IVP, CapitalG, and Nvidia. This new funding deal more than doubles that valuation in about five months. Just as recently as September 2025, the company was valued at approximately $2.15 billion.

      The driving force is demand. Customers like Cursor, Mercor, and OpenEvidence use Baseten to manage models, and the company claims that some clients have significantly reduced their expenses by switching to open-source alternatives, with one reportedly completing a task at about 30% of the cost compared to a proprietary option.

      This new funding comes amidst a gold rush for AI inference, as the technology that enables models to operate efficiently has become highly sought after within the industry. Cerebras, a maker of inference chips, went public earlier this year, and Fireworks AI aims to enter the same market for model serving.

      Additionally, it arrives during a price war. Many popular open-source models now originate from China, such as DeepSeek and Moonshot AI, and even Nvidia has introduced an open family known as Nemotron. With OpenAI reportedly considering substantial price reductions to remain competitive, this pressure spells trouble for leading labs and, theoretically, presents an opportunity for those who assist companies in using the more affordable options.

      The risk accompanies the opportunity. Baseten is currently valued as a beneficiary of falling costs, even as those declines are putting pressure on margins throughout the sector. A $13 billion valuation on a company at the heart of a price war represents a bet that the tools for this endeavor outlast the gold rush itself.