Baseten is raising $1.5 billion. Its valuation has gone from $5 billion in January 2026 to a reported $13 billion now. That is five months. The speed is worth pausing on, because it reflects something real happening in the market, not just investor enthusiasm.
The company does a specific, unglamorous job. When an AI system answers a question, generates a document, or checks an insurance claim, something has to process that request in real time. That processing step is called inference. Baseten handles it. It pulls computing capacity from around 20 different cloud providers, then routes each request to whatever option is fastest and cheapest at that moment.
The revenue trajectory backs up the valuation jump. Baseten's annualized revenue has moved from roughly $200 million to $600 million. That is a tripling of real sales, not just a paper number. Customers include Cursor, Mercor, and OpenEvidence, and some have reportedly run certain AI tasks at about 30 percent of the cost they paid using a proprietary provider.
That cost gap is the whole business case. Proprietary AI models from OpenAI and Anthropic are good, but they charge platform-level prices. Open-source models have closed most of the quality gap in the past year, and companies running AI at scale are starting to notice the difference on their bills. Baseten is the bridge: it gives businesses access to cheaper models without requiring them to figure out the technical side themselves.
The funding round structure is worth understanding. Some investors are coming in at an $11 billion valuation, others at $13 billion. This split-price tactic is increasingly common in AI fundraising. It lets a company claim the higher number as its headline valuation while the lead investors who took on more risk pay a lower price. It is not fraud, but it does mean the $13 billion figure is partly a ceiling, not a consensus.
For businesses outside the AI industry, Baseten is not something you need to buy or use directly. But the dynamic it represents is worth tracking closely. AI running costs are becoming a real budget line. The companies and tools your business eventually uses to automate work, process documents, or serve customers will all sit on infrastructure like this. The cheaper and more efficient that layer gets, the more AI-powered services will drop in price.
The inference market overall is large and growing fast. One estimate puts it at over $100 billion today and roughly $255 billion by 2030. The money flowing into Baseten is a bet that the layer between a business and its AI model is where the value gets captured, not in the model itself.
Baseten has now raised close to $2 billion in total. It also acquired a machine-learning research startup called Parsed earlier this year and signed a strategic agreement with Amazon Web Services. The company is building infrastructure it expects to become as standard as cloud computing itself. Whether the valuation holds at $13 billion or not, the underlying need it is serving is not going away.