Baseten Nears $1.5 Billion Raise at Up to $13 Billion Valuation as AI Inference Demand Explodes

Baseten Nears $1.5 Billion Raise at Up to $13 Billion Valuation as AI Inference Demand Explodes

The Vault — AI Edition | Business | June 19, 2026

Five months ago, Baseten was a $5 billion company. This week it is reportedly closing in on a $1.5 billion funding round that would value it at as much as $13 billion — a roughly 160% leap in less than half a year, and one of the clearest signals yet that the money chasing artificial intelligence has shifted decisively from training the models to running them.

The deal, first reported by The Wall Street Journal on Thursday, June 18, would more than double the valuation Baseten set in January, when it raised a $300 million Series E at $5 billion. That round, in turn, came just nine months after a $150 million Series D that valued the company at $2.15 billion. The acceleration is dizzying even by the standards of the current AI cycle.

A split-priced round and a record run-rate

The most striking feature of the new financing is its structure. Rather than a single clean number, the round is dual-tiered: some investors are coming in at an $11 billion valuation, while others are paying up to $13 billion for the privilege of getting in at all, according to people familiar with the talks cited by the Journal.

So-called split-priced rounds have become an increasingly common tactic in late-stage AI deals, allowing startups to advertise an eye-catching headline number while giving anchor investors a more favorable entry point on paper. TechCrunch, which covered the report, noted the device is now a recognizable feature of the venture playbook in 2026.

The round is co-led by a heavyweight syndicate — Altimeter Capital, Conviction, Spark Capital, Sands Capital and Wellington Management — and the appetite appears genuine. Baseten's annualized revenue run-rate surged from roughly $200 million at the start of the first quarter to about $600 million by its end, a tripling in a matter of months. As recently as May, the company had been in talks to raise $1 billion at an $11 billion valuation, with some investors reportedly floating offers as high as $15 billion.

Betting that open source is good enough

Founded in 2019, Baseten sits in a corner of the AI stack that was, until recently, easy to overlook: inference, the computationally expensive work a model does after a user submits a prompt. Baseten builds the software layer and rents multi-cloud compute — sourcing capacity from about 20 different cloud providers — that lets enterprises deploy, optimize and fine-tune AI models on their own data.

Its core thesis is that open-weight models have matured to the point where many businesses no longer need to pay premium prices for closed offerings from OpenAI or Anthropic. Companies can download an open model at no cost, adapt it, and hand the operational headache of serving it to a provider like Baseten.

"At the highest level, what's happening generally in the market is that the open-source models are getting very, very good," Baseten co-founder and CEO Tuhin Srivastava told the Journal. "And as open-source gets better, we are growing with it."

That ambition is not modest. After Baseten's September 2025 Series D, Srivastava told PYMNTS that inference is only one piece of the puzzle. "For us, inference is one part of AI infrastructure. Beyond that, there's training, evaluation, fine-tuning. We really want to own that entire loop," he said. "We want to build the next AWS for inference."

The inference gold rush

Baseten's ascent is inseparable from a broader repricing of where value sits in the AI economy. For three years, capital and attention flowed toward the foundation-model labs and the staggering cost of training ever-larger systems. But training is a one-time, episodic expense. Inference is recurring, and it scales with usage — every prompt, every agent action, every embedded AI feature is a paid event. As models get woven into ordinary products and workflows, the volume of inference is compounding.

Baseten's founders made the point bluntly when they announced the Series E in January: "As AI becomes embedded in every product and workflow, inference isn't just growing — it's becoming one of the largest markets ever created. We're building the software that powers it."

Venture investors have heard the message. The publication The Next Web has dubbed the moment an "inference gold rush," and Baseten is far from the only beneficiary; rivals and adjacent infrastructure plays have drawn comparable enthusiasm. The economics are seductive: if open-weight models are converging on the quality of closed ones for routine enterprise tasks, then the durable margin may live not in the models themselves but in the unglamorous machinery that serves them cheaply, quickly and reliably.

There is a cost dynamic underneath all of this worth naming. Closed-model API prices remain high, while open-weight alternatives are free to download and improving fast. Baseten's pitch — route each request to the most competent, least expensive model that can do the job — turns that gap into a business. Its growth tracks an explosion of applications built on open models, and the run-rate numbers suggest enterprises are voting with their inference bills.

What to watch

The financing is described as nearing completion rather than closed, so the final terms — and which tier of the split price dominates — remain to be confirmed. Three questions will define whether the $13 billion sticker proves prescient or premature.

First, durability of the revenue curve: tripling a run-rate in a single quarter is extraordinary, but inference is a competitive, capital-intensive business where compute costs and pricing pressure could compress margins as larger cloud incumbents move in. Second, the open-versus-closed thesis itself: if frontier labs cut prices aggressively or open-weight quality plateaus, Baseten's central wager weakens. Third, the structure of the round is a tell — split-priced deals can signal both intense demand and investor caution about a single clean valuation.

For now, Baseten is the clearest emblem of a market that has decided the next great AI fortune will be made not in building the models, but in running them.

---

Sources:

- [PYMNTS — Baseten Nears $1.5 Billion Funding Round as Inference Demand Surges](https://www.pymnts.com/news/investment-tracker/2026/baseten-nears-1-5-billion-funding-round-as-inference-demand-surges/) - [TechCrunch — AI inference startup Baseten reportedly raising $1.5B months after its last mega-round](https://techcrunch.com/2026/06/18/ai-inference-startup-baseten-reportedly-raising-1-5b-months-after-its-last-mega-round/) - [The Next Web — Baseten raises $1.5bn at up to $13bn for AI inference](https://thenextweb.com/news/baseten-1-5bn-round-13bn-valuation-ai-inference)