Bloomberg reported on 30 October that Baseten, which sells AI inference software and computing capacity, “is set to raise new funding that would more than double its valuation to about $30 billion”, citing people familiar with the efforts.
Reuters reported in June that Baseten said it raised $1.5 billion at a $13 billion valuation. By our arithmetic, about $30 billion is 2.3 times that June mark, roughly $17 billion of added value in about four months.
Business Pill · TOKENS PER WATT
A short Business Engineer explainer on the economics of producing AI tokens. It teaches the general idea only and does not discuss Baseten.
The key insight: Bloomberg reports Baseten is set to raise at $26 billion and $30 billion in two tranches; against the $13 billion valuation Reuters reported in June, the $30 billion tranche is 2.3 times, by our arithmetic.
According to Bloomberg, the round is “set to include two separate prices, or” tranches. The people said “the first segment of which values Baseten at $26 billion”, and that “The second tranche values Baseten at $30 billion”. Bloomberg’s readable text does not give the round’s size or its lead investors.
The $26 billion figure matches earlier talks. On 23 September, Bloomberg reported that Baseten “is in talks for a funding round that could value it at $26 billion, people said, up from $13 billion in June”, and noted that Axios first reported details of the round. Bloomberg wrote then that “Baseten and Modal declined to comment”.
In June, Reuters reported that the $1.5 billion round “was led by U.S.-based investors Sands Capital and Wellington Management”, as Baseten said, and that the company said “its revenue has grown 20-fold over the past year”. Reuters described it as “Baseten’s fourth capital raising in 18 months”.
Baseten valuation marks as reported: $13 billion in June 2026 (Reuters), talks at $26 billion on 23 September (Bloomberg, Axios first) and a round set at $26 billion and $30 billion in two tranches on 30 October (Bloomberg, citing people familiar). The October round is set, not closed. $30 billion divided by $13 billion is 2.3x, our arithmetic.
The Business Lever: Selling Inference Below Lab Prices
Reuters wrote in June that Baseten sells software and infrastructure using “what it says is a cheaper alternative to providers such as OpenAI and Anthropic”. Baseten said it “would use the funds to expand computing capacity, software and hiring”.
A day before Bloomberg’s report, Baseten released joey-1, a decision model it prices at “$0.01 per million tokens”. Its blog says the model “costs a tenth of mammoth-1” and serves the average 2,000-token request at a P50 latency of 20 ms. Earlier on 30 October we covered IREN naming Perplexity Computer as a GPU tenant, a day on which Perplexity also tied a model rollout to having sufficient token capacity.
The Structural Read
As we read the two Bloomberg reports, the $26 billion figure first surfaced as talks on 23 September and now appears as the first tranche, with a second tranche at $30 billion.
As we read Reuters’ June report, Baseten pitched itself on price: a cheaper alternative to OpenAI and Anthropic, with revenue it said had grown 20-fold in a year.
As we read Baseten’s joey-1 post, the company is also selling its own models, pricing joey-1 at $0.01 per million tokens on its Model APIs.
Michael Tolo, partner at Blackbird, to Reuters, 23 June 2026
“It’s a signal of conviction,”
Three Implications
FOR BASETEN Bloomberg’s readable text gives no round size or lead investor, so how much new money comes in at each price is not public in what we read.
FOR JUNE’S INVESTORS Reuters reported the June round was led by Sands Capital and Wellington Management at $13 billion; the October prices are 2.0 and 2.3 times that, by our arithmetic.
FOR INFERENCE BUYERS Reuters wrote that Baseten says it is a cheaper alternative to OpenAI and Anthropic; joey-1 is priced at $0.01 per million tokens, per Baseten’s blog.
The framework says “AI has crossed from the training era to the inference era.”
As we read it, Baseten is priced on that shift: Bloomberg reports a $30 billion top price for a company Reuters described as selling inference infrastructure. The framework was published in March 2026 and does not discuss Baseten.
What Is Not Established
Bloomberg’s readable text says the round is set to happen, not that it has closed, and gives no amount raised, lead investor or revenue figure. Baseten had not commented on the report in what we read. The 2.3 times multiple compares the $30 billion second tranche with June’s $13 billion valuation; the first tranche, at $26 billion, is 2.0 times June’s mark by our arithmetic.
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
Bloomberg reports Baseten is set to raise at $26 billion and $30 billion in two tranches, against the $13 billion Reuters reported in June. By our arithmetic, the $30 billion tranche is 2.3 times the June mark. Separately, Baseten’s blog prices its joey-1 model at $0.01 per million tokens.
A note on sourcing. The October round is reported by Bloomberg, citing people familiar with the efforts; only the opening passages of its article were readable to us. June figures come from Reuters’ report of Baseten’s announcement, and the joey-1 price from Baseten’s blog. We did not contact Baseten. Figures labelled our arithmetic are our own.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
