Fireworks AI in Talks to Raise at $30B to $40B, Roughly Double July Mark

Fireworks AI is in early talks to raise funding at a valuation of $30 billion to $40 billion, roughly double the $17.5 billion price it set in its July Series D, according to a Business Insider report citing people familiar with the matter. At the top of the range, the company would be worth ten times its Series C mark.

The caveat is timing. The talks are preliminary, the target could change, and Fireworks declined to comment. No round size appears in any report on the discussions. The Information reported in late September that Fireworks had considered raising at about $30 billion, so the new range extends above that figure. In plain words: the number is a ceiling under negotiation, not a signed price.

What has investors actually paid for?

Fireworks was founded in 2022 by engineers from the team behind PyTorch, the open-source AI framework developed at Meta. CEO Lin Qiao previously led PyTorch at Meta. Her co-founders span Meta, Google and compiler expertise:

• Benny Chen, former Meta ads infrastructure lead
• Chenyu Zhao, who led Google's Vertex AI
• Dmytro Dzhulgakov and Dmytro Ivchenko, both PyTorch veterans at Meta
• James Reed, who worked on the PyTorch compiler
• Pawel Garbacki, who led core machine learning for Meta's news feed

The platform lets companies train and serve custom versions of open models on their own data. Fireworks lists Uber, Shopify, Doximity, Geico and Revolut as customers, and says Cursor and Harvey build specialised models on it — Harvey's legal AI being one example.

The funding history is steep. Tech Funding News first reported $4 billion valuation talks in July 2025. A $250 million Series C followed in October 2025, led by Lightspeed Venture Partners, Index Ventures and Evantic, bringing total funding to $327 million. On July 16, 2026, Fireworks announced a $1.505 billion Series D at a $17.5 billion valuation, led by Atreides Management, Index Ventures and TCV. Evantic, Lightspeed and Nvidia participated as existing investors, and 17 investors contributed in total, including 20VC, Bessemer Venture Partners, Insight Partners, Lone Pine Capital, Menlo Ventures and Ontario Teachers' Pension Plan.

Fireworks said at the Series D that it had passed $1 billion in annualised revenue, up fivefold from its last round. Daily token volume nearly tripled, from 15 trillion to more than 40 trillion. About 95% of those tokens come from models customised on customers' own data. No newer revenue figure has been published since.

How does the valuation compare with rivals?

Fireworks is not repricing alone. Baseten raised $300 million at $5 billion in January, then a $1.5 billion Series F at $13 billion on June 22, led by Altimeter Capital, Conviction and Spark Capital. Bloomberg reported on September 23 that Baseten was in talks at about $26 billion. Modal Labs raised $355 million at $4.65 billion in May, then announced a $750 million Series D at $15.75 billion on October 14, led by Accel, and passed $1 billion in annualised run rate. Together AI raised $800 million at $8.3 billion on July 1, led by Aramco Ventures.

The multiples are where Fireworks stands out. By TFN's calculation, $30 billion to $40 billion equals 30 to 40 times the $1 billion run rate disclosed in July. The Series D priced the company at about 17.5 times. Modal's $15.75 billion sits at no more than roughly 16 times its run rate, and Baseten's $13 billion was about 22 times a reported $600 million run rate from the first quarter. Fireworks has not published a newer revenue figure, so its true multiple depends on growth since July. Baseten and Modal also priced earlier rounds in tranches, meaning headline valuations blend different entry prices. The multiples, in other words, rest on numbers that may already be stale.

What is not in the pitch?

An October security incident exposed API keys and tokens in some customer accounts — a fact absent from the fundraising coverage. Whether investors discount for it, or for the stale revenue figure, will show in the final term sheet.

The market backdrop is large. IDC raised its 2026 forecast for AI infrastructure spending to $497 billion on July 21 and expects $1.08 trillion in 2029. Servers made up 97.6% of first-quarter spending, so the figure tracks the compute these startups buy, not the software they sell. If the talks close anywhere near $40 billion, Fireworks will have raised the bar for every inference rival now queuing for repricing.