The Den Reviewthe briefing for executives who run AI
Wednesday, November 11, 2026

The Denominator

Subscribe
Trending

Model Labs · Inference Funding

Fal Reportedly Doubles Its Valuation to $16 Billion Since April

It raised $1.3 billion and passed $800 million in annualized revenue. Compute costs keep inference margins thin.

· 2 min read

A technician walks a long aisle between racks packed with copper-colored components and cables in a large industrial data hall.
Photograph: AI-generated

Fal, a San Francisco startup that hosts AI image, video and audio models, doubled its valuation to $16 billion from $8 billion in April with a Series F led by Bessemer Venture Partners, according to a published report.

Margins at inference providers run thin because compute is costly to buy or lease, the report said, citing several people familiar with their financials. It gave no margin for Fal itself, so the valuation says little about whether a provider’s prices can hold.

What did Fal announce?

Fal announced the Series F together with a Series E led by Sequoia Capital, and Nvidia kept investing. The company said the money buys inference capacity and hires, and covers its 2027 plans.

The report put the two rounds at $1.65 billion combined, with Sound Ventures and T. Rowe Price in the Series F. By The Denominator’s arithmetic, that leaves about $350 million for the Series E. The report also said Fal has passed $800 million in annualized revenue.

Where do margins stand?

At $800 million in annualized revenue, a $16 billion valuation is about 20 times revenue, by The Denominator’s arithmetic. Revenue does not show what a provider keeps after paying for compute, and the report gave no margin for Fal.

Baseten, Fireworks AI and Modal have also raised large rounds at rising valuations, the report said. Their gross margins stay thin for the same reason, the same people said.

Who depends on Fal?

Entertainment studios, advertising agencies, e-commerce brands and game developers run image, video and audio models through Fal inside their production pipelines, the company said. Bessemer, which has backed Fal since its Series B, named Adobe, Canva and Shopify among the companies that have moved production volume to Fal.

Those customers buy inference from a provider in a business where margins run thin. If compute costs rose faster than prices, would a customer’s inference contract hold?

NVIDIAMicrosoftOpenAIAnthropicGoogleAWSJPMorganCapital OneAT&TCharterComcastT‑MobileUnitedHealthCVS HealthTravelersWalmartSalesforceDatabricksSnowflakeCoreWeaveStripeAccenture