Modal, the US cloud computing platform for AI workloads, has closed a $750 million Series D round led by Accel that values the company at $15.75 billion post-money. The company says it already exceeds $1 billion in annualized revenue (run rate), a figure it reports itself and doesn’t back up with audited accounts or profitability data.
Modal’s funding round in 30 seconds
- Modal raises $750 million with Accel leading and Kleiner Perkins joining.
- Its valuation goes from $4.65 billion in May to $15.75 billion, about 3.4 times higher in five months.
- According to the company, its annual revenue rate exceeds $1 billion; in May it was around $300 million.
- Its sandboxes for AI agents already bring in $300 million annualized.
- It doesn’t own data centers: it aggregates GPU capacity from more than 20 clouds.
The announcement, published on October 14 on the company’s blog and signed by CEO and co-founder Erik Bernhardsson, confirms what TechCrunch had reported in late September based on a source familiar with the deal. Besides Accel and Kleiner Perkins, the round included existing investors Menlo Ventures, General Catalyst, Redpoint, Lux Capital, Amplify Partners, and Bain Capital Ventures. Modal hasn’t detailed what it will spend the money on beyond continuing to build its product and expanding its platform toward higher-level layers, such as reinforcement learning and model post-training tools.
What Modal Sells and Why It’s Growing So Fast
Modal started as a way to run code in the cloud without managing servers: developers write functions in Python, and the platform handles spinning up containers, allocating GPUs, and scaling with demand. Bernhardsson explains in his post that the company discarded much of the usual software stack and built its own file system, container environment, and scheduler, aiming to get large models running fast in a market where GPUs are scarce.
On top of that, it has added inference services for language models, multi-node training with no capacity commitment, batch processing, notebooks, and HTTP servers. The fastest-growing product is its sandboxes, isolated environments where AI agents run code. Modal launched them in 2023, says it has created nearly 3 billion of them, and that they can scale up to a million in parallel. On their own, the company says, they already generate $300 million annualized.
The customer list it cites helps explain that growth. Coding-agent and automation companies such as Cognition, Ramp, Linear, and Hebbia use its sandboxes; Runway and Suno use them for video and music generation; Physical Intelligence, Generalist AI, and 1X for robot training; and Chai Discovery and Achira for drug discovery. DoorDash uses it to put experimental AI applications into production. The company says its usage has tripled since its Series C and grown more than 15 times in a year — figures it hasn’t broken down by product or by customer.
A Valuation That Triples in Five Months
The Series C came in May: $355 million at a $4.65 billion valuation, according to TechCrunch and The Next Web. At the time, according to Reuters, Modal had surpassed $300 million in annualized revenue. If the current figures are accurate, the revenue rate has more than tripled over that period, and the valuation has grown at a similar pace. It’s worth remembering that run rate is an annual projection based on a short recent period, usually the last month, and that in a usage-based consumption business it can rise and fall quickly.
The business model also leaves open questions. Modal doesn’t own data centers: it aggregates GPU capacity from more than 20 cloud providers and resells it with its software layer on top. That lets it grow without tying up capital in hardware, but it makes its margins dependent on the price at which it buys that capacity — a figure it doesn’t publish. The company has more than 200 employees spread across New York, San Francisco, and Stockholm.
The Race Among Inference Providers
Modal isn’t alone in this space. According to TechCrunch, citing Bloomberg, Baseten is close to a new round that would value it at $26 billion, double its June valuation, and Fireworks announced in July that it had reached $1 billion in annualized revenue, five times what it had a year earlier. Fal, which specializes in image and video generation, is also in talks with investors, according to The Information. These are companies that, like Modal, rent out GPUs and sell the ability to run models — a market that moves at the pace of AI spending and overlaps with the neoclouds driving up revenue by building their own infrastructure.
Inference cost is today one of the biggest line items in AI labs’ budgets, as shown by the case of Thinking Machines, which pays $65 million a year for inference. That demand is what’s sustaining the valuations of Modal and its competitors. Its continuity will depend on customers keeping up their spending and on GPU access not becoming more expensive than they can pass on.
Frequently Asked Questions
How much is Modal worth after the Series D?
$15.75 billion post-money, up from the $4.65 billion of its Series C in May.
Who led the round?
Accel, with Kleiner Perkins joining and previous investors Menlo Ventures, General Catalyst, Redpoint, Lux Capital, Amplify Partners, and Bain Capital Ventures also participating.
What does a $1 billion run rate mean?
That if the revenue from a recent period held steady for a full year, it would add up to more than $1 billion. It’s a figure reported by the company, not closed annual revenue.
Does Modal have its own data centers?
No. It aggregates GPU capacity from more than 20 cloud providers and offers its own software platform on top of it.
Sources:
