Modal passes $1B in annualized revenue, valued at $15.75B

Modal, an AI compute platform, announced on October 14 that its annualized run rate has surpassed $1 billion. It also closed a $750 million Series D at a post-money valuation of $15.75 billion. Accel led the round, with Kleiner Perkins participating, and existing investors including Menlo Ventures, General Catalyst, Redpoint, Lux Capital, Amplify Partners and Bain Capital Ventures all following on.

Co-founder and CEO Erik Bernhardsson wrote in the company blog:

“We recently crossed $1B in annualized run rate with no sign of slowing down.”

Usage up about 3x in five months

Modal gave two growth figures. Platform usage has grown roughly 3x since its previous round, the Series C, about five months ago, and by more than 15x year over year.

Part of that growth comes from Sandboxes, the isolated environments in which AI agents execute code. The company says this product alone now accounts for about $300 million in annualized revenue, has launched nearly 3 billion sandboxes in total, and can scale to 1 million concurrent sandboxes. Customers named in the blog include Cognition, DoorDash, Ramp, Linear, Hebbia, Runway, Physical Intelligence and Suno. One of them, a company called TypeSafe AI, processed 1 trillion tokens on Modal in the three days before its product launched.

On the compute side, Modal does not build its own data centers; its resource pool draws on more than 20 cloud providers. The team has more than 200 people across New York, San Francisco and Stockholm. The blog says only that the new money will go toward “continuing to build the next generation of products” and gives no breakdown of how it will be spent. The total number of customers and the size of its GPU fleet were not disclosed either.

Three rounds, side by side

Lined up, the milestones from Modal's past year look like this:

Reports differ on the size of the Series B, with figures of both $80 million and $87 million in circulation, though all put the valuation at around $1.1 billion. The Series C was also split into two tranches: the first batch of investors came in at a $2.5 billion valuation, and after demand outstripped expectations the second batch was priced at $4.65 billion.

Valuation and revenue have climbed almost in lockstep. By a rough calculation from the table, the ratio of valuation to annualized revenue at the three milestones is about 18x, 15.5x and 15.75x. In other words, the valuation has risen more than tenfold in a year, yet investors are not paying noticeably more for each dollar of revenue.

At the time of the May round, Bernhardsson credited most of the revenue growth to AI coding. Once an agent has written code, it needs somewhere to run it and a way to keep it isolated, which is exactly the business Sandboxes is in; the presence of coding-agent companies such as Cognition on the customer list fits that logic.

Building the lower layers itself

The blog spends considerable space on why Modal wrote its own file system, container runtime, scheduler and networking layer. AI workloads look very different from traditional web services, and cold starts, bursty scaling and GPU scheduling all had to be redesigned. In recent months it has shipped multi-node training, Modal servers and Endpoint Candidates, and moved the VM-based Sandboxes to general availability.

Companies that focus on sandboxes, such as E2B and Daytona, are in the same space, and several large cloud providers are building out their own agent runtime environments. Modal has not said what share of its revenue comes from its largest customers or how concentrated its customer base is. That is the data point missing from this announcement.

Bernhardsson closed the post with one more line:

“I'm obviously biased, but I can't think of a better time to be building software.”

Sources: Modal official blog, Reuters, CocoLoop, Dealroom. Annualized revenue, funding amounts and post-money valuation follow the basis used in Modal's announcement; the Series B amount is treated as a range across multiple reports.