Modal says its Sandboxes product, which runs AI-generated code, has reached a $300 million annualised run rate, with close to three billion sandboxes spun up. The figures point to a substantial business built around isolating code from AI systems, not just another cloud infrastructure funding headline.
Modal Labs Watch analysis
What happened
Modal says it built its own file system, container runtime, scheduler and networking to handle AI workloads, using a capacity pool spanning more than 20 clouds. The company’s Sandboxes product launched in 2023. It also says it is expanding into multi-node training, reinforcement learning, post-training and new inference features.
The details come from The AI Insider’s report, which covers Modal’s funding round and growth. Modal says its customers include DoorDash, Cognition, Ramp, Runway and Physical Intelligence.
Why it matters
Running code produced by AI models means giving that code somewhere to execute without handing it the keys to everything else. Modal’s reported scale suggests that isolated execution is already a meaningful commercial use of AI infrastructure. The company’s own software stack and multi-cloud capacity also show how much plumbing may sit behind a service that looks, to a developer, like a place to run a task.
A $300 million run rate is a company-reported measure, not a public breakdown of recognised revenue or profit. Still, nearly three billion sandboxes is a striking volume claim, and gives the business story more substance than a funding total alone.
Our read
The interesting signal is that AI-generated code needs its own execution infrastructure, and Modal says customers are paying at scale for it. That is worth watching as the company moves beyond sandboxes into training and inference. The useful next question is whether this reported activity translates into dependable capacity and clear security boundaries for customers, rather than simply impressive counters.
What to watch
- Whether Modal shares more detail on what its sandbox volume and run-rate figures measure.
- How its multi-cloud infrastructure supports reliability and capacity as demand grows.
- Whether its planned training and inference services become distinct products, or extensions of the same platform.
Discussion spark: For AI-generated code, should customers favour a specialist sandbox provider, or should execution isolation be a standard feature of the model platforms they already use?
Sources and evidence
- Modal Secures $750M Series D at $15.75B Valuation After Passing $1B Run Rate (16 October 2026, 17:07 UTC)
not affiliated with or endorsed by Modal Labs