Modal Triples Its Valuation to $15.75 Billion in Five Months Without Owning a Single Server

Modal Labs closed a $750 million Series D on October 14 at a $15.75 billion post-money valuation, Accel leading with Kleiner Perkins joining as a new backer. The number that should catch an investor's eye isn't the valuation itself — it's that Modal tripled its price tag in under five months, going from a $4.65 billion Series C in May to $15.75 billion now, while its annualized revenue run rate climbed from roughly $300 million to just over $1 billion over the same stretch. The company owns no data centers; it brokers spare capacity from more than 20 cloud providers and resells it as a developer platform, with its fastest-growing line, a product called Sandboxes that gives AI agents an isolated place to execute code, now accounting for about $300 million of that run rate on its own.

A multiple that didn't move

What's analytically interesting is what stayed constant. At the May round, Modal was priced at roughly 15.5 times its then-run rate. At this round, it's priced at roughly 15.75 times a run rate that's more than three times larger. In other words, Accel and its co-investors aren't betting on a re-rating of the business — they're underwriting a continuation of the same growth multiple against a much larger revenue base, which is a materially different bet than paying up for unproven momentum. That matters because two of Modal's closest comparables, inference-hosting rivals Baseten and Fireworks AI, sit in a similar band: Baseten raised at roughly $13 billion in June after reportedly fielding takeover-style interest near $26 billion, and Fireworks AI raised $1.5 billion in July at $17.5 billion against a run rate above $1 billion — also around 17.5 times revenue. Together AI, by contrast, priced its July round at roughly 7 times bookings rather than run rate, a function of its business leaning more on long-term reserved-capacity contracts than on-demand consumption. Lined up, the pattern suggests the market has converged on a roughly 15-to-18-times-revenue band for AI infrastructure middleware specifically, distinct from the far richer multiples attached to foundation-model labs and the far thinner multiples attached to capital-intensive GPU owners.

Since our Series C, usage has tripled and we've grown more than 15x year-over-year.

That line, from Modal's own funding announcement, is the clearest public figure the company has given on revenue growth, and it doesn't fully reconcile with the run-rate math — a tripling of run rate in five months from a $300 million base would land closer to 10x annualized than 15x, unless growth accelerated sharply within the period rather than compounding smoothly. Investors underwriting this round are therefore pricing a growth trajectory that's still accelerating, not one that's merely continuing at its prior pace, which raises the bar for what the next twelve months need to show.

Renting capacity instead of buying it

The structural bet here is distinct from the GPU-ownership model that dominates most AI infrastructure financing headlines. Modal doesn't buy chips or sign power purchase agreements; it arbitrages idle capacity across cloud vendors and sells a managed developer experience on top. That keeps its capital intensity low relative to infrastructure operators that are financing physical buildouts, and it's a structural reason the company can grow revenue multiples faster than balance-sheet-heavy peers — there's no depreciation schedule or debt-service burden sitting underneath the growth. It's also the model's key vulnerability: the moat is thinner. Any hyperscaler or model lab with surplus compute could, in principle, stand up a competing sandbox product, and Modal's customer list — DoorDash, Ramp, Linear, Cognition, and Runway, among others — reads as sophisticated technical buyers who could plausibly build in-house if the economics shifted.

The macro backdrop investors are underwriting

Gartner's current forecast puts worldwide spending on AI-optimized infrastructure-as-a-service at $42 billion in 2026, up 96% year over year, with inference workloads ($23.3 billion) now outpacing training ($19 billion) for the first time. That shift favors exactly the layer Modal, Baseten, and Fireworks occupy — serving models rather than building them — and helps explain why all three have found receptive capital markets even as debt-financed infrastructure deals elsewhere in AI have drawn more scrutiny. For investors, the read-through is that the inference-serving layer is being priced as a distinct, durable category rather than a transient arbitrage, but the 15-to-18-times band these three rounds have clustered around leaves little room for multiple expansion if growth decelerates even modestly from its current pace.