Modal $750M Series D: AI Cloud Hits $15.75B

Modal, the serverless cloud platform built for production AI, has closed a $750M Series D financing that values the company at $15.75 billion, more than tripling its valuation in just five months. The round was announced on October 14, 2026, by chief executive Erik Bernhardsson in a company blog post, and was led by Accel with participation from Kleiner Perkins and every existing investor, according to the announcement.

The $750M Series D arrives alongside a striking growth claim: Modal says it has crossed $1 billion in annualized run rate. According to PYMNTS, usage of the platform has tripled since the company's $355 million Series C in May 2026, while year-over-year growth exceeds 15x. That May round had valued Modal at $4.65 billion, so the new mark represents a dramatic re-rating of the company in half a year. For a startup founded only five years ago, the pace of the climb is remarkable.

The fundraising reflects a broader investor conviction that the execution layer of AI — where models, code and agents actually run — may become the most strategically valuable part of the stack. The $750M Series D marks one of the largest AI-infrastructure financings of the year. Developers already use Modal to move experiments into production, build agents, generate video, train robots and run drug-discovery workloads, according to the company's announcement.

Sandboxes: the $300 million engine behind the run rate

A single product line explains much of the momentum behind the $750M Series D. Modal's Sandboxes — hosted containers optimized for AI workloads — now generate roughly $300 million of run-rate revenue on their own, according to the announcement. Developers use the containers as training environments, and the service has become a natural home for AI agents and the code they generate, reported by SiliconANGLE.

The engineering behind Sandboxes is built for scale. Modal says the product can provision tens of thousands of containers per second, enabled by a custom scheduling system that parallelizes provisioning requests across servers. In a recent internal test, the company reported running one million concurrent containers, and to date it has spun up close to three billion sandboxes in total.

Agents get special treatment in the architecture. Sandboxes that host AI agents can attach a "sidecar" container equipped with auxiliary software — for instance, a tool that screens for hallucinations — running on the same physical server to reduce latency. A feature called Secret Injection lets agents use application login credentials without ever viewing them, a design meant to protect credentials from rogue agents and cyberattacks, according to SiliconANGLE's report.

Customer traction cited alongside the $750M Series D is spreading across high-profile workloads. TypeSafe AI's Jev processed one trillion tokens on Modal in its first three days, according to Oossa's coverage of the round, while DoorDash's AI team uses the platform to move experiments into production. Runway, Physical Intelligence and Chai Discovery run video generation, robot training and drug-discovery workloads on the service, reported by Oossa.

Why Accel calls AI infrastructure the fastest-growing market

Accel's investors framed the round as a bet on the compute layer itself. The $750M Series D also signals how investors now rank that layer relative to model builders. In a blog post accompanying the financing, Accel partners Rich Wong and Gonzalo Mocorrea described the infrastructure used to train, serve and run AI as the fastest-growing market in technology — and one still in its very early stages, according to the firm's statement. As the market shifts from AI-native startups to every company using AI tools, Modal supplies the building blocks, the partners added.

The broader market data supports the urgency. PYMNTS reported in 2025 that inference infrastructure was becoming the next decisive layer of the generative AI stack, because latency, cost, scale, energy use and deployment location determine whether an AI service works or fails. Enterprises deploying generative AI were already facing inference costs that can reach hundreds of millions of dollars annually at large deployments, according to that report.

For the agent ecosystem, the financing carries a direct message. Agent developers need fast, secure sandboxes to execute generated code, provisioned infrastructure that scales to zero, and isolation that keeps credentials safe when agents act autonomously. Modal's round suggests investors believe the platform that standardizes that runtime — for coding agents, customer-support agents and beyond — could capture a large share of AI spending. The financing wave around enterprise AI agents has repeatedly highlighted execution infrastructure as the bottleneck between promising agent demos and production deployments.

Moving up the stack

Modal is spending the new capital on product breadth. The company says it has rebuilt its core stack — file system, container runtime, scheduler and networking — to run across more than 20 cloud providers. Recent launches include multi-node training with RDMA-connected clusters, Modal servers for HTTP serving, and Endpoint Candidates for large-language-model inference, according to the announcement. Virtual-machine-based Sandboxes are now generally available.

The firm is also moving up the stack into reinforcement learning and post-training products, areas where agent developers increasingly need specialized tooling. The $750M Series D was joined by Menlo Ventures, General Catalyst, Redpoint Ventures, Lux Capital, Amplify Partners and Bain Capital Ventures alongside Kleiner Perkins, according to RuntimeWire's report on the round, leaving Modal with one of the deepest benches of venture backers in AI infrastructure. The deal sits alongside other large enterprise commitments, including the DataRobot agreement deepening agentic AI adoption across the Middle East.

The open question is whether Modal is becoming durable infrastructure for AI software or whether the valuation is mainly capitalizing on temporary scarcity in GPU capacity and a spending rush around agents. The run-rate milestone and the repeat participation of every existing investor suggest the backers are betting on the former. With the demand curve still steepening, the company says it plans to keep building what comes next, funded by the $750M Series D — and now it has three-quarters of a billion dollars to do it.