Modal Labs Inc., a provider of artificial intelligence infrastructure, today announced that it has raised $750 million in late-stage funding. 

Accel led the Series D round. It was joined by Kleiner Perkins and a half-dozen existing Modal investors. The company is now valued at $15.75 billion, more than triple what it was worth after its last funding round in May.

The valuation bump reflects Modal’s rapid growth. The company says that its annualized revenue has more than tripled since the May round to over $1 billion. 

Nearly a third of Modal’s revenue comes from Sandboxes, a cloud service that provides hosted containers optimized for AI workloads. Developers can use the containers as model training environments. Additionally, the service lends itself to running AI agents and the code they generate.

Modal says Sandboxes is capable of provisioning tens of thousands of containers per second. That scalability is made possible by a custom infrastructure management system, or scheduler. The scheduler parallelizes container provisioning requests across multiple servers to speed up processing. Modal says that it managed to run 1 million concurrent containers in a recent internal test.

Sandboxes that host AI agents are sometimes connected to a second container, or sidecar, equipped with auxiliary software. For example, it might run a tool that detects and filters hallucinations. Modal says Sandboxes runs sidecars on the same physical server as the containers they support to reduce latency.

Some agents require the ability to log into external applications. According to Modal, a Sandboxes feature called Secret Injection enables agents to use appliation login credentials without viewing them. That limited access protects credentials from rogue agents and cyberattacks.

Modal offers Sandboxes alongside a cloud service optimized for AI training workloads. It enables developers to spin up graphics card clusters with relatively simple scripts. The chips in Modal’s clusters exchange data using RDMA, a network protocol that speeds up traffic by shortening the route through which packets travel.

The company competes in the inference market with a service called Endpoints. It enables developers to deploy AI models on hosted infrastructure and make them available through an application infrastructure. If an enterprise doesn’t wish to build its own model, it can use a library of open-source algorithms provided by Model.

The company’s services run on hardware that it rents from more than 20 cloud providers. If one of the clouds experiences technical issues, customer workloads can continue running on the others.

Modal will invest a portion of its newly raised funding in product development. The company plans to launch services focused on reinforcement learning and post-training. Those are phases of the AI development workflow designed to improve the reasoning capabilities, safety and cost-efficiency of the AI being trained.

Photo: Modal