Modal Labs raises 750 million dollars in Series D funding at 15.75 billion dollar valuation

a cell phone sitting on top of a table — Jakub Żerdzicki / Unsplash

Modal Labs, a provider of artificial intelligence infrastructure, announced that it has raised 750 million dollars in a late-stage funding round. The company's valuation now stands at 15.75 billion dollars. This figure represents a more than threefold increase from the valuation established during its previous funding round in May. The company stated that its annual revenue has grown by more than three times since that May round, surpassing the 1 billion dollar mark.

Accel led the Series D funding round. Kleiner Perkins and six existing investors in Modal also participated in the transaction. The significant rise in valuation reflects the rapid growth of the company's operations and revenue base over the recent period.

Nearly one third of Modal's revenue is generated by its Sandboxes service. This cloud offering provides customized containers designed specifically for artificial intelligence workloads. Developers utilize these containers as environments for training models. The service also supports the execution of artificial intelligence agents and the code generated by those agents. Modal stated that Sandboxes can provision tens of thousands of containers per second. This scalability is enabled by a custom infrastructure manager or scheduler. The scheduler processes container provisioning requests in parallel across multiple servers to accelerate performance. In recent internal tests, the company reported running up to 1 million containers simultaneously.

Sandboxes hosting artificial intelligence agents sometimes connect to a second set of containers known as Sidecars. These Sidecars run supplementary software, such as tools that detect and filter hallucinations. Modal stated that Sandboxes run Sidecars on the same physical server as the hosting container to reduce data transmission latency. Some agents require access to external applications. A feature called Secret Injection allows agents to use application login credentials without viewing the data itself. This restricted access helps protect credentials from unwanted agents and cyberattacks.

Modal presents Sandboxes alongside cloud services tailored for artificial intelligence training workloads. These services allow developers to enable graphics card clusters using relatively simple scripts. Chips in Modal's clusters exchange data using RDMA, a network protocol that accelerates data transfer by reducing the path packets must travel. The company competes in the inference market with a service called Endpoints. Endpoints help developers deploy artificial intelligence models on hosted infrastructure and make them accessible through application infrastructure. Organizations that do not wish to build their own models can select from open source algorithm libraries prepared by Modal.

The company's services run on hardware leased from more than 20 cloud providers. If one cloud provider experiences a technical issue, customer workloads can continue to run on other cloud providers. Modal plans to use part of the newly raised capital for product development. The company intends to launch services focused on reinforcement learning and post-training processes. These steps in the artificial intelligence development workflow are designed to improve the reasoning, safety, and cost efficiency of artificial intelligence systems being trained.