Modal
Serverless cloud platform for AI development and inference
Modal provides a serverless infrastructure platform designed for building, training, and serving artificial intelligence applications. Founded by Erik Bernhardsson and Akshat Bubna, it allows developers to scale code across thousands of GPUs without managing underlying hardware. The company is based in New York, Stockholm, and San Francisco.
- Raised over $1.2 billion from top venture firms
- Runs over 40 million gVisor containers daily
- Infrastructure spans 20+ cloud providers globally
- Pay-per-second pricing with no idle costs
- Scales from zero to thousands of GPUs instantly
What it is
Modal is a cloud infrastructure provider focused on simplifying the deployment of data, AI, and machine learning applications. It offers a serverless environment where code scales automatically from zero to thousands of processors based on demand. The platform handles container scheduling, image building, and storage, allowing teams to focus on software rather than hardware management. Users pay only for the exact time their code executes, avoiding costs for idle resources. The service supports tasks ranging from large language model inference to computational biology and media processing.
How it came about
The company was established by Erik Bernhardsson and Akshat Bubna with the aim of improving the developer experience for AI workflows. Their team includes creators of open-source tools such as Seaborn and Luigi, alongside academic researchers and engineering leaders. Modal has secured more than $1.2 billion in funding from investors including Accel, General Catalyst, Redpoint Ventures, Lux Capital, Amplify Partners, and Creandum. The organization maintains offices in New York, Stockholm, and San Francisco, with hiring presence in London.
The compute
Modal operates a custom container runtime and scheduler designed for rapid startup of large images. It utilizes a distributed file system to load model weights and datasets at high bandwidth across regions. The platform supports serverless functions, secure sandboxes for untrusted code, and dedicated servers for low-latency inference. It integrates with open-source frameworks for reinforcement learning and fine-tuning. Hardware health checks run continuously to replace faulty nodes before they impact workloads. The system manages capacity across multiple cloud providers automatically.
Where the GPUs are
The infrastructure spans more than twenty cloud providers globally. Regions include various zones in the United States, Europe, Asia Pacific, Japan, Australia, United Kingdom, Canada, Middle East, South America, Africa, and Mexico. Modal routes workloads to appropriate hardware in the correct region without manual intervention. This multi-cloud approach ensures resource availability and reduces dependency on single providers. The company maintains physical team locations in New York, Stockholm, San Francisco, and London.
Customers and deals
Leading AI teams use the platform for production workloads. DoorDash employs it for internal tools and customer-facing applications involving inference and agent sandboxes. Cognition utilizes the infrastructure for reinforcement learning and real-time serving. TypeSafe AI reported serving one trillion tokens within three days of launch using Modal. Ramp uses the sandbox capabilities to build its Inspect tool. These organizations cite speed to market and reduced infrastructure complexity as key benefits.
Money and ownership
Investors have contributed over $1.2 billion to the company. Backers include prominent venture firms such as Accel, General Catalyst, Redpoint Ventures, Lux Capital, Amplify Partners, and Creandum. The pricing model is entirely usage-based, charging by the second for compute time. This structure aims to eliminate costs associated with reserved capacity or over-provisioning. The company remains privately held and continues to raise capital to support growth and infrastructure expansion.
Its place in AI
Modal positions itself as a critical enabler for production AI systems. It supports the deployment of open-weight models like Kimi K3.1 Pro. The platform facilitates complex tasks such as multi-node training and interactive coding agents. By abstracting hardware management, it allows engineers to iterate quickly on experimental models. The company contributes to open-source security projects, serving as a maintainer for gVisor, which runs millions of its daily containers.
Questions people ask about Modal
What type of computing does Modal provide?
Modal offers serverless cloud infrastructure specifically optimized for AI and machine learning workloads. It allows users to run code on CPUs and GPUs without managing servers, scaling automatically based on demand.
How does Modal charge for its services?
Pricing is entirely usage-based. Customers pay only for the seconds their code is actively running. There are no charges for idle time, reservations, or upfront capacity commitments.
Which companies invest in Modal?
Major investors include Accel, General Catalyst, Redpoint Ventures, Lux Capital, Amplify Partners, and Creandum. The company has raised over $1.2 billion in total funding from these and other industry leaders.
Where is Modal based?
The team operates out of New York, Stockholm, and San Francisco. They also have a hiring presence in London. The infrastructure itself is global, spanning multiple cloud providers across many regions.
Good reading and sources
There is no English Wikipedia article on Modal yet, so the facts on this page come from the sources listed above: modal.com/company, modal.com/. Some are the subject's own pages, so its claims about itself are reported as its claims. EveryNeocloud wrote the text in its own words, with AI assistance, and compares it with the sources to make sure no passage is copied.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.