Modal Raises $750M Series D as AI Infrastructure Demand Pushes Valuation to $15.75 Billion

Modal Achieves Triple Valuation Jump in Five Months

Modal, an AI cloud company specializing in infrastructure for developers, closed a $750 million funding round at a $15.75 billion valuation, marking a dramatic acceleration in the AI infrastructure market1

1

. The Series D funding represents more than a tripling of Modal's $4.65 billion valuation from its May Series C round, when it raised $355 million1

1

. Accel led the round with participation from Kleiner Perkins and all existing investors including Menlo Ventures, General Catalyst, Redpoint, Lux Capital, Amplify Partners, and Bain Capital Ventures1

1

.

The New York-based company's growth trajectory reveals the scale of AI infrastructure demand currently reshaping the technology landscape. Usage has tripled in just five months since the Series C, with the business growing more than 15 times year-over-year, according to co-founder and CEO Erik Bernhardsson3

3

. Modal now generates over $1 billion in annualized revenue with no signs of slowing down1

1

.

AI Sandboxes Drive $300 Million in Annual Revenue

Modal's fastest-growing product, Sandboxes, generates $300 million annually and has become critical infrastructure for AI applications1

1

. The service provides hosted containers optimized for AI workloads, enabling developers to use them as environments for training AI models while also supporting AI agents and the code they generate2

2

. Since launching in 2023, Modal has started close to 3 billion AI sandboxes1

1

.

The technical capabilities behind Sandboxes explain its rapid adoption. Modal's custom infrastructure management system can provision tens of thousands of containers per second, with the company successfully running 1 million concurrent containers in recent internal testing2

2

. The scheduler parallelizes container provisioning requests across multiple servers to accelerate processing, addressing the bottleneck many AI applications face when scaling.

For AI agents requiring auxiliary software, Sandboxes runs sidecars on the same physical server as the containers they support to minimize latency2

2

. A feature called Secret Injection enables AI agents to use application login credentials without viewing them, protecting against rogue agents and cyberattacks while maintaining functionality2

2

.

Enterprise Adoption Spans AI Natives to Robotics

Modal's customer base demonstrates the breadth of AI infrastructure demand across industries. Companies including DoorDash, Ramp, Linear, Cognition, and video AI firm Runway rely on Modal's platform1

1

. Robotics companies such as Physical Intelligence and 1X use Modal to train robots, highlighting how AI infrastructure extends beyond software into physical applications1

1

.

Bernhardsson emphasized that companies use Modal's platform to move apps from experimentation to production, build agents, generate lifelike video, train robots, and run drug discovery3

3

. This positions Modal at the intersection of multiple AI application categories, each representing significant market opportunity as enterprise adoption accelerates.

Accel partners Rich Wong and Gonzalo Mocorrea noted that as the market evolves from AI-natives to every company using AI tools, Modal provides building blocks for startups through to the largest companies globally3

3

.

GPU Scarcity Drives Cloud Infrastructure Innovation

Modal addresses critical constraints in AI development by renting GPU computing power from more than 20 cloud providers, building its own file system, container software, and scheduler1

1

. This multi-cloud approach ensures that if one provider experiences technical issues, customer workloads continue running on others2

2

.

"GPUs are scarce, and models are quite large and slow to load," Bernhardsson wrote, explaining the infrastructure challenges Modal solves1

1

. Modal's service for training AI models enables developers to spin up graphics card clusters using relatively simple scripts, with chips exchanging data through RDMA, a network protocol that accelerates traffic by shortening packet routes2

2

.

For inference workloads, Modal offers Endpoints, allowing developers to deploy AI models on hosted infrastructure and make them available through application interfaces2

2

. Enterprises not building custom models can access a library of open-source algorithms provided by Modal2

2

.

Reinforcement Learning and Post-Training on the Roadmap

Modal plans to invest its newly raised capital in product development focused on reinforcement learning and post-training services2

2

. These phases of the AI development workflow improve reasoning capabilities, safety, and cost-efficiency of models being trained, representing natural extensions of Modal's current offerings as AI applications grow more sophisticated.

Bernhardsson described the moment as a "new era of software" where new capabilities require new building blocks4

4

. He emphasized that increasingly capable AI agents will need secure virtual environments where they can execute code and eventually handle longer-running tasks4

4

.

The company, which employs more than 200 staff across New York, San Francisco, and Stockholm, operates in what Accel characterized as the fastest-growing market in technology3

3

. With enterprises deploying generative AI solutions facing inference costs reaching hundreds of millions annually, Modal's infrastructure layer addresses cost, latency, scale, and energy efficiency challenges that determine whether AI services succeed or fail3

3

.