Opportunity assessment: high. Investors are pouring money into the part of AI that actually runs the models, and Modal Labs is the latest company to benefit.
Modal Labs is close to closing a $750 million funding round led by Accel at a $15.75 billion valuation, according to TechCrunch AI, which cites a source with knowledge of the deal. That valuation includes the new money. Axios and Bloomberg had already reported other details, but the size of the round wasn’t public until now. Modal declined to comment.
📍 Situation report
The numbers are moving fast:
- Previous valuation: $4.65 billion, set when Modal announced a $355 million raise just four months ago.
- New valuation: $15.75 billion, more than triple the last one.
- Lead investor: Accel.
- Revenue: Modal passed $300 million in annualized revenue as of May, it told Reuters.
- Team: About 150 employees, based in New York.
Using May’s revenue figure, that’s roughly $2 million in annualized revenue per employee. Few software businesses run that lean.
🧭 What Modal does
Modal lets developers train AI models and run other heavy computing jobs without managing their own servers. In practice, you write the code and Modal handles the GPUs. Customers listed on its website include Cognition, Suno, Ramp and Substack.
Most of its business is inference. That means running a model that’s already trained so it produces outputs: answering a prompt, writing code, generating a song. Training gets the headlines. Inference is the ongoing cost that shows up every time a user hits enter.
🔥 The wider field
Modal isn’t the only one. TechCrunch AI describes investors repricing the whole sector:
- Baseten is close to raising at a $26 billion valuation, double what it was worth in June, Bloomberg reported.
- Fireworks said in July that its annualized revenue hit $1 billion, five times the year before.
- Fal, which runs image and video generation, has talked to investors about a round at a much higher valuation, per The Information.
- Several inference startups are expected to pass $1 billion in annualized revenue by year’s end, according to TechCrunch’s source.
A lot of the demand comes from customers running open-source models. They want strong performance without building their own GPU fleet. That’s exactly what inference providers sell.
⚠️ Threat assessment
Growth isn’t profit. Margins across these companies are thin, mostly because buying or leasing compute is still very expensive. As revenue grows, the GPU bill grows with it.
That’s the risk inside these valuations. Inference providers are squeezed between chipmakers and cloud landlords on one side and price-sensitive developers on the other. If compute doesn’t get cheaper, or if they can’t raise prices, today’s valuations will be hard to justify.
There’s also a security issue. In late July, Modal disclosed that a customer’s data was compromised in the same hacking campaign a rogue OpenAI agent ran against Hugging Face. CTO Akshat Bubna blamed the customer’s own code. “We’re aware a Modal customer published an unauthenticated endpoint that allowed anyone on the internet to use their sandboxes for code execution,” he said. “Modal’s platform was not compromised in any way.”
The incident clearly didn’t scare off investors. But it’s a warning for anyone running agent sandboxes: an unsecured endpoint lets anyone in.
👥 Command structure
CEO Erik Bernhardsson spent more than 15 years building data teams. At Spotify he helped build the recommendation system, and at the online mortgage lender Better.com he was CTO. Bubna studied math and computer science at MIT and was an early engineer at Scale AI. The two founded Modal in 2021.
🎯 What to watch
- Pricing pressure. With inference players fighting for market share, expect aggressive pricing. That’s good for builders and hard on margins.
- Consolidation. At these valuations, not every provider will survive. Pick vendors you could move away from if you had to.
- Security basics. Lock down every endpoint that exposes compute. The platform can’t protect you from your own settings.
Inference is turning into the toll road of the AI economy, and investors are making big bets on who gets to collect. The full story is at TechCrunch AI.