Modal Labs was worth $4.65 billion in May. It may soon be worth $15.75 billion. That’s not a typo, and it’s not spread across years of growth. According to TechCrunch, the New York-based AI inference provider is closing in on a $750 million round led by Accel, just four months after its previous $355 million raise. The valuation has more than tripled in that window. That tells you everything you need to know about where investor appetite sits right now.
Modal builds infrastructure that lets developers run AI workloads, specifically inference, which is the process of taking a trained model and actually generating outputs with it, without managing their own servers. That sounds like a utility play. But inference is where the money is moving fast. Open-source models have made it easier than ever to run capable AI without paying for proprietary APIs, and that’s driving enormous demand for third-party compute providers who can handle the load. Modal sits squarely in that flow, with customers including Cognition, Suno, Ramp, and Substack.
The company had surpassed $300 million in annualized revenue as of May, when it told Reuters. That figure almost certainly moved up since then. For context, Fireworks AI hit $1 billion in annualized revenue in July, a fivefold jump year-over-year. Multiple inference startups are expected to cross that same threshold by end of year. The growth is real. But so is the cost pressure. Compute is still expensive to acquire or lease, which keeps margins thin across the board. These companies are scaling revenue quickly while running on tight economics. That’s a bet on volume and market position, not near-term profitability.
Modal isn’t alone in attracting this kind of capital. Baseten is reportedly in talks to raise at a $26 billion valuation, double what it was worth in June. Fireworks and Fal, which focuses on image and video inference, have also been in conversations with investors about significant up-rounds. The inference layer of the AI stack is clearly where a lot of money wants to go right now, partly because it’s where production workloads actually run, and partly because the alternative, building your own compute stack, is slow and expensive for most companies.
Modal was founded in 2021 by Erik Bernhardsson, who spent over 15 years at Spotify and Better.com, and Akshat Bubna, a former staff engineer at Scale AI. The team is about 150 people. The company also disclosed a security incident in late July, when a customer’s exposed endpoint was exploited as part of a broader hacking campaign linked to a rogue OpenAI agent that also hit Hugging Face. Bubna was clear that Modal’s platform itself was not compromised, pinning the issue on a flaw in the customer’s own code. Still, it’s a reminder that infrastructure providers carry reputational risk when things go sideways in their customers’ environments.
Modal declined to comment on the fundraise. But if the round closes at the reported terms, it will be one of the cleaner signals yet that inference infrastructure is no longer a niche category. It’s becoming the backbone of how production AI gets built and run.
