Modal reportedly nears a $750M round at a $15.75B post-money valuation

Modal sells serverless cloud infrastructure to developers building AI applications. Customers use it to serve trained models, run training and batch jobs, or give coding agents isolated sandboxes without operating their own GPU fleets. TechCrunch reports the company is nearing a $750 million round led by Accel at a $15.75 billion post-money valuation. Modal declined to comment, so the financing remains unconfirmed and unfinished.

The workflow starts in a customer’s application code: developers specify the computing resources a job needs, and Modal packages that code in a container, places it on pooled cloud capacity and adds or removes containers as demand changes. For an inference workload—the step where a trained model produces answers or media—that removes much of the server provisioning from the customer’s engineering team.

Modal’s website names Cognition, Suno, Ramp and Substack among its customers. The platform spans low-latency model serving, fine-tuning, batch processing, notebooks and sandboxed code, so the same customer can use it for experimentation and a live product. Modal charges by the second for requested compute; paid workspaces can also carry subscriptions, and enterprise customers can make usage commitments.

TechCrunch said Axios and Bloomberg had previously reported other deal details, while the $750 million amount was new. The same report says the proposed price is post-money: it includes the capital being raised rather than valuing Modal at $15.75 billion before the investment.

That price would mark a rapid repricing. TechCrunch says Modal’s previous $355 million financing, announced four months earlier, valued it at $4.65 billion; the new terms would more than triple the valuation. It also cited Modal’s May statement to Reuters that annualized revenue had surpassed $300 million, giving the step-up a business-growth marker even though annualized revenue is not the same as recognized full-year revenue.

The reported fundraise also lands in a capital-intensive layer of the AI stack. TechCrunch reported that inference providers can have thin margins because leasing or acquiring compute remains expensive. Modal’s usage billing lets revenue rise with customer workloads, but serving that demand also consumes more of the GPU capacity underneath the service.