The archive · Developer & Business Tools · Product decision · 2021–2026
Modal bets devs never manage GPU infra again; $7M seed to a $2.5B valuation in 5 years
A serverless cloud that runs code on GPUs with zero infrastructure; Modal went from a 2021 experiment to ~$50M ARR and a ~$2.5B valuation by 2026.
Modal Labs
What the business is
Serverless cloud that runs AI and data workloads: code first, GPUs autoscaled, billed only for the compute actually used.
Starting capital：$7M seed raised in early 2022
How it started
Erik Bernhardsson — builder of Spotify's music-recommendation system, later CTO of Better.com — started Modal Labs in 2021, convinced the fragmented, manual data-engineering toolchain was the wrong shape for large organizations. By early 2022 he had a $7M seed round and a small team.
What happened
In October 2023 Modal raised a $16M Series A led by Redpoint (with Lux Capital, Amplify Partners and Definition Capital), bringing total funding to $23M. Its Rust-built container system scaled to hundreds of GPUs in seconds; Substack and Ramp were named customers. The generative-AI boom then turned a generic serverless-cloud bet into an inference platform.
How it ended up
$87M Series B at a $1.1B valuation announced around September 2025; in February 2026 TechCrunch reported General Catalyst was in talks to lead a round at roughly $2.5B, with annualized revenue around $50M.
Background
Modal is a serverless cloud for AI and data workloads: developers declare code and hardware specs in a single file, and Modal's Rust-built container system scales to hundreds of GPUs in seconds, billing only for the compute actually used. Erik Bernhardsson, who built Spotify's music-recommendation system and later became CTO of Better.com, started the company in 2021 convinced the fragmented, manual data-engineering toolchain was the wrong shape for large organizations.
The bet did not need the market to change to be right — but the market changed anyway. When Modal raised a $16M Series A led by Redpoint in October 2023, generative AI was pushing model deployment onto GPUs, and early customers included Substack and Ramp. The pitch — run code in the cloud without managing infrastructure — turned out to be precisely what AI inference developers wanted.
By September 2025 Modal had raised an $87M Series B at a $1.1B valuation. In February 2026 TechCrunch reported that General Catalyst was in talks to lead a round near $2.5B, with annualized revenue around $50M. The same wave that pushed Baseten, Fireworks AI and the vLLM spinout Inferact to billion-dollar-plus valuations found Modal already standing in front of developers.
What has to be true
- The wedge was personal frustration: Bernhardsson had lived the pain at Spotify and Better.com, so the product matched a real job to be done.
- Serverless plus pay-per-use was a genuine category fit, not a gimmick, for teams whose GPU bills were exploding.
- The platform went general availability at the moment model deployment became every team's problem.
- Code-first developer experience spread the product through engineers rather than a sales team.
What can be applied
Build for the shift about to happen, not the market as it stands: a generic serverless data platform looked niche in 2021, then every AI team needed exactly it.
Aftermath
The February 2026 report called the ~$2.5B talks early and noted CEO Erik Bernhardsson denied active fundraising; General Catalyst had not commented. Modal competes in an inference market where Baseten raised $300M at a $5B valuation and Fireworks AI secured $250M at $4B within months, per the same TechCrunch coverage. As of September 2026 the company continues to operate and scale its self-serve developer platform.
Sources
- AI inference startup Modal Labs in talks to raise at $2.5B valuation, sources say
- Modal Labs lands $16M to abstract away big data workload infrastructure
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card