A data infrastructure company just hit a $30 billion valuation in a market that supposedly doesn't care about infrastructure anymore.
The Summary
- Vast Data raised $1 billion in Series F funding, tripling its valuation from roughly $10 billion to $30 billion with Nvidia backing
- The round includes a secondary offering, signaling early investor liquidity ahead of an IPO the company is actively preparing for
- At $30 billion, Vast is now valued higher than most legacy storage companies despite being founded in 2016, reflecting the premium investors place on AI-era infrastructure
The Signal
Vast Data makes software that unifies data storage, a decidedly unsexy category that just became the hottest place in enterprise tech. The company's core product consolidates what used to require separate systems: file storage, object storage, and database functions all running on a single platform. In the AI training era, that matters. A lot.
The math here tells you everything. Vast tripled its valuation in a market where most late-stage companies are struggling to hold flat. Nvidia doesn't casually back infrastructure plays. They back companies building the roads their chips will drive on. Vast's storage architecture is optimized for the kind of massive parallel data access that AI model training demands.
"The company is readying itself for an IPO." — Renen Hallak, Vast Data CEO
The secondary component of this raise is the real tell. Early investors are getting liquidity before the public offering, which means two things: Vast has enough demand to let insiders sell without spooking new money, and the company is confident enough in its IPO trajectory to start the pre-public liquidity dance. That's not a move you make unless your revenue numbers can survive public market scrutiny.
Here's what separates Vast from the last generation of storage companies: they built for GPU clusters from day one, not as an afterthought. Traditional storage systems bottleneck when you throw hundreds of GPUs at them simultaneously. Vast's software architecture assumes that's the baseline use case. Every major AI lab and cloud provider is now solving the same problem: how do you feed data to thousands of chips fast enough that your $500 million training run doesn't stall waiting for storage I/O?
Key market context:
- Legacy storage companies trade at 2-4x revenue multiples
- AI infrastructure companies are commanding 15-25x multiples
- Vast positioned itself in the second category while solving problems in the first
The IPO timing matters. We're in a narrow window where public markets still have appetite for AI infrastructure stories, but that won't last forever. Vast is moving while the getting is good. The company hasn't disclosed revenue, but at a $30 billion valuation, investors are pricing in a path to several billion in annual recurring revenue. That's achievable if every major AI training operation becomes a customer, which is exactly the bet Nvidia is making.
The Implication
Watch who else announces storage rounds in the next quarter. Vast's valuation reset just moved the goalposts for every competitor. If you're building anything that touches AI infrastructure, the money is here now. But the clock is ticking. By late 2027, investors will want to see actual margins, not just revenue growth.
For companies buying infrastructure: your storage architecture will be a bigger determinant of AI training costs than most people realize. The difference between good and great storage can be the difference between a $10 million and $50 million training run. Vast's valuation suggests customers are finally doing that math.