- VAST’s CEO told Fierce securing the new funding wasn't a matter of survival but growing its credibility with enterprise customers
- The company is adding new capabilities to its AI Operating System as the technology's needs change
- VAST will be ready for an IPO later this year. But it hasn't decided yet if it'll pull the trigger
It’s been a wild ride for VAST Data over the decade. From a scrappy data storage startup, the company grew to hit unicorn status in 2020 and by 2023 was valued at $9.1 billion. Now, the company just closed a $1 billion fundraising round valuing it at $30 billion. But to hear CEO Renen Hallak tell it, VAST doesn’t need the money.
“We’re generating more than $100 million in cash every quarter now,” he told Fierce. “So, what will we use the money for? That’s a good question.”
Then why fundraise at all? Well, Hallak said there are two key reasons. First, fundraising rounds provide an opportunity for the company to partner with new investors.
Second, Hallak said VAST wants to be equipped with a strong balance sheet because “we’re doing very large projects now and our customers are making big bets on us.” Since VAST isn’t a public company, a fundraising round like the one it just pulled off is one way for it to signal to customers that “we’re here forever.”
Asked whether VAST is planning to pursue an IPO, Hallak said the company expects to be prepared for one later this year – but it will wait for the right moment to file.
“We don’t need the money, so there’s no rush to go public…But we’re also selling more and more into enterprises and they like to buy from public companies. That gives them confidence,” he said. “So, the question that’s interesting in my mind is ‘can we grow faster as a private company, or will we be able to grow faster as a public company?’ Once we convince ourselves that the answer is the latter, that, I think, is a good reason to pull the trigger.”
Growing the new AI stack
Originally focused on storage, VAST has more recently been focused on building a new software stack for AI in the form of its AI Operating System. The OS sits above the compute layer (think GPUs, CPUs, etc.) and below the model and application layer, and it is effectively designed to function as both the brains and circulatory system for the data that is AI’s lifeblood.
It is this OS that has tripled VAST’s worth in the eyes of investors over the last two years.
“What’s become apparent is that the old stack is not good enough for these new AI applications and models,” he explained. “As people start to put AI into production…they realize that they need to move all of these new applications to the new stack and they need to move all of their data to the new stack so that it becomes accessible to these new AI workloads.”
But as Hallak explained, AI’s needs are changing. Two years ago, the primary challenge VAST was trying to solve was creating a big enough data pipeline for model training. But inference will require more distributed infrastructure with low latency and high resilience. Agents, meanwhile, will require more observability and policy governance. The complete reinforcement learning loop will require all of these components. These are the capabilities VAST is building, Hallak said.
In addition to investing in product development to expand its stack, Hallak said VAST will beef up its go-to-market channels and expand its presence geographically. Back to his original point, though, “All of those things are happening, but the way the business is built we’re doing those without burning cash. So, I can’t say with a straight face that’s why we raised the money.”
Telco opportunity
VAST is spending plenty of time wooing enterprises to its platform. But they're not the only ones using its software. Hallak said the company also has a lot of telco customers. That might surprise you given he stated at VAST’s recent customer conference that telcos have been “irrelevant for a long time,” having ceded value to the apps and services that run on their networks.
Fierce pressed the CEO on whether he sees an opportunity for telcos in the AI era. His response was a resounding “yes.” Telcos, he noted, have both power and space at edge locations, as well as network assets to carry data.
“I think they are sitting on a goldmine and they just need to realize they’re sitting on that goldmine,” he said. “They have all of the physical and logical building blocks for them to be major players in this AI revolution, and they just need to make a decision that they want to be.”
Fierce noted that several telcos have tried to make the edge play in the past – including Verizon, MobileEdgeX and VaporIO – but failed due to the lack of a killer use case. As we noted in a recent story, at this point, it seems that use cases are either so critical they need to run on-device or on-prem, or are insensitive enough to latency that processing can occur at a centralized data center.
Hallak responded that previous attempts were ahead of their time and that the proliferation of inference will change the game entirely.
“I don’t think that architecture scales to the size that we’re going to need over the next few years…All of that data cannot make its way into a central data center. It will live at the edge,” he concluded. “I think there’s no way around this idea.”