• D.C.
  • BXL
  • Lagos
  • Riyadh
  • Beijing
  • SG
  • D.C.
  • BXL
  • Lagos
Semafor Logo
  • Riyadh
  • Beijing
  • SG


‘This stuff needs to become real’: Snowflake’s Sridhar Ramaswamy on getting returns from AI

May 2, 2025, 5:16am EDT
ceobusinessNorth America
Sridhar Ramaswamy
Kris Tripplaar/Semafor
PostEmailWhatsapp
Title icon

The Signal Interview

When the data warehousing company Snowflake went public in September 2020, investor excitement for its cloud-driven growth story propelled its first-day valuation to $70 billion — more than 100 times its annualized revenues at the time. That made it the biggest software IPO on record, but the enthusiasm didn’t last.

Snowflake, which helps companies store and analyze large volumes of data, had come to fame with a technical accomplishment that allowed clients to scale up their computing power without buying more storage. Yet by February last year, its shares had fallen below their 2020 price, as it became clear that AI was ushering in a period of disruptive technological change. On the day it appointed Google veteran Sridhar Ramaswamy to succeed Frank Slootman as CEO, a disappointing financial update sent its stock down another 20%.

“I told my team that they had to earn their way back to a higher stock price, that no one was going to give it” to them, Ramaswamy says.

Rethinking the company’s product lineup was the obvious priority for the new CEO, who earlier in his career ran Google’s entire advertising business, overseeing everything from search to shopping. But just as important, he says, was figuring out Snowflake’s future: “How do we take new products to market in a company that has been enormously successful with an old product?”

AD

Companies can go from being hungry for success to just expecting success for all that they do, Ramaswamy warns. “Part of what I’ve tried to drill into every Snowflake person’s way of thinking is the need to strive for success, the need to be excellent on an ongoing basis, the need to be gripped to capture opportunities that are there. Because, who knows? Your president might announce tariffs that shut the world down.”

‘No-regrets’ moves to turn crisis to opportunity

Snowflake supports AI models from OpenAI to DeepSeek, and the AI spending cycle looks likely to shelter it from some of the pressures other companies will face in a trade war. “Our core product is very strong,” Ramaswamy says, pointing to its high recurring revenues, but he still expects an impact if current economic uncertainties persist. “Any reasonable CFO is going to say, ‘I don’t really want to commit large dollars over the next three years,’” he predicts.

Now, he is using that prospect to inject a sense of urgency into the company he joined two years ago when Snowflake bought Neeva, the privacy-focused search company he launched after leaving Google.

AD

“Whenever there is a crisis, there are no-regrets moves that you should take. It falls a little bit into the bucket of, ‘don’t let any crisis go to waste,‘” Ramaswamy observes. He is looking carefully at hiring, spending and performance management, and points to his appointment of a chief people officer as a way to put more “back to basics” HR rigor into the 7,500-employee business.

“You cannot talk about performance if you’re not willing to have a conversation about what excellence means,” he notes, saying leaders “need to have conversations with people about expectations and what happens if they don’t meet expectations.”

Turning AI ‘shopping’ into tangible results

That includes not just telling staffers what they need to do to improve their performance, but helping them do so, and Ramaswamy has focused on helping Snowflake’s 3,000 salespeople adapt to the fast-changing technology backdrop.

AD

“I have a lot of sympathy for our sales folks,” he says. “They have to go pitch a product in… an environment that they don’t always understand, talking to experts who sometimes know way more about these areas than they do.” It’s one thing for Snowflake to add AI capabilities to a product; it’s another for a salesperson to be able to speak eloquently about what that product can do for a customer, he notes. “You can’t get 3,000 people educated on a new thing overnight. It’s very hard, especially when they’re uncomfortable with it.”

His solution has been to create a dedicated team of AI experts that can help the wider sales force with early pitches to clients. He also made a point of meeting each of Snowflake’s 100 largest customers in his first five months.

Despite the complexities he sees in this technology shift, Ramaswamy’s watchword is that his people need to “demystify” AI. “Our motto for AI on Snowflake is that it’s going to be easy for you to see what value you can get as a customer,” he says. So he will tell clients that he’s happy to send a team over to run a half-day hackathon to show what they can do by applying AI to their data.

Forcing teams to show tangible, relevant examples of what AI can do is “by far the single biggest thing that CEOs are not doing,” he says. “It’s easy to talk highfalutin strategy. This stuff needs to become real.”

Too many executives are still “shopping for AI,” Ramaswamy says, likening what he sees to a consumer enjoying the fleeting dopamine hit of an online purchase before wondering whether it will need to be returned. The technology cannot be abstract for executives, he says, adding that he subscribes to a host of different AI products so he can better understand their capabilities.

“You begin to get more and more of a feel for what’s the utility that you can expect from AI, so just immerse yourself. It’s a tool. It’s a means to an end. It doesn’t automatically create business value,” he says, adding that he expects enterprise adoption of AI to be “slow and steady,” and driven by simple applications that are able to show real value creation.

Skillset lessons from a startup setback

Ramaswamy, a professorial graduate of the Indian Institute of Technology Madras and Brown University, earned the respect of many peers in Silicon Valley for launching Neeva, a search engine which relied on subscriptions rather than advertising to protect users’ privacy. But the admiration it earned was not matched by commercial success, and he shut the product down even before the sale to Snowflake went through.

What Snowflake was buying was Neeva’s people, and Ramaswamy told them that “getting acquired like that meant that we really had to embrace the aspirations and the North Stars of our new home.” He had seen many of Google’s acquisitions “go horribly wrong” because of a lack of alignment between the incoming team and their new employer, he recalls. Once you shut down your old product, “you have to embrace the new place.”

Even so, he admits that his heart “still skips a beat” when he reads about a related product, or when he bumps into a former Neeva subscriber, like Uber CEO Dara Khosrowshahi.

“The world of startups is tough, man,” he admits, adding that he has used the experience to stress the need for resilience to his children. “I tell them that it’s very easy to just deal with success. You’ve got to deal with failure and internalize that a lot more than your success.”

Running a startup was an amazing experience, Ramaswamy says, but “in retrospect, I think Neeva was a bad fit for me, more because I dropped too much of what I was good at.”

He learned a lot at Google about how to run a business, recruit people, and handle product strategy, policy, and business development, he says. Now, when people consult him about changing jobs, he tells them to make a list of their skills. It can take years to get good at something, he says, and people should think hard before making any move in which they would set those skills aside.

“You’re not going to like being mediocre at a whole bunch of other things,” he says. In the new role, you will find “incredibly motivated people that have been working on them for 20 years. You will compare yourself to them and feel pretty miserable.”

If Ramaswamy’s role at Snowflake feels like a return to something he knows he can succeed at, he is not dropping his sense of urgency. In 2003, the year that he arrived at Google, the search giant leased the Mountain View campus once occupied by Silicon Graphics International, a feted computer graphics pioneer that ended up going bankrupt. Ramaswamy, who warned at Semafor’s World Economy Summit that “search is under attack” from AI, used the old SGI buildings like a corporate memento mori, to remind colleagues of the perils great companies face if they go astray.

“We’re in Silicon Valley, right? This is the land of Only the Paranoid Survive,” he says, in reference to the 1988 book by Andy Grove, who ran Intel before the chipmaker lost its way. That line, Ramaswamy reminds people, was “said by a company that failed to be paranoid.”

AD
AD