A MESSAGE FROM OUR SPONSOR

Become a more intelligent enterprise

Imagine your entire company working in perfect sync. Strategy, operations, tech, and AI – all connected. At PwC, we design the solutions that can help you get there.

After nine months of collaboration, OpenAI unveiled a new, high-performance chip built with Broadcom and designed for A.I. inference, which it called “the first step in a multi-generation compute platform designed for initial deployment by the end of 2026.” The chip needed a name, of course, and they landed on “Jalapeño,” a title so ridiculous that it almost distracts from the achievement. I wonder if ChatGPT recommended that one.

And now for the main event…

It sounds like a B-plot straight out of HBO’s Silicon Valley: In April, Allbirds, the struggling shoewear company once favored by tech guys, announced that it was pivoting to A.I. and would begin buying up G.P.U.s. In hours, the company’s stock, which had once fallen so far it was in danger of being delisted from the Nasdaq, shot up 600 percent.

But Nadia Carlsten, the new C.E.O. of the company now calling itself Smartbird, is completely serious. Over the past several weeks, Smartbird has sold all its assets and shoe-related I.P. to the American Exchange Group for $40.7 million, formally changed its name, and secured $100 million in convertible debt to finance its transformation—along with an enticing incentive package for Carlsten. “It’s kind of funny how these things come about,” she told me when we connected Tuesday afternoon.

A former product chief at Amazon Web Services with a Ph.D. in engineering, Carlsten had just stepped down as C.E.O. of a small Danish A.I. infrastructure business when Allbirds reached out with an unusual pitch—leverage the company’s public listing to effectuate a sort of reverse merger into selling A.I. services. “It just ended up being a conversation about how committed they are to this and how much freedom I would have for the strategy,” Carlsten said. “For me, it was exciting that it was a bit of a blank slate.”

In many ways, the move recalls the peak of the last tech hype cycle, when a handful of similarly left-for-dead public companies suddenly rebranded themselves as crypto plays. Perhaps most infamous was Long Island Iced Tea, a small beverage-maker that renamed itself Long Blockchain Corp. in 2017, a move that sent its shares soaring some 200 percent, if only momentarily. (The S.E.C. delisted the stock in 2021, shortly before filing charges against three people involved for insider trading.) Naturally, I brought up this history to Carlsten. “Convince me,” I said, “not to be a cynic.”

A MESSAGE FROM OUR SPONSOR

Become a more intelligent enterprise

Imagine your entire company working in perfect sync. Strategy, operations, tech, and AI – all connected. At PwC, we design the solutions that can help you get there.

Carlsten, who has been on the job for only a week, was entirely unfazed. “Sometimes with A.I., there’s a huge mismatch between what people are saying and what’s actually doable,” she admitted. Nevertheless, she said, “our strategy is definitely doable. We have the capital to execute it. I have the experience to do it.” Done correctly, she added, “the origin story doesn’t really matter anymore. The results will speak for themselves.”

Smartbird’s new business plan isn’t all that radical. In short, Carlsten intends to use the company’s new financing to buy and lease dedicated compute to create single-tenant, private A.I. clouds for mid-market clients who require strict data sovereignty—pharma companies, finance, even governments. “One of the key things is we’re not competing head-on with a hyperscaler or large neocloud,” Carlsten said, acknowledging bigger rivals like CoreWeave, Crusoe, Nebius, Cerebras, and Groq that have committed tens of billions to build-outs of their own. Legacy companies like Dell, Nvidia, IBM, and HPE have also been building new business lines around selling hybrid, private A.I. factories-as-a-service. But Carlsten still thinks there is more demand and an unsaturated market.

In particular, Carlsten sees the market opportunity in bespoke services. “What we’re going after is customers that either cannot be on the clouds—and this comes down to sovereignty and security—or they have certain workloads that they don’t want to do on the cloud.” This, she said, is common with large pharmaceutical companies. In general, she said, companies are becoming far more sophisticated in how they’re adopting and integrating A.I., developing dual strategies that include on-premise deployments; the tech, she contended, is no longer an “experiment on the side, and that’s exactly why now is the time to be having those conversations with those customers.”

But to execute in that area requires trust, for starters, a talented team, and lots of electricity—in a space where these things are in relatively short supply. On that first point, Carlsten said, customers are looking for somebody who can execute on these build-outs quickly and reliably. But why they’d put faith in a company that was selling sustainable slip-ons just six months ago is an open question.

Carlsten knows it will require the kind of relationship-building that takes time. She said that many of the customers she talks to aren’t interested in becoming A.I. infrastructure operators themselves, but they want to hire someone to manage the constantly changing tech to their specific needs. “Our pitch,” she said, “and where we are providing usefulness in this market, is we are one service, we’re a managed service, and we’re basically taking off this operational burden while still allowing them to have the full control that they want. It’s not for everybody—but for the people that need it, they really need it, and right now they feel like their only other option is do-it-yourself.”

Because the idea is to work closely “with a few targeted customers,” Smartbird doesn’t need to pour billions into massive data centers like their neocloud contemporaries. And because there are fewer chips at play and no need to acquire massive swaths of real estate while meeting gargantuan electrical requirements, there are fewer barriers to entry. “We don’t have to put up a lot of money into speculative infrastructure upfront,” Carlsten said. “Our strategy is also land-and-expand with a lot of these customers, so a lot of them will start with relatively modest-size clusters—think hundreds and thousands of G.P.U.s rather than 10,000 G.P.U.s—so it’s a little bit easier for us to manage the business and expenditures that way.”

Carlsten added that the company is pivoting with its stock intact, which means quick access to the necessary capital. Rather than “spending six to 12 months raising from venture,” she said, the company can start having conversations with potential customers immediately. Meanwhile, the stock—which peaked at $578 per share in 2021 and now trades at a little over $4—also serves as a useful recruitment tool, and Smartbird’s status as a public company enables it “to partner with other companies and even do some acquisitions,” she added. “I think of it more as a currency, in that it’s going to be helpful for different business structures and different business partnerships.”

Well, maybe. On the contrary, it’s going to be hard for Carlsten to enter the great A.I. talent war with an underwater stock, a challenged brand narrative, and the questionable thesis that Smartbird is going to penetrate a market that the hyperscalers and neoclouds simply haven’t conquered yet. Still, there’s some validity to Carlsten’s point of view: One of the strongest opportunities in A.I. surrounds the underlying infrastructure (picks, shovels, etcetera). Sure, the idea of a private A.I. factory-as-a-service might have a small customer base, but the trend, according to industry sources, is that more and more companies are exploring on-premise deployments not just to achieve better security, but also to escape the pricing uncertainty of cloud-based A.I.

Of course, successfully executing on the strategy is another thing entirely. The elements that make this moment so fruitful for A.I. may also sow a correction. But for what it’s worth, Allbirds already lived through one investment hype cycle and somehow came out the other end. If the tide rolls out, well, they’ve been there before.

That’s all for today. I’ll see you next week.

Ian

Need help? Review our FAQ page or contact us for assistance. For brand partnerships, email ads@puck.news.

You received this email because you signed up to receive emails from Puck, or as part of your Puck account associated with {{customer.email}}. To stop receiving this newsletter and/or manage all your email preferences, click here.

Puck is published by Heat Media LLC. 107 Greenwich St., New York, NY 10006