Smartbird, the AI infrastructure company that was Allbirds two months ago, named Nadia Carlsten chief executive on June 17, 2026, handing her a $700,000 salary, roughly $9 million in stock, and a balance sheet flush from a $43 million shoe-business sale and a $100 million stock-market raise. She has no team, no office, and no customers. She is hiring all three.
Carlsten, a former AWS executive with an engineering PhD, ran European compute firm DCAI before taking the Smartbird seat. She is starting from Amsterdam and said the first hire on her list is a leader for infrastructure operations. The shoe business officially wound down the day before she began.
The pivot itself, announced in April, read as a meme-stock maneuver: a struggling direct-to-consumer brand swapping a tired category for the hottest one in the market. It worked. The cash is real, the ticker is intact, and now Carlsten has to convert a stunt into an operating company.
Key facts
- 01Allbirds sold its shoe business for $43M and raised another $100M from the stock market before rebranding as Smartbird.
- 02Nadia Carlsten, a former AWS executive and ex-CEO of European compute company DCAI, started as Smartbird CEO on June 17, 2026.
- 03Carlsten's package includes a $700,000 annual salary and roughly $9M in stock awards.
- 04Smartbird targets customers needing hundreds to thousands of chips per deployment, not hyperscaler-scale clusters.
- 05Carlsten aims to have compute clusters deployed for several customers by the end of 2026.
Smartbird's pitch is narrower than the neocloud boom that fueled valuations at GPU brokers and inference startups. Rather than arbitraging chip prices against GPU-time or token revenue, Carlsten wants to sell managed, single-tenant compute to customers who need direct control of the servers running their models — typically for data-sovereignty, regulatory, or business-model reasons. At DCAI she worked with Novo Nordisk and other European firms in pharmaceuticals, energy, finance, and the public sector.
Customer deployments will be measured in hundreds to thousands of chips, not the hyperscale clusters that anchor OpenAI or the frontier labs. Carlsten says her buyers prize control over scale.
That positioning puts Smartbird up against Hewlett Packard's single-tenant managed AI compute service and Equinix's data-center business, rather than against AWS or a neocloud. It is a real market, but a slower-growing one than the hyperscale layer, where unit expansion compounds quarter over quarter. Carlsten expects to have clusters deployed for several customers by the end of 2026 — a deliberately modest first-year target.
The contrast with the rest of the AI infrastructure market is sharp. General Compute came out of stealth last month touting a $300 billion chip order, the kind of headline number that has come to define this cycle. Smartbird is going the other direction, and Carlsten argues her customers don't need that volume to make the economics work.
Pricing is the obvious soft spot. Public cloud providers run chip utilization around the clock to drive per-hour costs down, and a single-tenant managed offering cannot match that on raw cost-per-token. Carlsten's bet is that customers with specialized workflows will extract more value from dedicated servers than from cheaper shared ones — a thesis that holds in regulated industries and breaks in commodity inference.
There is also a governance footnote. Allbirds was a public benefit corporation, a structure intended to lock in its sustainability commitments. That status went away with the AI pivot, a reminder that PBC charters bend when boards want them to. OpenAI, also a PBC, has leaned on the structure to enshrine its AI-safety mission; the Allbirds case suggests the lock is softer than the marketing implies.
What Smartbird actually becomes depends on whether Carlsten can recruit an infrastructure leadership team in a market where every neocloud, hyperscaler, and frontier lab is hiring the same people. She is starting from a standstill against competitors that have been deploying clusters for two years. The cash gives her runway; the calendar does not.
The Smartbird story is a useful tell on where the AI infrastructure market is heading. The headline-grabbing layer — billion-dollar chip orders, gigawatt buildouts, orbital data centers — is consolidating around a handful of players with hyperscaler economics. The opportunity Carlsten is chasing, managed sovereign compute for regulated buyers, is genuine but slower, more service-heavy, and structurally capped by how many pharmaceutical, energy, and public-sector customers actually want to run their own clusters. A $143 million war chest buys time to find out, but the test isn't the pivot — it's whether a sole-founder startup with a public-company balance sheet can ship infrastructure faster than HPE and Equinix already do.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
