Smartbird CEO explains how the former Allbirds is rebuilding for AI infrastructure

Smartbird CEO Nadia Carlsten is making the case that a former footwear company can become a credible AI infrastructure supplier. In a conversation published by Business Insider on October 30, she discussed rebuilding the company formerly known as Allbirds—and persuading customers and investors that its change of business is not a gimmick.

Speaking with Business Insider chief correspondent Ben Bergman as part of The Long Play, Carlsten framed the shift as reinvention. “I think every company needs to reinvent itself every now and again,” she said.

The business behind the new name

The strategy predates that interview. In an August 19 shareholder letter, Carlsten wrote that the former Allbirds had sold its footwear assets, changed its name and brought in new leadership. She described the company’s public-market access and existing corporate foundation as starting advantages, while warning that a listing alone would not create a defensible business.

Her market argument centers on a distinction between companies building AI models and companies using them. She expects more demand from inference—the everyday running of AI in products, workflows and agents. In her view, those businesses will need infrastructure shaped around different requirements from the enormous systems used to train frontier models.

Smartbird’s intended answer is dedicated infrastructure delivered through a managed service. Customers describe what they need AI to do; Smartbird designs, buys, deploys and operates the underlying computing systems. The pitch is that customers can get infrastructure tailored to their needs without building an internal team capable of running it themselves.

Carlsten’s examples include financial institutions handling sensitive data, governments with requirements about where data is kept, and businesses facing rising costs as AI use grows. The proposed service can include systems at a customer’s own premises. Performance, cost, security and control are the requirements she says would shape each deployment.

That is not a claim that every enterprise should leave shared cloud services. Carlsten explicitly wrote that shared infrastructure would remain sensible for many workloads. Smartbird is targeting organizations whose needs do not fit a standardized platform, rather than presenting dedicated computing as the default answer for everyone adopting AI.

Being right about the growth of AI will not make every infrastructure investment a good one.

Nadia Carlsten, August 19 shareholder letter

From reinvention to deployment

The staffing work discussed in the new interview has a specific purpose in that plan. In August, Carlsten said she was building a small team with deep technical experience in AI infrastructure. She argued that expertise gained from deploying and operating systems—not simply owning more graphics processors—would become Smartbird’s lasting advantage.

Her proposed operating model also leaves room to change hardware suppliers. The letter says Smartbird is not tied to one technology or provider and intends to adopt specialized technologies as they emerge. Carlsten’s argument is that understanding a customer’s workload should come before deciding how many processors to supply.

For the next stage, Carlsten has named operational measures she intends to share as Smartbird grows. These concern whether customer interest turns into usable infrastructure, rather than the size of the company’s hardware purchases alone:

• Customer demand: the quality of the demand Smartbird attracts.
• Capacity: how much computing infrastructure is contracted and how much is deployed.
• Deployment speed: how quickly Smartbird puts that infrastructure in place.