Allbirds Trades Shoes for AI Infrastructure, and the Market Remembers the Dot-Com Era

Allbirds has decided not to rescue its shoe business but to reinvent itself entirely. The company, once a symbol of "sneakers for the tech elite," has announced a turn toward AI infrastructure, and the market reacted instantly, as if it were the late 1990s again.

A 180-degree turn

On April 15, Allbirds announced that it had signed a $50 million convertible financing agreement with an institutional investor. The company plans to use the money not to relaunch its shoe brand but to buy high-performance GPUs and start a business renting out AI compute. The new name is ready too: after the deal, the company wants to be called NewBird AI. The financing is expected to close in the second quarter of 2026 and requires shareholder approval.

The move follows an even sharper turn. On March 30, Allbirds said it was selling its brand, intellectual property and some assets to American Exchange Group for about $39 million. For a company that was worth more than $4 billion at its peak, this looks less like evolution and more like an actual exit from its former business, along with an attempt to use a public shell to enter a new market.

Under its own plan, NewBird AI wants to:

• buy a fleet of high-performance GPUs;
• lease compute capacity to customers under long-term contracts;
• grow over time into GPU-as-a-Service and an AI-native cloud;
• expand partnerships with operators and enterprise customers.

Why the stock soared

The market's reaction was almost a caricature. On April 15, Allbirds shares closed up 582%, and on April 16 they lost about 36%. The chart looks more like a burst of speculative frenzy than a cool revaluation of a business. The pivot hit the market's nerve: anything tied to AI infrastructure currently draws instant investor attention.

The story does have some logic. Demand for compute to train and run models is indeed enormous, and access to top-tier GPUs remains scarce. Allbirds says outright that it wants to profit from that scarcity: buy hardware and provide access to customers that the spot market or hyperscalers cannot serve quickly and reliably.

But the euphoria is not driven by that alone. Investors saw a familiar plot: a company whose previous business failed attaches itself to the hottest tech trend and gets an instant premium just for a new narrative. That is why the Allbirds story is so often compared to the dot-coms, an era when a single right word in a name or strategy could temporarily rewrite market capitalization.

Where the main risk lies

The most uncomfortable part of this story is that AI infrastructure is not just a fashionable signboard. It is a capital-intensive business that requires access to scarce chips, long-term power and site agreements, cooling, operational expertise and a lot of money. In a market where new data centers are counted in billions of dollars, $50 million looks more like an entry ticket than a serious competitive advantage.

"Back then the magic word was the internet, now it is AI."

That is why the dot-com bubble comparison sounds here not like a joke but like a warning. Allbirds has a clear idea: the GPU shortage is real, and demand for renting compute is real too. But between "we are buying chips" and "we are building a sustainable infrastructure business" lies a huge distance. So far the market has priced in the AI story above all, not the company's proven ability to execute such a plan.

What this means

The Allbirds case shows how overheated everything linked to AI remains. Even a company from a completely different industry can see its stock surge after announcing entry into the compute infrastructure market. For the industry, this is a signal of a real capacity shortage; for investors, it is a reminder that the fashion for technology buzzwords almost always runs ahead of actual execution.