
On April 15, Allbirds, the sustainable sneaker company that once made Silicon Valley’s favorite shoe, announced it was selling its entire footwear business to American Exchange Group for $39 million and rebranding as “NewBird AI,” a GPU-as-a-Service infrastructure company. The stock surged 582% in a single session. 300 million shares changed hands, about fifteen times normal daily volume. MEME ETFs bought in. By day two, it was already giving back a third of the gain.
The Dutch had tulips. The British had the South Sea Company, a firm that promised to trade with South America and instead traded in its own hype until half of Parliament was bankrupt. Charles Mackay wrote about this in 1841 in Extraordinary Popular Delusions and the Madness of Crowds: people don’t invest in things, they invest in narratives about things, and the narrative only needs to last long enough for you to sell it to the next person. Hyman Minsky formalized this two centuries later — stability breeds complacency, complacency breeds speculation, speculation breeds the moment where everyone realizes they’re holding air. The buzzword is just the delivery mechanism. Sure, the technology changes, but the human wiring that makes someone look at a shoe company rebranding as a GPU lessor and think “I should buy this” has not been patched since the Mesopotamians started speculating on barley futures.
The Origin Story, and Then the Decline
Allbirds was founded in 2015 by an industrial engineer and a former New Zealand soccer player. They pioneered wool and eucalyptus fiber sneakers, marketed them as sustainable, and became a Bay Area status symbol — the shoe you wore to a Series B pitch to signal you cared about the planet but also understood unit economics.
They IPO’d in November 2021 at $15. Closed day one at $28.64. Market cap topped $4 billion. The company never turned a profit. Revenue fell from $298 million in 2022 to a $134 million run rate by 2025. They closed every full-price U.S. store except two outlets. Cash burn hit $68 million a year against $67 million in reserves. The 10-K carried a going concern warning, the Nasdaq sent a delisting notice, and they executed a reverse stock split just to stay listed.
The problem was the same one that kills most DTC darlings: making popular sneakers is hard, holding core customers while chasing broader appeal is harder, and big players can reproduce your look using petroleum, which is cheaper.

The Pivot
Here is the genius AI-pivot plan (per the SEC filing): raise $50 million through a convertible financing facility from an unnamed institutional investor, acquire high-performance GPU hardware, and lease compute capacity to enterprise AI customers. The company will target “customer demand that spot markets and hyperscalers cannot reliably service.” Shareholders vote on the asset sale, the financing, and the name change on May 18.
The filing also asks shareholders to approve removing “references to the company being operated for the environmental conservation public benefit.” So the shoe company that sold sustainability in every step is now asking permission to literally stop caring about the environment so it can lease GPUs. The company needed a new name because it’s selling the old one.
Here’s the scale problem. CoreWeave, an actual neocloud company that pivoted from crypto mining to GPU infrastructure in 2019 and now trades at a $74 billion valuation, plans to spend $30 to $35 billion on capex this year. That’s 635 times Allbirds’ entire war chest. Nebius, a smaller player, is budgeting $16 to $20 billion. A $50 million investment is a drop in the bucket, so they are really going for the Hail Mary.
Other companies that pivoted to AI at least started with adjacent infrastructure. CoreWeave was already running GPU clusters for crypto mining. The technical DNA carried over. Allbirds’ expertise is in merino wool supply chains and eucalyptus fiber sourcing, and none of that transfers to GPU procurement.
We’veHeardThisStoryBefore.com
In 1999 and 2000, companies appended “.com” to their names and watched stock prices double overnight. Not a single change in operations, revenue, or customers. The Nasdaq hit a P/E ratio of 200. It fell 78% by October 2002.
In December 2017, Long Island Iced Tea, a beverage company with $1.5 million in quarterly revenue, rebranded as Long Blockchain Corp. The stock surged 380%. The company never did anything with blockchain, and the SEC delisted it and filed insider trading charges.
Last summer, a Florida toy maker, a Japanese hotel operator, a nail-salon chain, and an electric-bike company all rebranded as crypto-treasury companies to buy bitcoin. Most of those tokens have since collapsed. A karaoke machine company called Singing Machine renamed itself Algorhythm Holdings and claimed to have AI technology for trucking efficiency. It tanked trucking stocks on the news.
NewBird AI is the latest entry.

What Happens Next
The French economist André Orléan has this concept he calls “mimetic rationality” — the idea that in financial markets, the rational move is often to imitate what you think others are about to do, regardless of whether the underlying thesis makes sense. You’re betting on the consensus about reality, but not reality itself. And if enough people bet on the consensus, the consensus becomes self-fulfilling — until it doesn’t.
We saw this play out in real time with the Iran situation a few weeks ago. Prediction markets spiked on U.S. military action before any public announcement. Oil futures moved. Defense stocks moved. Someone knew, or enough people believed that someone knew, and the distinction stopped mattering. The mechanism of betting on what other people are about to believe was identical to what happened with Allbirds. The only difference is that the Iran trades were anchored to something real. NewBird AI is anchored to a press release.
Keynes called this the beauty contest problem in 1936. You don’t pick the face you think is prettiest. You pick the face you think everyone else will pick. Allbirds at $17 was never a bet on GPU infrastructure. It was a bet that enough people would see the 582% headline, pile in for another 20%, and you’d be out before the music stopped. Some people timed it right, bu most won’t. The stock was already down 31% by day two, and the shareholders haven’t even voted yet.
The market doesn’t need to believe a story to trade it. It just needs to believe that someone else will believe it for slightly longer.
See ya, folks. Stay curious.
—J&E