How do you save a failing business? Say it with me now: "Pivot to AI."

In this frothy boom time of financial exuberance, the intoxicating elixir of AI riches has charmed many an executive, and attracted hungry investors looking for the next big thing.

The early days of the generative AI explosion saw logical pivots. Cryptomining firms, facing increasing headwinds in Bitcoin prices and the death of the blockchain dream, quickly shifted to AI workloads. With much of the same core challenges, and crucially, access to power, the pivot has wholly proved successful.

But now, several years into the golden age of data centers, we're starting to see more confusing pivots. First, you had former politicians get in on data center land speculation. Now, you've got shoe brands.

This week, much of the industry has been talking about the conversion of shoe brand AllBirds to that of potential compute business NewBird AI.

Once valued at $4 billion, the eco-brand fell out of fashion and lost 99 percent of its share price since 2021. Now, the owners of the company have said that they will sell the brand to American Exchange Company for $39m, and instead develop an AI compute business.

This pivot, which is still in the early investigatory stage, is backed by $50m in funding the company says it will get from an undisclosed institutional investor later this year.

"NewBird AI expects to use initial capital from the Facility to acquire high-performance GPU assets, which will be deployed to serve customers requiring dedicated access to AI compute capacity," the company said in press release.

"NewBird AI’s long-term vision is to become a fully integrated GPU-as-a-Service (GPUaaS) and AI-native cloud solutions provider. Over time, the Company intends to grow its neocloud platform by expanding its compute and service offerings, deepening partnerships with operators and customers, and evaluating strategic M&A opportunities."

The company, best known in recent years for losing sales, has no experience in the data center space. It has no experience in the AI space, the compute space, or even in the B2B space. It has no special access to land, power, talent, or technology.

Even its war chest in the tens of millions is not that inspiring in a sector where billions are raised and spent regularly. A single data center can easily blow past a couple hundred million dollars, and the number just keeps on rising.

Rather than buying a whole data center, NewBird looks to be set to just buy GPUs or equivalent and place them in another's data center. "The Company will initially seek to acquire high-performance, low-latency AI compute hardware and provide access under long-term lease arrangements, meeting customer demand that spot markets and hyperscalers are unable to reliably service."

While we're talking less than 200 racks filled with Nvidia's latest gear, finding markets that hyperscalers or neoclouds haven't saturated and yet have modern liquid-cooled data center space, available quickly and at a low cost, is quite a big ask (all while NewBird will need to develop its own software stack).

In its own press release, the company notes that "North American data center vacancy rates have reached historic lows, and market-wide compute capacity coming online through mid-2026 is already fully committed." This, on the one hand, sounds great for investing - but in making more space for capacity, not in just joining the queue of those looking for space. NewBird's answer to the leasing crunch appears to be to try to lease.

In its SEC filing, the company said that its entry into the market "may include, without limitation, sales, leases, sale/lease-back transactions and other monetization structures, are anticipated to be capital-efficient and scalable, with customers bearing substantially all operating, maintenance, and infrastructure costs, allowing us to grow our asset base and revenue without a commensurate increase in operating overhead."

That approach sounds good in principle, should it be able to find space to lease, but the low asset business is usually expected to bring something to the table.

After a period of freewheeling excess, where companies with limited history and lofty dreams were able to sign major data center deals, the industry is starting to coalesce around experience and proven deliverability.

With major IPOs around the corner, and an increasing eye on reigning in costs and getting compute live, we have seen deals be walked back or stripped down.

A new entrant with limited funds, already at the back of the queue on long-lead orders like GPUs (or data center gear, should they go further down the chain), has little to offer potential customers as way of proof that they can deliver mission critical compute on time and on cost.

Then there's the fact that, by the time they could come to market, the industry could look very different. In a bullish view of continued growth, NewBird will face the fact that all the companies that began pivoting or expanding several years ago have projects well underway - plus, assuming scaling laws hold, $50m will get you even less far.

As for the more negative view, where this is a bubble that bursts, then NewBird will launch just as the market contracts and there is a glut of in-market cheap compute.

Even the biggest proponents of perpetual growth admit the reality could be somewhere in the middle, with OpenAI's Sam Altman admitting that "someone’s gonna get burned” due to AI overinvestment.

Now, NewBird may thread the needle (or should I say eyelet?), it may find a data center in need of $50m at the right time, it may even - somehow - become the next CoreWeave.

Companies should not be discouraged from reinvention - from Nintendo moving on from playing cards, American Express surviving being a mail carrier, or Play-Doh dropping being a cleaning business.

Admittedly, they usually had some industry or technological reason for their pivot, and were not just chasing the latest hype wave, but casting off unprofitable ventures for pastures new is keen to corporate survival.

If NewBird's team believe they have a pathway from shoes to AI compute, more power to them, I wish them luck.

What is concerning, however, is the market's willingness to believe this pathway exists at this stage. Given what we know, there is currently no reason to believe that the company is worth more than the $39m in assets on hand and $50m on the way - it has announced no contracts, no plans, and it hasn't even confirmed this idea has moved beyond the exploratory phase.

Nonetheless, shares in the public company rose some 582 percent. Its market cap hit $148m in days.

These are not signs of a rational market, nor of one that understands data center or AI compute development. Retail investors risk getting burned on speculative pivots from companies with no clear AI compute strategy, while many of the biggest data center and AI firms remain locked behind private markets.

Unhealthy speculation risks turning the sector into a bubble, if it isn't one already.

While I wish NewBird well, I am left waiting for the other shoe to drop.