AI IS WHITE HOT… IS THIS NEXT? WHITEFIBER (WYFI – NASDAQ)

QUICK FACTS

Trading Symbols: WYFI
Share Price Today: $45
Shares Outstanding: 38.4 million
Market Capitalization: $1,509 million
Cash: $76 million
Debt: $222 million
Enterprise Value: $1,656 million

BETTING THAT LONG-TERM FUNDAMENTALS DRIVE SHORT-TERM MOMENTUM

Every so often the market hands you a stock where the story is better than the numbers and the tape is going to be driven by the story.

That’s Whitefiber (WYFI -NASDAQ) in a nutshell.

Whitefiber is a sub-scale AI datacenter landlord. They have one large good-looking contract that will drive growth for the next 2 years. They have a controlling shareholder that is pulling the strings and providing cash. And a share price that is way, way over-bought, but could go a lot higher once that has been worked off.

Whitefiber owns one valuable resource that AI customers are desperate for. Not GPUs. Not memory. They own ready-to-go power and the land to fill it with server racks.

In our tipsy-topsy world of AI, that’s a moat. At least for a while.

Datacenters have become the bottleneck.

Cerebras (CBRS – NASDAQ), which offers a faster GPU for inference, was asked on their call a couple days ago what was the constraint to their growth.

CEO Andrew Feldman replied: “Demand is not the constraint. Supply is not the constraint. The constraint is data centers”.

The amount of datacenter capacity expected to be needed over the next 5 years is something approaching astronomical. And it is still not enough.

Growth in the AI workload segment is expected to compound at 30% growth from now until 2030.

While much of the growth will fall on Amazon (AMZN – NASDAQ), Google (GOOGL – NASDAQ), Microsoft (MSFT – NASDAQ) and the like, there is plenty of room for a well placed small-cap to go along for the ride.

That is exactly where Whitefiber fits in.

DRIVING GROWTH OVER THE NEXT TWO YEARS

Whitefiber has 2 small datacenters operating right now, both in Montreal. They have a third small datacenter in Montreal that is looking for a tenant. A fourth, far larger, datacenter in North Carolina is planned to be commercial operational by the middle of this year.

Finally, there is a fifth facility planned for Paris that was just announced.

Whitefiber also has leased datacenter capacity of 6.0 MW in Iceland that doesn’t seem to be core to their future operations.

These datacenters won’t compete with the hyperscaler’s on size. The biggest, in North Carolina, is 1/20th as big as the gigawatt facilities being built by Amazon, Meta and Google.

Whitefiber is targeting datacenters near commercial hubs, like Montreal and Paris, where they can deliver low-latency inference – AI lingo for faster answers.

Whitefiber is also not buying GPUs and racks. Their focus is on colocation.

This is where Whitefiber is a landlord. They build a powered-up shell and it is up to the customer to fill it with equipment.

Their flagship North Carolina datacenter, called NC-1 is, which will be the driver of growth over the next couple of years, is a 100% colocation facility.

The lease of space in NC-1 is expected to lead to stair-stepping growth over the next few years.

That growth is already largely baked into existing agreements. A signed tenant, Nscale, agreed to a 10-year $865 million lease on the facility.

Billings to Nscale are expected to start mid-2026. Barclays is estimating that Whitefiber’s revenue is about to take a big step up in the second half, followed by another big step in early 2028.

The impact to the bottom-line is expected to be equally large. Analyst estimates are for upwards of $200 million in EBITDA to Whitefiber by 2028.

At $40, and with 39 million shares outstanding, Whitefiber trades at a market cap of $1.5 billion and an enterprise value of $1.66 billion.

The biggest part of the debt is $230 million of 4.5% convertible senior notes due 2031, which Whitefiber completed in January 2026. There is a $60 million credit facility tied to the smaller Montreal facilities. There is another $20 million tied to the Iceland operation. Finally, there is a $100 million term loan facility from Bit Digital, a related party I’ll touch on in a bit.

They are going to have to raise some cash to fulfill their growth pipeline. Building out both stages of NC-1 by 2028 is going to require in the neighbourhood of $1 billion of capital.

Adding that to the current valuation puts the stock at an enterprise value of $2.7 billion. With $200 million of EBITDA in 2028, or about 14x EV/EBITDA.

That may not sound particularly cheap, but in today’s AI valuation world, it actually isn’t too bad.

THE NORTH CAROLINA FACILITY

NC-1 is a retrofit of an old building in Madison, North Carolina.

Retrofit means the physical data center is already stood up. It is just a matter of upgrading cabling and cooling to meet the requirements of an AI data center.

NC-1 has 99 MW of gross capacity, but all of that capacity isn’t available right now. The deal with Nscale is for the first stage of NC-1, which is going to deliver 54 MW of gross capacity. The other 45 MW is expected to be stood up by late 2027.

Gross capacity is the total load coming into the datacenter. Not all of that load becomes useful and directed to running GPUs, memory modules and the like. A meaningful chunk is consumed by the cooling system, by facility overhead like lights, controls, and is lost to power conversion and in distribution.

Nscale is going to get about 40 MW for their datacenter racks once all these factors get their piece.

Maybe the most important thing about this 54 MW is that the power is already contracted. Because of all the datacenters being built, the gird in many parts of the US is backlogged for years. The value of NC-1 is that the power supplier in the area, Duke, is already delivering the 54 MW to the site.

Who is Nscale?

When the landlord depends on one tenant, the big question is: is the tenant good for it?

We are looking at Whitefiber stock here, but it is as much the performance of Nscale that matters.

Nscale is a London-based AI infrastucture builder. It actually shares some similarities to Whitefiber. Both companies were born out of crypto-mining and have pivoted to AI.

Most importantly, Nscale is very well funded. In fact, in September 2025, Nscale secured the largest Series B financing (Series B meaning the second major financing round for a company) in EU history, raising $1.1 billion.

Nscale is valued at ~$15 billion, which is not far off mid-tier public datacenter “neoclouds” like Terawulf (WULF – NASDAQ) and even in the ballpark of the large names like Coreweave (CRWV – NASDAQ).

It also means that Nscale is around 10x the size of Whitefiber.

Nscale already builds GW scale campuses across Europe. Their Stargate Norway project is a partnership with OpenAI for a 230 MW facility initially, scaling to over 500 MW through a second phase – or about 10x the size of NC-1. Their Microsoft Narvik deal, a 230 MW campus also in Norway, is 5x NC-1.

The 40MW facility at NC-1 is a rounding error compared to their other capacity. Which is to say, Nscale appears to be about as good of a counterparty to a major datacenter that you could ask for.

Why did Nscale choose Whitefiber?

Whitefiber offers speed-to-market in the United States. Nscale gets a ready-to-go datacenter without having to tie up its own capital and queue time to build a US site from scratch.

Nscale is not without risk. They are still pre-profit, they are burning through a large amount of capital standing up datacenters in Norway, UK, Portugal, Iceland and the US, all at the same time. And for Whitefiber, they mostly depend on this single counterparty, which is never ideal.

Still, for a small player looking for their first big deal, you could do a lot worse!

CLOUD SERVICES

While the colocation deal at NC-1 with Nscale is going to be the growth driver for Whitefiber, today colocation is just a small part of revenue for Whitefiber.

The much larger contributor comes from their Cloud Services business.

Whitefiber provides cloud services through its leased Iceland datacenter and its two other operating datacenters, MTL-1 and MTL-3, which as the names suggest, both operate in Montreal.

The Iceland facility is barely mentioned by the company, though it was their first operating datacenter. There is also a mention in their filings that an agreement with the customer using the Iceland facility, Greenblocks ehf, was terminated earlier this year.

MTL-1 was energized over four years ago, well before the current AI/liquid-cooling buildout. It predates the whole GB200/Blackwell density era. That means it isn’t really an AI datacenter. It could be someday, but when I see there are 14 separate customers taking up a relatively small 3 MW of power, it is more like that these are corporate clients.

MTL-3 is fully leased by one customer, Cerebras, which is undoubtably an AI customer.

Cerebras is a chipmaker that provides a competitive inference chip to what Nvidia offers. While Cerebras is much smaller than Nvidia, they are a real company and offer a real solution. They have begun to offer access to their ASICs by selling compute from datacenters they operate, of which MTL-3 would be one of.

Again, it is an example of a high-quality customer that Whitefiber has been able to attract.

THE LEOPOLD ANGLE

A lot of the reasons to go long Whitefiber don’t have much to do with the business. What matters is where we are in the market right now and what the market cares about.

For example, Leopold Aschenbrenner’s Situational Awareness fund held 1.75 million shares of Whitefiber as of the date of his last 13-F filing, which was at the end of March.

That alone could be enough to send the stock higher if the momentum crowd catches onto Whitefiber.

Leopold is a hot new star in the investing world, and his fund is well known for making big bets on AI names and scoring big wins as a result. His fund was an early investor in Beam Energy (BE – NASDAQ), which has gone up about a gazillion percent in the last year. FinTwit loves this young man like few others.

This isn’t a big position for Leopold’s fund – it amounts to 0.54%. But it is a big position for Whitefiber. It amounts to almost 5% of outstanding stock.

More important is that in this environment, any association with a name like Leopold is enough to set FOMO into the stock. It is not hard to imagine a scenario where retail starts to use Leopold’s other big winners as “case studies” of what is going to happen next with Whitefiber.

THE BIT DIGITAL KNOT

Whitefiber isn’t really an independent company. It is a carve-out from Bit Digital (BTBT – NASDAQ).

Bit Digital sold about 20% to the public last August. They kept roughly 80% and voting control.

The CEO and CFO are the same for both companies. Whitefiber is run and controlled by Bit Digital.

That shows up in the financing. Whitefiber has a $100 million facility funding for the NC-1 buildout that is provided by Bit Digital and B. Riley.

Bit Digital has been honest about the relationship. They described the debt financing as providing better yields than they are getting on their Ethereum staking.

Bit Digital has pledged not to sell their shares this year. But the lock-up is over so after the end of this year so at that point sales of the shares are fair game.

Bit Digital is also another way to play Whitefiber. Given the large stake and the meteoric rise in Whitefiber stock, Bit Digital is actually only pricing in their Whitefiber stake into the current share price, meaning that if you buy Bit Digital stock, you get their Ethereum staking business for free.

WHAT SHOULD WE EXPECT NEXT?

That is a good question!

On any given day Whitefiber’s stock could be up or down 10%.

None of those moves reflect what is going on with the business itself.

Datacenters take 12-18 months to build and get up and running. Whitefiber’s runway is pretty well known for the next 2 years.

Longer term, what really matters for Whitefiber is AI adoption.

But I don’t think you even want to think about that with Whitefiber. You can’t look at this stock as a long term play here. Especially after the run it has had.

What you have to be looking for here is a short-term continuation of this move up. A consolidation at around the current level and then another move that can take it into the $60s or, hopefully, higher.

That move will be underpinned by the valuation comps that I showed earlier. You want to see investors looking at Coreweave, Terawulf and the like and pointing out that Whitefiber is about half the price.

It could also just as easily be underpinned by short covering. Short interest in the stock has been rising and is now 10% of shares, which is a very large percentage of the float (remember that most shares are held by Bit Digital).

Together, this could lead to another move up. But honestly, I don’t know if you want to overstay your welcome with the stock. If we get another leg, it may be best to take the gains and move on.

The reason is simply what I just said about the business. We are looking at number 2-3 years out when we are valuing these stocks. AI is moving incredibly fast, which makes looking out that far, something that is always perilous, even more so in this case.

We just don’t know what AI will look like in 2028. Will it be ubiquitous? Will it be restricted? Will we have figured out ways to run compact models locally? Or will datacenters dot the countryside?

I have no idea. Which is why I can give the go ahead on Whitefiber as a short-term play. But I hesitate to make any predictions for the long run.