Videos·Oct 23, 2026·7 min

Wall Street Says This $2 AI Power Stock Is Worth $16. I Checked Who Said It. (FRMI) (Fermi Stock)

FRMI

Chapters

Did you catch it?

3 quick questions on this video. It's step 3 of the Read a stock like an auditor path, and passing here counts there too.

1. Why does Blake say Fermi's average price target of about $16.60 can mislead you?
2. What did Blake find when he checked which firms set Fermi's price targets?
3. In Fermi's director vote, what happens if you own the stock and just ignore your ballot?

Transcript

What I say in the video, from its captions. Press a time to play the video from there.

Fermi at $2.65 with a $16.60 average price target

Fermi, the AI power company Rick Perry co-founded just closed at $2.65 a share, 91% below its high from last fall, and yet the 8 Wall Street analysts who cover it still have an average price target of about $16.60. That's more than six times the stock price, but when I checked who set those targets, the newest ones are way lower and at least six of the eight firms were banks that helped sell Fermi's IPO in the first place. And next week could decide a lot because there's a shareholder vote the day before Halloween and Halloween itself is the deadline for the only binding lease this company has ever signed.

So, whether you own it, you're tempted by that target, or you think the whole thing is a mirage in the Texas Panhandle, This is the week that matters.

What Fermi is building (Project Matador, explained with Hank's truck stop)

Picture a guy named Hank who decides to build the biggest truck stop in the panhandle with 50 pumps, a giant parking lot, and his own power plant out back, before a single trucking company has signed up to fuel there. That's basically Fermi. It's building a campus called Project Matador on about 8 square miles of Texas Tech land near Amarillo with plans for up to 11 gigawatts of power, gas first, and nuclear later to rent out to AI data centers. Before the IPO, Rick Perry went on Fox News and called AI data centers the new gold rush and said he'd seen a site in Texas that could be the largest data center in the world.

[Rick Perry] "AI-driven data centers are going to be the absolute new gold rush, if you will, of the 21st century." And just like Hank, the spending comes first. Fermi has put more than $1.5 billion into the site so far, and it still has no revenue. Which is fine as long as the truckers actually show up.

How the stock fell from its $21 IPO

Fermi started trading on October 1st of last year at a $21 IPO price, opened at $25, and the guy who bought on day 1 because Rick Perry plus AI sounds like a sure thing, felt like a genius for about a day. Then in December, Fermi said its first prospective tenant had backed out of a $150 million deal to help fund construction, and the stock lost about a third of its value in a single day. In April the board removed co-founder Toby Neugebauer as CEO and he's been fighting them in court ever since.

While his side says it still owns about 22% of the company. So Hank lost his first trucker, then fired his business partner. and the partner still owns a big chunk of the truck stop.

The TensorWave lease, AMD and the Halloween deadline

But in August, Fermi finally signed a real tenant. An AI cloud company called TensorWave agreed to lease 222 megawatts for 15 years worth about $6.5 billion in rent. Now here is the catch. TensorWave is a startup, so the lease needed somebody big behind it. And earlier this month, the chip giant AMD signed a credit support agreement for it, and the stock jumped by double digits that day. That's Hank's one trucker showing up with his rich uncle as a cosigner, except the uncle's signature comes with conditions, and the deal still hasn't closed.

The closing date was already pushed once from the end of September to Halloween, and if conditions aren't met, either side has the right to walk away. Fitting, because this chart already looks like a horror movie. And those conditions aren't small. Fermi still has to finalize the guaranty and line up project financing big enough to actually build the thing.

The cash problem: going-concern warning and convertible notes

Because money is the real pressure here. At the end of June, Fermi had about $62 million of cash plus $29 million of restricted cash, and it spent over $600 million on the site in just the first half of the year. Its own quarterly filing warned of substantial doubt about its ability to keep going for the next year, which is accountant-speak for "we need more money," although management says its plans resolve that doubt. So in July, it borrowed over $400 million through convertible notes, which basically just means bonds that can turn into stock later, on top of more than $500 million of debt it already had at the end of June.

This is where the doomer in your group chat says, "Told you so." And honestly on cash, he has a point.

Who set the $16.60 price target

So with all of that going on, how do 8 analysts still land on an average target around $16.60? A price target is just an analyst's guess at where the stock will be in about a year. And the five targets set since late July average about $10.60: Cantor at $8, down from $12 last week, UBS at $6, Evercore and Mizuho at $11, and Stifel at $17. Meanwhile, the highest target on the whole list was set back in February, when the stock was trading somewhere between $7 and $12.

That's like checking February's forecast to decide if you need a jacket today.

The banks rating Fermi helped sell its IPO

And here's the important part that most people never check. Cantor, UBS, Evercore, Mizuho, Stifel and Berenberg were all named on the cover of Fermi's IPO prospectus, which means they helped sell the shares to begin with. That doesn't mean their analysts are wrong or dishonest because research and banking are supposed to be walled off from each other, but it's worth knowing that the banks grading the stock got paid to sell it. In Hank terms, the appraisers are the same guys that sold you your shares.

The October 30 vote (Rick Perry is on the ballot)

Which brings us to the vote: on the day before Halloween, shareholders vote on re-electing two directors, and one of them is Rick Perry himself. And this year, the bar is unusually high because each director needs a majority of all outstanding shares, not just the votes cast and the proxy says, "Abstentions and broker non-votes count the same as votes against." Said another way, if you own Fermi in your brokerage app and you just ignore the ballot, you basically voted no.

Meanwhile, Neugebauer's side owns about 22%, says it isn't asking anybody for their votes and wants the board to hire an independent bank to test a possible sale against the current plan. So the anxious trader is refreshing his proxy email like it's a stock chart. Nobody knows how that vote lands and I won't pretend that I do either.

The bull case: $6.5 billion of contracted rent

Now to be fair, there is a real bull case. If the TensorWave lease closes with AMD behind it, Fermi has about $6.5 billion of contracted rent, or over $400 million a year on a company worth about $1.7 billion today. Its new CEO, Lee McIntire, who spent about 40 years on giant construction projects, says Fermi will be producing power by the end of 2027, and that they should be judged on the electrons they actually put out. [Lee McIntire] "Because we're going to be really measured on electrons that are put out."

Electrons, not press releases. Honestly, that's the right scoreboard.

The bear case: financing and dilution

Chad already did the napkin math on a 6-bagger and the trad wife asked the only question that matters, which is who's paying to build it. Because that, my friends, is the bear case in one sentence. The lease isn't closed, AMD's support comes with conditions, the financing isn't lined up, and every month without it burns more cash. And if Fermi has to raise that money by selling a pile of new shares at $2 and change, today's shareholders get diluted, which basically just means your slice of the pie gets smaller, even if the pie eventually gets built.

For what it's worth in June, about 864,000 of Rick Perry's shares were sold at just over $7. But that was through options the founders gave pre-IPO investors, and he still holds about 15.8 million shares.

The 3 dates to watch

So here's how I would think about it. The vote tells you who runs the board. Halloween tells you whether there's a tenant at all, and the November 12 earnings tells you how much cash is left to build with. If the lease closes and real financing gets announced, the bull case gets a lot more believable. If it slips again or money comes from a pile of new shares, that $16 average is February's weather forecast. As for Hank, he's still got 50 pumps, a power plant, and one trucker whose rich uncle hasn't finished signing.

And on Halloween we find out if the truck actually pulls in. Now I'd love to know would you buy Fermi before Halloween, after Halloween, or never? Tell me in the comments. YouTube picked this next video for you, and unlike Fermi, it's already plugged in.

The stocks in this video

More videos

Search all 61 videos →

This is education, not investment advice, and figures are as of the day the video came out. Full disclaimer. Spot a mistake? Tell me.