On November 4, 2026, something distinctly awkward happened in US equities.

During the day, Wall Street firm Bernstein cut price targets on seven bitcoin miners at once, the deepest by 54%. IREN went from $100 to $55, RIOT from $35 to $18, and CIFR from $23 to $15.

That same evening, Hut 8, another miner pivoting into AI data centers, posted third-quarter results showing net income of $190 million, up from just $50 million a year earlier. On paper, excellent news. The stock fell 20% that night.

The ones not making money got marked down. The one making serious money got hammered too. What is Wall Street afraid of?

The answer lies partly in the US election that had just wrapped up, and partly in these companies' own books.

Texas Has a New Governor, and Building Data Centers Just Got Harder

Start with the election, held the day before the selloff.

In the November 3, 2026 midterms, Democrats swept, taking both chambers of Congress and flipping at least seven Senate seats. The Washington Post called it a reversal of a scale not seen since 1974.

But what is keeping mining executives awake is the change of governor in Texas. Democratic candidate Gina Hinojosa defeated Abbott, who was seeking a fourth term, becoming the state's first Democratic governor in more than three decades. Her campaign promise was blunt: freeze approvals for new data centers on day one, scrap tax incentives, and impose strict limits on how much water and power these facilities can draw.

When Texas sneezes, the entire AI data center industry catches cold, because it is the top state in the country for data center construction. More than a quarter of all data centers planned nationwide through 2030 are slated for Texas, including the flagship site for Stargate, the mega AI project.

And it is not just Texas. Newly elected Democratic governors in Ohio and Georgia all campaigned on restricting data center construction. The reason is straightforward: voters are fed up with facilities soaking up water and power and pushing up electricity bills. An NBC exit poll found 76% of voters oppose a data center being built near their home.

Local officials once treated new data centers as an achievement to run on. Now the votes are in attacking them. Goldman Sachs estimates that after this election, roughly 45% of all planned US data centers could be stalled in the near term.

Of the 474 Gigawatts in the Queue, Only 1.8% Have Actually Been Energized

Worse still, even if the new governor leaves you alone, the Texas grid is already close to bursting.

Large projects queued for interconnection on the Texas grid (ERCOT) now total 474 gigawatts of demand, 90% of it data centers. Of that enormous backlog, only 1.8% has actually secured power and started operating.

It is like 100 people lined up outside a restaurant with fewer than two getting through the door. The new governor taking office amounts to adding a padlock to that door.

Which makes the logic behind the price target cuts simple: it is not that AI no longer needs data centers. It is that securing power and getting them built became vastly harder overnight.

Of That $190 Million in Profit, $413.8 Million Came From Bitcoin's Rally

Go back to Hut 8's November 4 earnings release and you can see how sharp Wall Street's eye was.

The strangest thing about this report is that the bad news and the good news are both real.

On the surface, third-quarter net income was $189.5 million versus $50.6 million a year earlier, roughly 3.7 times higher. Open the books, though, and the money did not come from leasing data centers. It came from crypto.

Of that $189.5 million, as much as $413.8 million was digital asset gains. In plain terms, the bitcoin on the company's balance sheet appreciated. Those coins were not sold for cash; the gain was simply booked under accounting rules.

Nor does all of that $189.5 million in paper profit belong to Hut 8 shareholders. After stripping out more than $58 million attributable to other shareholders, including those of listed subsidiary American Bitcoin, net income attributable to Hut 8 itself was only $131 million.

All Three Core Business Lines Shrank, While Interest Topped $100 Million in a Single Quarter

Squeeze out the bitcoin bubble and look at the actual operating business, and the numbers are ugly.

Hut 8's third-quarter revenue was just $71.1 million, missing Wall Street's $75.9 million estimate and down 15% from $83.5 million a year earlier. All three segments contracted: power generation revenue fell from $8.4 million to $1.3 million, digital infrastructure from $5.1 million to $1.4 million, and even the largest line, compute (crypto mining plus cloud services), came in at only $68.4 million.

Stripping out coin price swings entirely, core operating profit (adjusted EBITDA) was $33 million for the quarter, versus $30.5 million a year earlier. A full year of work, and the core business is essentially running in place.

The core business is not earning much, and the spending is alarming. General and administrative expenses jumped from $25.9 million a year ago to $99.5 million this quarter, of which $64.7 million was stock-based compensation for executives and staff. Heavier still is interest: to fund equipment purchases and AI data center construction, quarterly interest expense surged from $8.6 million to $105.5 million.

That is why the stock dropped 20%. Wall Street saw it clearly: the $190 million in profit is paper wealth handed over by bitcoin, while more than $100 million of interest every quarter and a 15% decline in the core business are the real bleeding.

The $26.6 Billion in Contracts Is Real, but the Rent Does Not Arrive Until 2028

To be fair to Hut 8, the picture is not all bad. The other side of this report holds something substantial.

During the third quarter, Hut 8 signed a major contract for phase two of its Beacon Point campus, leasing out 352 megawatts of capacity for 15 years at a total value of $9.8 billion. The tenant is the same highly rated large enterprise as in phase one, which doubles that customer's footprint at the campus to 704 megawatts and brings total contracted value there to $19.6 billion. Add the separate River Bend project and Hut 8's long-term AI data center contracts total $26.6 billion.

To get to the day those facilities start collecting rent, the company has stocked up. It just secured a $1.07 billion revolving bank credit facility, and holds roughly $6.6 billion in cash and short-term investments plus nearly $700 million in bitcoin. It even sold more than 3,000 bitcoin for real cash, spending $250 million on buybacks and raising its future repurchase authorization to $1 billion, signaling to the market that money is not the problem.

On one side, a $26.6 billion order book and a pile of cash. On the other, a shrinking core business and more than $100 million in quarterly interest. So why did the market sell first and ask questions later?

Because distant water does not quench present thirst. Beacon Point phase one, the site behind the multibillion-dollar contract, will not be delivered until the third quarter of 2027, and phase two not until the second quarter of 2028. For the next seven or eight quarters, those new facilities will generate zero rent while the company keeps pouring money into construction and servicing bank interest, carried by its deposits, loans, and bitcoin sales.

Which explains both the panicked 20% selloff and why KBW still has the nerve to keep a price target above $90. The contracts are hard cash, and so is the grind of surviving the two-year gap before they pay.

Valuation Shifts to Its Third Framework: Forget the Site Plan, Show Me the Power Hookup

Once you understand the election and the earnings, you can see that Wall Street's method for valuing these miners has changed completely.

Over the past few years the market has run through three frameworks. The first was bitcoin's price. Back then a miner was essentially leveraged bitcoin: coin up, stock up. The second was who held the most power capacity, because AI was short of electricity. Get into a grid queue with a few hundred megawatts on the drawing board and the stock could double, even with nothing built.

Starting this week in November 2026, the market has moved into the harshest framework yet: stop showing me site plans. Show me whether the power is connected, whether the lease is signed, and when the rent hits the bank.

The companies Bernstein cut hardest share the same flaw. Too much of their market value rests on data centers still to be built. With policy tightening in Texas and elsewhere, a large share of those paper plans may never get power.

Conversely, the storm is a windfall for some companies. If Goldman is right that 45% of new planned capacity gets blocked, data centers become scarcer. Firms that already have interconnection approvals, signed long-term tenants, and in some cases AI rent flowing in get the benefit of policy wiping out their competition for them.

Who Is Collecting Rent, Who Is Paying Interest, and Who Is Still Telling a Story

From here, this cohort of miners turned AI landlords will separate sharply.

The most comfortable group has power connected, anchor customers signed, and rent already coming in. CoreWeave and Applied Digital sit here. The more local governments restrict new construction, the more valuable their existing capacity becomes.

The second group has the hardest time: contracts signed, facilities still under construction, no rent yet. Hut 8 belongs here, along with WULF and CIFR. They are racing the clock, praying construction stays on schedule while keeping their funding intact. Even with $6.6 billion on the balance sheet like Hut 8, an interest bill of more than $100 million every quarter is genuine pressure.

The third and most dangerous group is the small companies that hold only land, a number in the interconnection queue, and an AI story. With the policy wind shifting, Wall Street has stopped listening to stories. Without a power approval and a signed lease, the door to issuing equity or borrowing to build is essentially shut.

The Shovels Are Still Being Forged in the Workshop

Everyone knows the line: in a gold rush, sell shovels.

But that line comes with a condition. The shovel has to be made, sold, and paid for before it counts. If it is still being forged in the workshop, the interest on the high-rate loan you took to buy iron ore may crush you first.

History runs the same script, whether it was railroad building in the 19th century or the fiber laid during the 2000 internet bubble. Demand for railroads and networks was absolutely real, but most of the companies that borrowed heavily to lay track and cable died on the road just before dawn, undone by long build times and high interest.

The AI revolution is real. That does not mean everyone building data centers for it will survive to collect the payout.