Ledgers don't lie, but headlines can. Hut 8's latest 10-Q shows roughly $7 billion in cash and equivalents. That number is real in the accounting sense. It is also misleading. Only $233.6 million sits unrestricted. The other $6.8 billion is locked inside two AI data-center subsidiaries. If you bought HUT because you thought the company held a $7 billion war chest, you bought a balance sheet that exists mostly behind a covenant wall.
I have run this exact type of audit before. In 2017, I tore through ICO listing criteria and found tokens without auditable contracts. The market was pricing narratives. It still is. The question is not whether Hut 8 has cash. The question is who can spend it, and when.
Hut 8 is no longer a pure bitcoin miner. It is trying to become a digital infrastructure company. The market has rewarded miners that pivot to AI hosting. Core Scientific signed a long-term AI hosting contract with CoreWeave. IREN built its own GPU clusters. Hut 8 wants the same premium. But its AI projects are not operational. River Bend and Beacon Point are under construction. They have raised $3.25 billion and $4.25 billion in subsidiary-level notes. Those proceeds are restricted to building and debt service. The first interest payment does not start until November 2026.
This is not a hidden fraud. It is a classification problem. The $6.8 billion sits in restricted accounts controlled by the project notes. This is standard project financing: debt proceeds flow into a build reserve, then a debt-service reserve. The parent cannot redirect those dollars to mine more bitcoin or buy GPU chips. Parent shareholders do not own that cash directly. They own a residual claim on a construction project that has not yet produced a dollar of AI revenue.
The operating engine is stretched. Second-quarter adjusted EBITDA was only $10.4 million. Interest expense was $51.2 million. That is a 0.2x interest coverage ratio. The core business does not earn enough to service the debt it has already taken on. Cash flow from operations was negative $32.8 million in the first half. Interest income of $27.1 million helps, but the AI projects have no revenue yet. The entire bridge is held up by unrestricted cash, bitcoin collateral, and expectations. Expectations are not a liability line item.
Interest capitalization is the hidden earnings switch. The River Bend and Beacon Point notes carry rates of 6.13% to 6.19%. During construction, that interest can be capitalized to the project balance sheet. The parent's income statement does not feel the full weight of those coupon payments yet. But if a project is delayed, or if the construction timeline breaks, capitalized interest must be expensed. That would convert a buried line item into a direct drag on earnings. The same mechanism applies to the $51.2 million quarterly interest expense. The number looks manageable only because the AI projects are not yet paying operating costs. Once they do, the coverage ratio will matter more than any headline cash figure.
Bitcoin is both the collateral and the crutch. Hut 8 holds 17,316 BTC. Of that, 9,376 are in custody, 3,090 are pledged to buy miners, and 4,850 are collateral. The FalconX loan is $200 million at 7%, due April 2027. If the collateral is 4,850 BTC at roughly $100,000 per coin, that is $485 million of collateral for a $200 million loan. LTV is about 41%. A 130% maintenance margin would put the loan under stress near $52,000 to $65,000 BTC. That scenario is not imminent. But the asymmetry matters. If BTC falls, Hut 8 must post more collateral or sell coins. If BTC rises, the liquidity trap loosens. This is high beta in its purest form.
The cash number also masks ownership frictions inside the bitcoin stack. The merged group holds 17,316 BTC, but 8,002 of those coins sit under American Bitcoin. Hut 8 did not allocate status buckets between the two entities. It did not disclose how many bitcoin back the FalconX loan. That ambiguity matters. If some of those coins are not freely owned by the parent, then the bitcoin reserve narrative is less clean than the headline suggests.
Alpha hides in the friction between chains. In this case, the friction is not between Ethereum and Solana. It is between the parent balance sheet and the project subsidiaries. The restricted cash is real cash, but it is not free cash. It is project capital with a covenant box around it. The distinction changes how you should value HUT.
The contrarian read is uncomfortable. Retail sees $7 billion of cash and an AI pivot. Smart money sees a construction escrow, a parent with a thin free-cash cushion, and no signed AI customer. The market is pricing HUT as if the pivot is already complete. It is not. What HUT has proven is the ability to raise debt. That is not the same as proving the ability to operate an AI data center. Core Scientific has a signed contract with CoreWeave. Hut 8 has financing. Financing is a bridge, not a destination.
The $7.5 billion in project notes is not a blessing. It is a commitment. A construction overrun or delay does not wait for AI demand. It waits for cash. If the project needs an equity injection, parent shareholders will be diluted. If the project fails, the SPV structure contains the damage. But it also means parent shareholders have no direct claim to the restricted cash. They own an option on the project's future cash flows. Options decay if milestones slip.
Regulation is the one area where HUT is better than 99% of crypto projects. It is a Nasdaq-listed company. The SEC forces it to disclose the restricted cash, the FalconX loan, and the project notes. You can audit these numbers in the 10-Q. But transparency is not liquidity. A footnote that explains the cash is restricted is not the same as a bank account you can draw on. The 2017 ICO mania taught me that disclosure documents are only as valuable as the analyst reading them. Most market participants will not read the footnote. They will see $7 billion and move on. That is the edge.
The real insight is not that Hut 8 has less cash than reported. The real insight is that the cash is not owned by the entity you bought. The restricted millions are owned by bankruptcy-remote subsidiaries. The parent owns the equity in those subsidiaries. Equity in a construction project with no revenue is not the same as spendable cash. In my 2020 DeFi arbitrage build, I learned to separate deployable liquidity from dashboard liquidity. This is the same discipline on a corporate scale. Deployable capital is the only capital that matters for survival.
The market's next repricing trigger will be a signed hyperscaler contract. If River Bend or Beacon Point locks a long-term AI tenant, the restricted cash becomes productive capital. The current valuation can absorb that change quickly. If no contract appears by the time interest payments start in November 2026, the discount rate rises. Conviction without verification is just gambling.
The first line of defense is the $233.6 million of unrestricted cash. That is real. It can pay operating losses for a while. But at the current burn rate, it is not infinite. The second line of defense is the bitcoin position. If BTC keeps climbing, the FalconX loan becomes easier to manage. If BTC stalls or dives, HUT faces a margin loop. The third line of defense is the project financing itself. Huge, but locked.
No one should treat this as a bankruptcy scenario. Yet. The balance sheet is not insolvent. It is mismatched. There is long-duration illiquid project value, and near-term obligations that depend on capital markets and bitcoin price. That mismatch is exactly what volatility exposes. Volatility exposes weak foundations first. A weak foundation here is not the mining business. It is the assumption that restricted cash is available cash.
What changes now? The burden of proof shifts to management. The next 10-Q must show construction milestones, signed contracts, or at least a clear drawdown schedule. If the company treats restricted cash as operating cash in its earnings narrative, treat that as a red flag. If it breaks out the liquidity picture clearly, that is a signal of discipline. Efficiency is the enemy of complacency. Hut 8 needs to be efficient with every dollar of the $233.6 million.
Hut 8's transition is real. The capital is real. The AI infrastructure build is real. But none of that makes the stock a buy. The stock is a leveraged call on three variables: bitcoin price, construction execution, and AI customer demand. All three must move in the same direction. If that happens, the restricted cash will turn into revenue. If one of them breaks, the cash number will not save the stock. Structure survives the storm; chaos does not. Can you still call it cash when you cannot spend it? The next two quarters will answer that question.
