Hut 8's $7 Billion Cash Mirage: The Restricted Balance That Couldn't Be Spent
The code whispered secrets the whitepaper buried. This time, the code was the footnotes of Hut 8's 10-Q. On its face, the Nasdaq-listed miner held roughly $7 billion in cash. The press release celebrated Bitcoin reserves, AI ambitions, and the war chest for a pivot. Then I read the restricted cash schedule. Only $233 million is actually spendable. The other $6.8 billion is locked inside two special-purpose vehicles, River Bend DC LLC and Beacon Point DC LLC, funded by subsidiary notes that the parent company does not guarantee. This is not a liquidity story. It is a liability story wearing a cash costume.
I have spent years dissecting protocol whitepapers and corporate filings. The behavior is identical: management emphasizes the asset, buries the encumbrance. Hut 8's balance sheet is not a monolithic pool of capital. It is a collection of cages. And the birds inside are not free to fly.
Context: Hut 8 began as a Bitcoin miner, one of the largest publicly listed in North America. In 2025, it merged its mining arm with American Bitcoin while dual-tracking toward AI data center infrastructure. The market narrative is familiar: miners dying from halving economics should rebrand as 'digital infrastructure providers.' Hut 8 announced two AI data center projects. River Bend carries $3.25 billion in project financing; Beacon Point carries $4.25 billion. Combined subsidiary notes: $7.5 billion. Restricted cash reported: $6.8 billion. Interest on these notes begins in November 2026. Principal matures in May 2028 and May 2030. Parent company Hut 8 Corp is completely outside the guarantee structure.
This is project finance 101. The problem is that Hut 8's investor communications allowed the market to treat the $7 billion as a treasury. It is not. The 10-Q line item 'cash and cash equivalents' includes restricted accounts. Under GAAP, a company can hold a huge cash balance and be thirty days from insolvency. I learned that during the 2017 ICO audits, when teams flashed token treasuries that were actually locked in illiquid presale wallets. Same song, new blockchain.
Core teardown begins with the line item. Total cash and equivalents on the June 30, 2025 balance sheet: approximately $7.0 billion. Restricted cash: $6.8 billion. Unrestricted: $233 million. The gap is not a rounding error. It is 96.7% of the cash entering the narrative.
The restricted funds are not available to pay vendors, cover margin calls, or plug operational losses. They are segregated in construction reserve accounts and debt service reserve accounts. The note proceeds were never earnings. They were borrowings. Hut 8 borrowed $7.5 billion, spent or reserved $6.8 billion, and the leftover is what the company can actually spend.
But wait. The notes total $7.5 billion, and the restricted cash is only $6.8 billion. Where did $700 million go? The 10-Q does not fully explain. Possible answers: issuance discounts, legal fees, bond insurance, or early construction disbursements. In my experience, this gap becomes the first place a project discovers cost overruns. The reserves are designed to cover interest and principal, not construction inflation. If copper prices, power interconnection costs, or GPU delivery delays eat the float, the parent must inject equity. And the parent has $233 million. That is my core red flag: the project notes are large enough to impress, but the free cash buffer is one construction change-order away from emptiness.
Now run the operating cash flow. Q2 adjusted EBITDA was $10.4 million. Interest expense was $51.2 million. The interest coverage ratio is 0.2 times. Q2 net loss was $177.1 million, including $138.6 million in digital asset impairment losses. First-half operating cash flow: negative $32.8 million. Put simply, the mining business is not paying for itself. Hut 8 is a company with a negative carry. It borrows at 7%, lends via Bitcoin volatility, and has no operating segment that generates enough cash to cover interest.
Management will likely capitalize construction-period interest for River Bend and Beacon Point. That is an accounting choice, not an economic one. Capitalized interest is a financial instrument pretending to be an asset. If the projects are delayed, the capitalized interest is written down or expensed. A delay is not just a delay. It is an accounting detonator.
The Bitcoin side does not rescue it. Hut 8 reports 17,316 BTC across the group. The split: 9,376 held in custody, 3,090 pledged to purchase miners, and 4,850 posted as loan collateral. The custody bucket worries me. Is that Hut 8's own Bitcoin, or Bitcoin belonging to American Bitcoin or third-party clients? The filing does not allocate status buckets between Hut 8 and American Bitcoin. American Bitcoin holds 8,002 BTC. If those coins are included in the 17,316 headline, the parent's effective holding is roughly 9,314 BTC before pledges. That is a materially different asset position. Between the lines of the ABI lies the intent. The American Bitcoin structure is a black box. Without a legal allocation table, any net asset value calculation is fiction.
The FalconX loan adds another layer. $200 million borrowed, 7% annual interest, due April 2027, secured by Bitcoin. If the collateral is the disclosed 4,850 BTC, and Bitcoin trades at $100,000, the collateral value is $485 million against a $200 million loan. Loan-to-value: 41%. A typical maintenance margin of 130% means the lender requires $260 million of collateral. The Bitcoin price would need to fall below $53,600 to trigger a margin call. That is not a 1980s horror movie. It is a 2022 sequel. When Bitcoin dropped below $55,000 in past cycles, leveraged miners faced exactly this choice: post more coins or sell into a falling market. It didn't loop, it drained.
Then there is the SPV architecture. River Bend DC LLC and Beacon Point DC LLC are separate legal entities. Hut 8 Corp is not a guarantor. Noteholders can only attack the project assets if the projects fail. Parent shareholders cannot touch the restricted cash even if the projects succeed. The cash belongs to the project, the project belongs to the LLC, and the LLC's economics flow to bondholders first, equity holders last. This is a classic liability firewall. It protects the parent from downside, but it also means the market's $7 billion cash narrative should be revised to $233 million of free liquidity plus an option on future projects. Logic does not lie, but architects often do.
The entity map matters more than any single line item. Hut 8 Corp sits at the top. Below it, a mining operations subsidiary, a digital asset custody entity, and then two AI special-purpose vehicles. American Bitcoin sits as a separate venture, partly controlled by Hut 8, but its Bitcoin is not automatically available to the parent. The restricted cash is not Hut 8's cash. It is the cash of River Bend DC LLC and Beacon Point DC LLC. The parent's access to that cash is limited by the note indentures, the reserve account terms, and the lenders' consent. That is institutional centralization in its purest form: the assets that look like a fortress on the consolidated balance sheet are actually a series of locked rooms, each with a different key. The key-holders are not the shareholders.
Where does Hut 8 stand competitively? Core Scientific signed a 12-year AI hosting agreement with CoreWeave. That contract delivers visible revenue. IREN built GPU clusters using its own power. Bitfarms has been slower to pivot. Hut 8 has two construction projects, no disclosed AI customer, no signed hyperscaler lease, no megawatt target, and no utilization milestone. The financing package is impressive on paper. But in the AI infrastructure race, capital is the entry ticket, not the winning lottery ticket. The customer is the contract. The contract is the cash flow. The cash flow is the valuation. Read the function calls, not the press release.
Let me add one quantified observation that most coverage missed. The adjusted EBITDA of $10.4 million is computed before digital asset fair value adjustments. That means it includes gains from selling mined Bitcoin at a profit, but it excludes the impairment on the coins that were held. Under the new fair value accounting standard, miners can show gains when prices rise and losses when prices fall. Hut 8's June-quarter loss was $138.6 million in digital asset losses. The market is not pricing Hut 8 as a portfolio of Bitcoin plus a real-estate construction loan. It is pricing it as an AI growth stock. That mismatch is exactly where the disappointment enters. When revenue is three quarters away, the stock trades on promises. The promise here is a construction schedule.
There is also the matter of note covenants. The restricted cash sits in reserve accounts. Draws are almost certainly limited to contracted construction milestones. That means the money is not disbursed to Hut 8 management. It is disbursed to contractors, directly, after third-party validation. In theory, this prevents runaway spending. In practice, it means the parent company cannot redirect funds to any other division. If the AI projects must pause for a power interconnection delay, the reserve account waits. The project debt still accrues. The construction loan still matures. And the parent still needs to pay $9 million per quarter in FalconX interest from operations that generate nothing close to that. The math is unforgiving.
The market narrative is a lagging indicator. For months, HUT traded as a leveraged AI play because the cash balance gave comfort. My expectation is that the next quarter will show continued cash burn, no material AI revenue, and a gradual decline in the unrestricted cash. The restricted balance will stay high, but that high balance will not be translatable into operational runway. This is the classic gap between accounting assets and financial capacity.
Based on my audit experience, whenever a company highlights a cash balance that is 96.7% restricted, the press release is doing the opposite of disclosure. The 10-Q is the truth. The truth is that Hut 8's AI pivot is a construction financing event, not an operating success. I am not predicting bankruptcy. I am predicting repricing.
The next 10-Q should show the status of the restricted cash draws. If the construction reserve account is declining faster than physical progress, that is a warning. If the debt service reserve is being tapped before interest payments begin, that is a structural failure. Those are the function calls that matter.
But the bulls are not wrong about the direction. Hut 8 is making a rational pivot from commodity Bitcoin mining to digital infrastructure. The scale of project-level financing, $7.5 billion at 6.13 to 6.19 percent, is not available to random newcomers. Some institutional lender ran a diligence process and concluded the land, power, and sponsorship are credible. The SPV structure protects the parent from construction blowups in a way that balance-sheet financing would not. And the Bitcoin position is still substantial. If Bitcoin appreciates, the FalconX loan collateral becomes safer, and the custody coins can later be sold to fund the parent's liquidity. The market ultimately prices future cash flows. If Hut 8 signs a hyperscaler contract in the next two quarters, today's liquidity panic will be a footnote. I have seen worse structures recover. I have also seen better structures collapse. The difference is never the size of the cash balance. It is the existence of a revenue contract.
Until a revenue contract appears, the only truthful liquid number is $233 million. I will track three data points. A signed AI lease with a tenant with actual balance-sheet capacity. The megawatt load under construction, not just the financing. The legal ownership of American Bitcoin's 8,002 BTC. If those disclosures arrive, the restricted cash becomes an asset. If they don't, Hut 8 is a miner holding a construction loan, not a data-center empire. What good is a $7 billion balance sheet if you cannot spend it?