CleanSpark's $2.3 Billion Wager on One Data Center

On September 18, CleanSpark priced $2.276 billion of 7.875% senior secured notes due 2031. The shares jumped more than 8%, which is how the market usually responds when a company it is unsure about suddenly looks committed. The commitment is real. The question is whether the math works.

This bond offering is not general corporate debt. It is project financing for one facility — a 175-megawatt data center in Sandersville, Georgia — built for one tenant under a 20-year triple-net lease. CleanSpark is putting nearly its entire company behind that single bet, and the financial test is clear: can the Sandersville lease service the debt that funds it, and can the company survive the years before rent arrives?

The deal in numbers

The notes carry a 7.875% coupon, which means roughly $179 million a year in interest for the next five years before maturity. That interest does not come due on day one — proceeds from the offering will fund the remaining buildout costs, reimburse CleanSpark for equity contributions already made, and fill a debt service reserve account. Delivery to the tenant is expected to begin in the fourth quarter of 2027.

On the revenue side, the lease with a high-investment-grade technology company is estimated at $6.6 billion over the initial 20-year term, or up to $11.6 billion if both five-year extension options are exercised. CleanSpark says the project will contribute approximately $330 million in annual net operating income, with a cumulative NOI margin of nearly 100%. At that run rate, the lease should cover the $179 million in annual interest payments on the new notes and still have $150 million left over.

That math works — once the facility is complete and the tenant moves in.

The gap before the cash arrives

Between now and Q4 2027, CleanSpark must build a facility estimated at $1.75 billion to $2.1 billion to complete and then wait for revenue. The $2.276 billion in new debt closes that funding gap, but it also means the company is taking on annual interest obligations that exceed its current annual revenue.

As of June 30, CleanSpark reported $138 million in quarterly revenue, down 30.5% from the prior year. For the nine-month period ending June, total revenue was $455.6 million. Annualized, that is roughly $607 million — but management acknowledged "currently challenging bitcoin mining economics" and the revenue trend is downward, not stable. The company also reported a net loss of $239.8 million in the most recent quarter and a nine-month accumulated loss of nearly $1 billion, driven by Bitcoin valuation write-downs and weaker mining margins.

The balance sheet shows $202.6 million in cash and $814.9 million in Bitcoin holdings as of June 30. Total assets stand at $2.7 billion against $1.9 billion in liabilities, with $1.8 billion in existing long-term debt — primarily zero-coupon convertible notes due 2030 and zero-coupon convertible notes due 2032, which require no cash interest payments today. Stockholders' equity was $800 million.

Add the new $2.276 billion in secured debt to the existing $1.8 billion and CleanSpark will be carrying roughly $4.1 billion in total obligations against a $3.4 billion market capitalization. That is not a traditional debt-to-equity ratio — the company now operates two fundamentally different businesses, one losing money and one not yet started. The point is simpler: this is a leveraged build on a balance sheet that is already stretched thin.

Why the structure matters

The notes are not unsecured. They are issued by a subsidiary, CSDC Finance I, LLC, and secured by first-priority liens on substantially all assets of the issuer and CSRE Properties Sandersville, LLC. The notes are guaranteed by the property subsidiary and backed by CleanSpark's completion guarantee. This is how project finance works: the lender gets first claim on the data center assets, and the parent company promises to fund any shortfall during construction.

For CleanSpark shareholders, that structure is both comfort and exposure. It isolates the debt to the Sandersville project — the lenders cannot reach the rest of the company unless the facility fails. But the completion guarantee means if construction runs over budget, the parent company writes the check. And if the project fails entirely, the parent company has invested equity that is gone.

The tenant is described as a "high-investment-grade, leading global technology company." CleanSpark declined to name them, but the credit quality matters because the entire thesis depends on that tenant taking delivery, staying for two decades, and making rent payments on time. A missed milestone during construction could result in rent abatements or even lease termination, and the buildout cost of $10 to $12 million per megawatt leaves no margin for error.

What the market is seeing

The 8% pop in shares is not a vote of confidence in the existing mining business. It is a vote that the market believes Sandersville is a real asset worth building. The price at 98.5% of face — a slight discount — suggests bond investors want a little extra yield for taking credit risk on a Bitcoin miner turned data center developer. The effective yield to maturity is roughly 8.2%, which is not cheap, but it is not distressed either. It is the pricing you get for a speculative-grade borrower building an asset with a committed tenant.

There is also a Texas angle worth noting. The Sandersville tenant signed a letter of intent and exclusivity arrangement covering CleanSpark's entire Texas portfolio — up to 885 megawatts across 718 acres. That is a letter of intent, not a signed lease, and it is not part of the bond proceeds. But if even half of it materializes, the implied revenue expansion is enormous.

The investment case, plainly stated

CleanSpark is no longer just a Bitcoin miner. It is a company attempting to convert scarce, grid-connected power infrastructure into long-term data center leases. Sandersville is the first test of whether that conversion works at scale. The $2.276 billion bond offering is the bridge between the mining business that pays the bills today and the infrastructure business that is supposed to pay them tomorrow.

For an investor, the decision turns on three questions:

First, can CleanSpark deliver the Sandersville facility on time and on budget? The long-lead items have been ordered and prepaid, and the equity portion has been funded. That de-risks execution but does not eliminate it. A construction delay pushes revenue further out and burns through cash.

Second, will the mining business survive until the data center revenue arrives? Bitcoin mining margins have compressed, revenue is down 30% year over year, and the company is reporting losses. The existing convertible notes do not require cash interest, which buys breathing room. But if the mining cash flow dries up further, the company may need to raise equity or sell assets.

Third, is the tenant credit quality as solid as described? A "high-investment-grade global technology company" is reassuring, but the name is undisclosed and the lease is still conditional on construction milestones. If the tenant walks, the asset becomes a $2 billion data center with no rent check.

CleanSpark is not a cigar butt at depressed levels with hidden asset value. It is a company executing a pivot that, if it works, transforms its economics entirely. The Sandersville lease turns 175 megawatts of power capacity into $330 million in annual, nearly pure-margin NOI — a return profile that mining will never match. But the bridge to that future is $2.3 billion in debt with interest payments that exceed current annual revenue, a construction timeline that extends well into 2027, and a mining business that is struggling today.

The valuation gap is not between price and book value. It is between price and the Sandersville NOI that may or may not arrive. If the facility opens, the tenant stays, and the lease generates what CleanSpark says it will, the debt service is covered with substantial margin. If any of those steps fails, the debt service is not.

That is the wager. The notes are priced. The build begins in earnest. The market will find out by Q4 2027 whether a Bitcoin miner can become an infrastructure landlord.