Cipher Mining Inc. CIFR

Key data

The business

Cipher runs warehouses full of computers next to cheap electricity, mostly in west Texas. Until recently every machine did one job: mine bitcoin, racing to earn new coins. That business is fading. The reward per machine was halved in 2024, mining got more crowded, and last quarter it brought in just $24.8M, down 43% from a year earlier and near the lowest quarterly revenue in the company's recent history. So Cipher is doing what much of its industry is doing, converting those power-connected sites into data centers it can lease to artificial-intelligence firms that need somewhere to run their chips. It holds land and enough grid power, measured in megawatts, that a single large tenant would dwarf everything mining earns. The moat, such as it is, is the power itself: interconnection queues run years, and Cipher already holds the plug. What it does not yet hold is a signed anchor tenant.

The numbers

Revenue is shrinking while the company spends like it is about to triple.

Every quarter since the Q3 2025 peak of $71.7M has been smaller, and the bottom line is ruled by non-cash swings. The $267.5M loss last quarter and the $734.2M loss in Q4 2025 are mostly the accounting mark on stock warrants, which rises when the shares rise, not cash going out the door. Operating loss last quarter was $78.5M.

The clean signal is not on the income statement at all.

In eighteen months long-term borrowings went from nothing to $5.45B while cash sits at $831.8M. Capex ran $488M in 2025 and is running far higher this year. This is a leveraged construction project wearing a miner's ticker.

On book value you pay 14.7 times, more than double the five-year high of 7.0 and about five times the peer at 3.0. That multiple is not paying for what Cipher owns or earns, both going backwards; it is an advance against 2028. Analysts model revenue climbing from $224M last year to about $953M by 2028 and $1.5B by 2029, the whole jump resting on data-center leases that mostly are not signed. The one thing this memo believes the market does not price honestly is that the lease is being treated as done. The single print that settles it is a filed anchor-tenant contract with a term, a megawatt number and a dollar figure. Until that filing, you own the buildout, not the lease.

Management

The people closest to the company are selling, not buying: zero insider purchases in twelve months against $87.4M sold across sixteen filings, led by V3 Holding, a founder-linked vehicle, unloading about $72M in two days in early June. Plan status is not disclosed in the filings, so I cannot separate pre-scheduled from opportunistic, but the direction is one-way. Pay leans hard on equity, with stock-based compensation near half of revenue over the trailing year, an enormous share that says the company pays its people in a currency the market, not the cash flow, funds. The one constructive tell is deferred revenue growing to $27.8M total as of June 2026, up from $1.7M current portion six months earlier, meaning someone is prepaying for capacity.

How it fails or surprises you

The debt clock (downside). Long-term borrowings hit $5.45B against $831.8M cash and operating cash flow that was negative $208M last year. Interest on that stack runs into the hundreds of millions. If lease income does not arrive on schedule, the next refinancing sets the terms and equity sits last in line. Watch quarterly cash against the interest bill.

Mining runs out before the lease arrives (downside). Mining fell to $24.8M last quarter and keeps sliding. That shrinking stream is the only revenue bridging Cipher to its first big tenant. If bitcoin drops or the buildout slips two quarters, the gap is funded with more shares or more debt. The number to watch is quarterly revenue holding near $25M.

The anchor lease lands (right tail). Peers that converted power sites to AI leases, the CoreWeave and Google-backed deals across the sector, re-rated two to five times on the announcement. Cipher holds the scarce input, grid-connected megawatts, and one hyperscaler contract at the scale it is building would validate the entire $953M-by-2028 model. The market is not paying for a signed deal today. Watch for an 8-K naming a tenant.

Closing thoughts

A specific print settles this. Nothing about the next mining quarter matters much; what matters is whether Cipher turns its power into a signed hyperscaler lease with real term and dollars. That 8-K is close to binary: a large anchor tenant converts the stock into a contracted-cash-flow story and 14.7 times book stops looking absurd, while continued silence, another quarter of shrinking mining, and a rising interest bill turn the leverage against the equity. A small deal or a letter of intent without terms leaves you exactly where you are now, paying above every historical valuation multiple for a promise. The near-term left tail is the fatter one, because the debt is real and dated while the lease is hoped-for. What is at risk if the debt clock wins is most of the equity, against a right tail that could be several times the money if the lease lands large.

The bet is still that Cipher rents its cheap-power sites to AI companies and gets paid before the $5.45B in loans comes due. It breaks if mining keeps sliding and no anchor tenant signs, and the one pair of numbers that tells you first is cash, $831.8M and falling, against long-term debt, $5.45B and rising. No signed lease, no thesis.