On 5 August 2026 TeraWulf's (Nasdaq: WULF) second-quarter release described the nominal value of its signed leases as "~$27B+". The 10-Q filed the same day lists, in the lessor note, "minimum lease payments receivable on commenced leases" of $1.457bn. Same company, same day, a difference of 18.7 times.
Nobody is lying. The two numbers measure two different things. This piece does one job: it separates them, and gives a method anyone can reproduce from the filings.
How contract value is computed
The company's "contracted revenue over the initial term" is the nominal sum of base rent over the initial term: undiscounted, excluding renewal options, excluding power pass-through. All three pieces of evidence sit in the filings: the August 2025 deck describes buildings CB-3 and CB-4 as "~$370MM average annual contracted revenue", which × 10 years = $3.7bn; the Abernathy cash-flow model prepared for note investors shows rent starting at $264m and escalating 3% a year, which summed over 25 years gives $9.63bn, 1.3% from the company's "~$9.5 billion"; the Anthropic lease lists $19bn (20 years) and $33bn (30 years with renewals) separately, so the base figure excludes renewals.
Contract value therefore answers the question: if every building is finished on time and every tenant pays every year in full, how many nominal dollars arrive. Every one of its premises lives inside an "if".
The accounts recognize only what has commenced
ASC 842 requires a lessor to list future minimum lease payments only for leases that have commenced. If the building is not finished and the lease has not started, not a dollar enters that table.
How to find it: open the 10-Q, locate the lessor section of the "Leases" note; the table is usually headed "Future minimum lease payments receivable" or "maturity analysis of lease receivables", laid out by year with a total at the bottom. TeraWulf's table at 30 June 2026: $40m for the rest of 2026, $92m in 2027, $130m in 2028, $142m in 2029, $146m in 2030, $906m thereafter, $1,457m in total. Six months earlier (31 December 2025) the figure was $308m.
Divide it by WULF's $27.2bn share of nominal contract value and you get 5.4%. The ratio does not mean the other 94.6% will fail; it means the other 94.6% depends on things that have not yet happened — the buildings finished, the power connected, the leases commenced. Each has a schedule, and each schedule keeps moving.
Why this number deserves more attention than contract value
Three reasons. First, it is mandated by the standard: the company cannot change the method, and it is comparable across quarters. Second, every jump in it corresponds to a verifiable event — one building commences, and ten years of that building's rent enter the table. TeraWulf's next large jump is the full commencement of CB-4 (168 MW): at $1.9m per MW-year × 10 years, the table should move from $1.457bn to above $4.5bn, and the February 2027 10-K will show whether it did. Third, it compresses three things into one figure — power access, construction funding, build progress — and if any one of them stalls, the number does not move.
Contract value has none of these properties. It is fixed on signing day and does not change for two years no matter what happens.
The same ratio across peers
| Company | Initial-term contract value | GAAP future minimum rent on commenced leases | Ratio |
|---|---|---|---|
| WULF | $27.2bn | $1.46bn | 5.4% |
| CORZ | >$24bn | $6.09bn | c.25% |
| GLXY | ≥$15.1bn | $4.75bn | c.31% |
| RIOT | $9.8bn | $0.63bn | c.6% |
| CIFR / HUT | $11.4bn / $26.6bn | 0 | 0% |
Source: each company's latest 10-Q / 10-K lessor note and earnings release, with MW and contract value aligned to critical-IT and initial-term bases; APLD's lessor table includes leases not yet commenced, a different basis, and is omitted.
The thing to read in this table is not who is high and who is low but stage: CIFR and HUT have not commenced a single building, and a zero says nothing about their contracts' quality; CORZ has delivered 437 MW, so its ratio is naturally higher. How the ratio moves over time at one company carries more information than the cross-section.
The second trap: gross versus critical IT
Megawatts come in two bases too. Gross is grid connection; critical IT is the load the tenant's equipment can actually use. The ratio between them is the PUE, typically designed around 1.25 in this industry, so 100 MW gross corresponds to 80 MW of critical IT.
The same company reporting the same asset in both bases in the same period can be traced line by line in TeraWulf's filings: the November 2025 release described the Fluidstack leases as "450 MW of capacity" while the deck of the same period said "366 MW contracted to Fluidstack", 366 ÷ 450 = 0.81; Core42 appears as 72.5 MW in one place and 60 MW in another; the "22.5 MW HPC capacity" of 30 September 2025 became "18 MW of critical IT" in the 10-K, a ratio of exactly 1.25. Across the sector, CIFR's "700 MW" is gross and 454 MW in critical IT; HUT's 1,330 MW converts to 949 MW.
There is only one rule: when you see a MW figure, ask which basis first. Rent per MW, computed by dividing contract value by capacity, changes by 25% when the denominator's basis changes.
Where these two points sit in the report
These two bases form the skeleton of Chapters 3 and 6 of our TeraWulf (WULF) deep-dive report (first edition): how the five unfiled leases are reconstructed, how nominal contract value and the GAAP receivable are each computed, the three power states at five sites, and what follows when they go into a valuation. The report passed four adversarial audit tracks, with all 25 revisions printed in Appendix C. Methodology at Research method; for the same class of basis problem see Vertiv stopped disclosing orders and One question, six answers on AI market size.
Research content, not investment advice.