WULF Q2 2026 - Post-Call Earnings Update

Event: Quartr Q2 2026 event Retrieved: 2026-08-05T14:30:00Z Comparison mode: maximal Scorecard contract: v1

Consensus scorecard

Carried forward unchanged from the preliminary report for this event.

Management and Q&A

• The largest new economic disclosure was the Fluidstack lease modification, which the release did not mention. CFO Patrick Fleury said CB-3 carried "about 43 tenant fit-out requests that were design changes," that the cost impact was "about $150 million at Lake Mariner that TeraWulf... agreed to fund," and that the recovery over the ten-year lease term is "over $300 million of rent" at "a mid-teens return in line with the yield on cost on the lease on those items."
• He combined that with a capacity increase not disclosed in the release: "if you take those changes along with the move from 162 to 168 critical megawatts, it's over $500 million of additional revenue over the initial lease term."
• Fleury explicitly chose equity over debt for that funding: "we very easily could have debt financed those. We specifically chose not to," and framed the effect as improving "the amortization and de-leveraging profile of the project."
• He gave the project capital structure in numbers. The WULF Compute project "was 75/25 debt equity. It's now being equitized even more, closer to 68/32," against peers he described as "levered 95% or 100%." His summary: "just because you can, doesn't mean you should."
• Fleury laid out the return hurdle arithmetic directly: "we are very open with our customers that we need a mid-teens return... If you take the way we finance projects, call it at roughly 80/20... If your debt costs 6%, 6.5%, that's five points of WAC. If your equity costs 25%, that's another five points of WAC. Your WAC is 10%." He said he has "seen some folks in our market signing deals that appear where their WAC is actually in excess of the yield they're earning on the lease. Like that doesn't work."
• On liquidity, Fleury said "We have a billion and a half of cash on balance sheet today," following the April equity raise and the joint-venture sale, and that "We don't need to run to the markets anytime soon." He expects to return to the debt capital markets to finance Hawesville, and for Muskie "probably first half of next year."
• On the Anthropic lease credit support, Fleury would confirm the fact but not the name: "the Hawesville lease will be supported by an investment-grade credit."
• He also disclosed the competitive process behind the lease: TeraWulf bought Hawesville in February and leased it "less than 6 months later," and "there were a bunch of other competitors for that capacity in that process that we have moved over to Eastern Kentucky, at Muskie." CEO Paul Prager separately called Anthropic "a multiple repeat customer."
• On the Kentucky utility structure, Nazar said Kentucky Power and AEP "have signed up a gigawatt of capacity with us under an LOA, a transmission agreement, then they go find the generation to support that," and that TeraWulf commits to the transmission build-out plus "a backstop for the energy." He confirmed "the credit that we posted to AEP was for the energy for the initial 500 MW allocation as well."
• The 250–500 MW annual contracting target was defended on capital and labor grounds rather than demand. Nazar: "Each of these projects at the upper end of the range is still nearly $5 billion of total capital... At our site in Lake Mariner, we peaked at over 1,000 electricians at the site. As you scale up, the ability to kind of procure the requisite labor gets more challenging."
• He also warned against straight-line modeling: "these large-scale infrastructure projects may not just work on a linear scale... Given the size of these projects, we think there's going to be more fits and starts to it."
• Prager stated the core screening principle twice in different words: "a position in the queue does not represent access to power," and "We're all about available contracted power."
• Asked to size the total opportunity, Prager declined: "I don't have an answer for you. Demand is extremely strong. At our Muskie site, we have the most active data room we've ever had, and with world-class credits as potential customers. I don't focus on that. I'm trying to focus on execution."
• Nazar gave a demand-timing datapoint: "That 2028 power, we think is becoming more and more kind of what's left on the table. Most of that 2027 capacity is either sold or pretty close to being sold."
• On labor and cost protection, Nazar separated the two inputs. Equipment: "we have these 12-month rolling forecasts with our various vendors... we've got deposits down and have firmed up those prices." Labor: "on a cost per hour basis, that's been generally moving up... it's difficult to kind of pin that down exactly, especially for a construction cycle that's 12 to 15 months long," and "that's where I think we have the most work to do. It's not a perfect science."
• Prager credited the EPC relationship for cost control: Fluor is "a top quality, world-class EPC that we have a history with on the power side," and the existence of a reference design now lets subcontractors "take more risk with us, and give us a fixed price contract."
• On the development pipeline, Prager said one of the three-to-five sites in final due diligence "is certainly going to be our first effort at international data center development," targeting "customers that want to move to some of the better opportunities in Northern Europe," with the site search led by Kerri Langlais. He added that the company likes "the Midwest quite a bit" and "the Southeast quite a bit," and studies West Virginia.
• Prager argued that state-level political action is a competitive filter in TeraWulf's favor: "as a result of what's happening in Texas and in New York... A lot of projects are going to fall away because they're not going to be credible, or they won't have the requisite experience to develop it. Or they won't have the cash to sort of post a collateral to get surety of power." He said New York policy makes "Lake Mariner and Cayuga Lake Hawkeye... worth so much more money today."
• Fleury framed the Texas grid constraint numerically: "The backlog in Texas... is now over 400 gigawatts, and batch zero is roughly 60 gigawatts... You basically have to build another California in the Texas grid to meet all of the batch zero projects. Combined cycles are 500 megawatts each, roughly, and take three years on average to build."
• Prager made regional diversity a stated design principle, using a submarine-design analogy, and said "Our customers don't want to be vulnerable to something that happens in one grid, one region, one fuel source," adding that "as you think about the move towards inference, I think folks want to move to the market."
• On tenant credit, Fleury said the screen extends years out: "These are 20-year leases... as you've seen all the hyperscalers take on more debt, their balance sheets today are not going to be their balance sheets in five, six years."

Guidance and KPI clarification

Updated neutral analysis

• The call did not change any reported figure. It added the single most important economic disclosure of the event — the Fluidstack lease modification — which appears nowhere in the release.
• That modification is net positive but reveals a cost dynamic worth tracking. TeraWulf absorbed roughly $150m of tenant-driven design changes at Lake Mariner and recovers over $300m of rent for them across ten years at a mid-teens return. Combined with a 162-to-168 MW capacity step, management puts the total at over $500m of incremental lease revenue. The economics work, but the mechanism is that the tenant requests changes, the landlord funds them, and the rent is re-struck — 43 change requests on a single building.
• Choosing to equity-fund those changes rather than debt-finance them, and moving the WULF Compute structure from 75/25 to roughly 68/32, is a deliberate de-levering. Against a balance sheet carrying $7,900.9m of total liabilities on $147.3m of stockholders' equity, that is a meaningful signal about how management intends to fund the next phase — and it also means the equity layer is being consumed rather than the debt capacity.
• Fleury's WACC arithmetic is the clearest statement of underwriting discipline made on the call, and it is testable: mid-teens yield on cost against a roughly 10% blended cost of capital assuming 6–6.5% debt and a 25% equity cost. He explicitly accused unnamed competitors of signing leases below their own cost of capital.
• The cash figure requires care. Management said approximately $1.5bn on the balance sheet on the call date, against $2,619.2m of cash and $3,028.6m including restricted cash at 30 June. The gap is a month of capital deployment plus whatever is restricted or committed, and the call did not reconcile it. On management's own number, liquidity has fallen by roughly $1.1bn from the reported cash line in about five weeks, before the ~$530m Abernathy proceeds arrive.
• Demand commentary was consistently strong and consistently unquantified. Prager refused to size the total opportunity, cited the Muskie data room as the most active in company history, and Nazar said most 2027 capacity is sold or near-sold with attention moving to 2028. None of that converts into a contracted figure beyond the Anthropic lease already disclosed.
• The 250–500 MW annual target was defended as a supply constraint, not a demand ceiling — roughly $5bn of capital per project at the top of the range and a labor market where Lake Mariner peaked above 1,000 electricians. Nazar's warning that these projects "may not just work on a linear scale" and will have "fits and starts" is an explicit caution against modeling smooth quarterly capacity additions.
• Labor cost is the acknowledged weak point in the cost structure. Equipment is largely locked at lease signing through rolling forecasts and deposits; labor per hour is "creeping up" across a 12-to-15-month build with only "some parameters" around it. Management said plainly this is where it has "the most work to do." That is the input most likely to pressure the $8–10m per MW guidance.
• The competitive framing — that Texas and New York policy action will remove less-credible developers — is a plausible argument that management has an obvious incentive to make. The supporting Texas figures (400 GW backlog against 60 GW in batch zero) are cited from market data, not company disclosure.
• The Anthropic credit support remains unnamed. Management confirmed only that it will be an investment-grade credit, so the counterparty backing approximately $19bn of contracted revenue is not yet identifiable from any source in this event.
• Notably absent from the entire Q&A: no analyst asked about the $755.7m warrant fair-value charge that produced the $(1.94) loss per share, about the Adjusted EBITDA swing to $(18.3)m, about the $83.9m of quarterly stock-based compensation, or about the 73.1% decline in bitcoin mining revenue. Six analysts asked questions and all six focused on capacity, contracting, and financing.
• One analyst referred to Muskie as a "2-gigawatt" site; management neither confirmed nor corrected the figure, and the release states up to 1 GW of contracted electric service from Kentucky Power. The 1 GW figure is the only one supported by a document.

Market context and limitations

• API Ninjas snapshot: WULF $19.24 on NASDAQ, volume 5,718,453, quote timestamped 2026-08-05T13:50:18Z, retrieved 2026-08-05T13:51:00Z. That quote was taken before the U.S. open and while the 8:00 a.m. ET call was under way, so it does not reflect the call. No post-call quote is presented here.
• Every reported actual carried into this update comes from the Quartr Q2 2026 earnings release document; every quotation and attributed statement comes from the Quartr transcript of the 2026-08-05 TeraWulf earnings call for event 663740.
• The Q&A section of that transcript was read in full — all 76 paragraphs. Prepared remarks were not read verbatim; prepared-remarks content in this update is taken from the earnings release rather than the transcript.
• The Quartr transcript for this event carries no speaker-name labels. Attributions above rest on speakers self-identifying in the text ("this is Patrick Fleury," "It's Paul," "This is Nazar here," "It's Naz here"). Paul Prager's title as Chairman and CEO and Patrick Fleury's as CFO come from the earnings release; Nazar's role is not stated in either source and is therefore not asserted here.
• The consensus scorecard is carried forward from the preliminary report without recalculation. Stage claims forbid corrections, so no scorecard revision is made here.
• The approximately $1.5bn cash figure is management's spoken statement on the call date and is not a reported balance-sheet figure; the reported figures are $2,619.191m of cash and cash equivalents and $3,028.608m including restricted cash at 30 June 2026. The two are not reconciled in either source.
• The 162-to-168 MW Fluidstack capacity change, the approximately $150m of funded design changes, the over-$300m rent recovery, the over-$500m total revenue increment, and the 75/25-to-68/32 capital structure shift are all management statements on the call with no corresponding disclosure in the release.
• Slide references made on the call (for example "slide 16, the phase four, the three to five pursued sites") point to a presentation not read for this report; the 2.9 GW pipeline figure was quoted by an analyst and is not verified against a document here.
• No analyst question addressed the warrant fair-value charge, Adjusted EBITDA, stock-based compensation, or the decline in mining revenue, so this update contains no management commentary on those items.

Source links

• Quartr Q2 2026 earnings call transcript
• Quartr Q2 2026 earnings release
• Quartr Q2 2026 slides
• WULF Q2 2026 preliminary report