Initiated 2026-10-01 · Price $14.80 (as of 2026-09-30 close, stockanalysis.com) · Mkt cap $7.38bn · Technology / AI data-center leasing (former bitcoin miner) · Model: verified

Rating: Sell — Conviction: Low

TeraWulf has signed real, long-dated leases with Fluidstack (Google-backstopped) and Anthropic, but at $14.80 the market already pays for all of them plus a large uncontracted pipeline. On contracted leases alone, ~$5.5bn of remaining build, $5.1bn of debt and the Google warrants leave too little equity value to support the price. The stock needs new contracts and smooth financing just to hold where it is.

Conviction tests (3a-v-c): T1 pass (base −82.0% at +1pp, −63.5% at −1pp; Sell at both) · T2 pass (Gordon $1.67, exit multiple $6.09; both Sell) · T3 fail (Google warrant count partly from press, not a filing) · T4 fail (pending ~$3.5bn Kentucky debt raise and new-lease announcements each could move value >25%)

Model value range vs price
Bear $0.00Base $3.88Bull $19.12Price $14.80

Business overview

TeraWulf is a bitcoin miner turning into a landlord of AI data-center capacity, leasing powered, cooled shells ("critical IT MW") to GPU operators who own the chips. FY2025 revenue was $168.5mm, of which $151.6mm was mining and $16.9mm HPC leasing. By Q2 2026 leasing was 71% of revenue ($31.9mm of $44.8mm). Lake Mariner (NY) had 102 MW live after CB-3 in July, and CB-4/CB-5 (336 MW) are due in phases from late 2026 into 2027. Justified Data (KY) has a 20-year, ~401 MW lease with Anthropic worth ~$19bn, delivering late 2027 to early 2028. Earnings drivers: MW delivered on schedule, rent per MW-year (~$1.6–2.4mm), and the cost of funding ~$8–10mm/MW of construction.

Competition

The thesis rests on winning and financing hyperscale leases. Here TeraWulf competes with other converted miners, Cipher Digital, IREN and Applied Digital, and with private developers. All three are loss-making and spending heavily; trailing revenue is $611mm at Applied Digital, $707mm at IREN and $191mm at Cipher (stockanalysis, 2026-09-30), against TeraWulf's $165mm. The pricing datapoint is TeraWulf's own contracts: ~$1.86mm per MW-year on Fluidstack (Aug 2025) and ~$2.37mm on Anthropic (Jul 2026, average including escalators). Pressure would show first in rent per MW on new signings and in financing spreads. On trailing EV/Sales the group runs CIFR 59.3x, IREN 25.5x and APLD 17.4x against WULF's 60.6x, so TeraWulf is priced at the top of the group on revenue that is still mostly ahead of it.

Bull case

  1. Contracted ramp delivered on time — Lake Mariner fills to ~440 MW through 2027 and Anthropic reaches 401 MW by early 2028, lifting revenue from ~$204mm (FY2026 est.) to ~$1.5bn in FY2028. Plays out if CB-4 and CB-5 are commissioned on the H2 2026 / early 2027 schedule and the Kentucky campus is energised by late 2027. Model: rev_growth
  2. Pipeline converts — Muskie (up to 1 GW, power Q4 2028), Chesapeake (up to 1 GW) and Lake Hawkeye (~320 MW) give a pipeline of about 2.9 GW. Another ~250 MW contracted at ~$2.2mm/MW-year would add ~$550mm of revenue by FY2030. Plays out if AI capex stays strong and at least one new lease is signed in 2027. Model: rev_growth, capex_pct_rev
  3. SBC normalises — SBC was $185mm in H1 2026, more than twice revenue. If it fades as the build ends, GAAP EBITDA margins approach ~74% on lease revenue with ~12% direct costs. Plays out if FY2027 SBC is below $200mm. Model: ebitda_margin
  4. Valued as lease paper — with Google's $3.2bn backstop and investment-grade support expected on Anthropic, the cash flows could earn data-center multiples (17x in Bull). Plays out if the Kentucky debt prices tight and the market treats tenant credit as near-investment-grade. Model: exit_ev_ebitda

Bear case

  1. Delays and no new contracts — transformer, cooling and backup-power lead times push CB-5 and Anthropic out ~two quarters; revenue plateaus at ~$1.44bn and the multiple compresses. Plays out if CB-4 lease commencement slips past Q4 2026 or the 2027 contracting target is missed. Model: rev_growth, exit_ev_ebitda
  2. Build cost overruns — ~$5.5bn of remaining build (est.), funded by debt and equity. A 15% overrun adds ~$0.8bn before any revenue arrives. Plays out if per-MW cost moves above the company's $8–10mm range. Model: capex_pct_rev
  3. Dilution and SBC persist — shares rose ~19% in seven months (420.1mm to 499.0mm). If SBC stays near $370mm a year, mature margins stay near 60% rather than 70%. Plays out if SBC and share issuance don't fall after the build. Model: ebitda_margin
  4. Financing and tenant credit — $5.1bn of debt, ~$3.5bn more to raise, and lessees (Fluidstack, Anthropic) that rely on backstops and private capital. Plays out if high-yield spreads widen or a neocloud tenant shows stress. Model: wacc, exit_ev_ebitda

Valuation & balance sheet

Metric (definition) Current Own history (range or 5y avg) Peers Source, as-of
EV/EBITDA (model basis: mkt cap − cash + debt + warrants net of Abernathy receivable; GAAP EBITDA) n/m (EV $10.45bn; FY2025 EBITDA −$97.6mm) n/m (EBITDA negative FY2023–25) n/m (peers loss-making or near zero) Model; 10-Q 2026-06-30
EV incl. remaining build / Base FY2028 EBITDA ~16x ($15.9bn / ~$1.0bn) n/a n/a on this basis Model (est.)
EV/Sales (aggregator, trailing) 60.6x n/a (unverified) CIFR 59.3x · IREN 25.5x · APLD 17.4x stockanalysis 2026-09-30
Contracted revenue (initial terms) ~$6.7bn Lake Mariner (Fluidstack) + ~$19bn Anthropic $0 before 2025 n/a (unverified) 8-K Aug 2025; Q2 2026 release

Model-implied value range (from model-summary.json; generic module, Gordon-growth DCF and exit-EV/EBITDA DCF, midpoints): Bear $0.00 · Base $3.88 · Bull $19.12 per share, i.e. implied returns of −100.0% / −73.8% / +29.2% vs $14.80. These ranges show how the bull and bear drivers translate into value; they are not price targets. The price sits two-thirds of the way from Base to Bull, so roughly the Bull case (on-time delivery, new contracts, a data-center multiple) is priced in; Base values only contracted leases. The exit method ($6.09) embeds a 13x Y5 multiple, a ~19% de-rating from the ~16x paid today on forward contracted EBITDA including the remaining build; Gordon ($1.67) embeds a lower terminal multiple. Bear floors at $0.00: debt and warrants exceed a still-positive enterprise value (~$2.5bn). Tail (Anthropic is over half of mature revenue), scratch copy without Justified Data: revenue +21%, +100%, +25%, 0%, 0%; margins 25–50%; 9x exit; 13.3% WACC: also $0.00.

Balance sheet: net leverage not meaningful on negative FY2025 EBITDA. Base path net debt/EBITDA is 8.2x in FY2028 and 6.0x in FY2030, with EBITDA/net interest 2.0–2.4x. Liquidity: $2.62bn cash, $409mm restricted, $530mm Abernathy proceeds due by April 2027. Nearest maturities: 0.50% converts ($612.7mm, in the money at $11.50) and 7.75% secured notes (carrying $3.02bn), both 2030; then converts of $675.9mm (2031) and $814.4mm (2032). Unrated in sources retrieved.

Model note: verified (LibreOffice matched all 3,207 formula cells). Unverified inputs: none. Assumptions without basis: none. No scenario CHECKs. Judgment inputs, labelled (est.): H2 2026 revenue, Core42 rent, the ~$5.5bn remaining build and SBC fade. Balance sheet is at 6/30, so Y1 capex is H2 only; full-year Y1 EBITDA double counts ~$0.50/sh (conservative). Converts at principal ($2.10bn, vs ~$2.35bn as-converted). Warrants at intrinsic value ($1.09bn); the 32.5mm tranche count is from press. No EPS guidance. WACC uses beta 2.0, a stated haircut from the observed 4.25.

Scenario stress test

Reasoned from the bull/bear drivers above. The model column comes from the scenario overlays (Base case + shock).

Scenario Effect Mechanism Magnitude Model Δ value vs Base ($/sh)
S1 Fast equity crash − Beta 4.25, 24% short interest: the multiple de-risks; leases unaffected over weeks High −$3.03
S2 Slow bear / recession − Contracted rent holds, but new leasing stops, the ramp slips and financing spreads widen High −$3.32
S3 Rapid rate shock − Long-duration, back-end-loaded cash flows, ~$3.5bn still to finance High −$2.62
S4 Slow rate grind − Same channel, grinding Med −$1.80
S5 Soft-landing cuts + Cheaper project debt and a lower discount rate on long-dated leases High +$3.22
S6 Recession-driven cuts − AI-spend pullback and tenant stress outweigh lower rates Med −$2.05
S7a Credit liquidity shock − Kentucky high-yield/leveraged-loan raise delayed; equity de-rates High −$2.89
S7b Slow default cycle − Levered neocloud tenants test the Google backstop; refinancing costs rise High −$3.48
S8 Stagflation − Transformer, cooling and labor inflation on ~$5.5bn of remaining build plus higher rates Med −$2.10
S9a Dollar spike 0 No material effect, not modeled (US assets, USD leases) Low $0.00
S9b Dollar slide 0 No material effect, not modeled Low $0.00
S10 Melt-up + High-beta AI infrastructure with heavy short interest: momentum and covering High +$3.85
S11 Energy supply shock − Leases are grid-priced in NYISO/MISO; residual mining power cost and risk-off Low −$0.47
S12 Mega-cap/AI derating − The core risk: an AI-capex unwind de-rates the group, stalls leasing and raises renewal doubts High −$3.88

S2 is ~86% of the Base − Bear gap; downside rows cluster near the floor because Base equity is only $3.88. Currently active/on watch per the playbook: S3 partially active (price leg met; the 2026-09-30 macro log shows the pace legs crossed too); S8, S10 and S11 on watch. S3 and S10 are the live rows here: rates hit the discount rate and the pending debt raise, while a melt-up is the main way the price could hold.

Model value change vs Base, by scenario
S12 Mega-cap/AI derating−$3.88S7b Slow default cycle−$3.48S2 Slow bear / recession−$3.32S1 Fast equity crash−$3.03S7a Credit liquidity shock−$2.89S3 Rapid rate shock−$2.62S8 Stagflation−$2.10S6 Recession-driven cuts−$2.05S4 Slow rate grind−$1.80S11 Energy supply shock−$0.47S9a Dollar spike$0.00S9b Dollar slide$0.00S5 Soft-landing cuts+$3.22S10 Melt-up+$3.85

What would change the call

Upgrades if: a new lease of ≥250 MW is signed at ≥$2mm/MW-year with investment-grade credit support, the Kentucky debt closes at or below ~8%, and CB-4 commences on schedule. Downgrades if: n/a below Sell. Conviction rises if CB-4/CB-5 slip, the Kentucky financing is delayed or priced above ~10%, or SBC and share issuance stay at the H1 2026 run-rate.

Watch items

Sources