TI
TERAWULF INC. (WULF)·Q3 2025 Earnings Summary
Executive Summary
- Q3 2025 was a transformational inflection: total revenue rose to $50.6M (+6% QoQ, +87% YoY) with first HPC lease revenue of $7.2M; Adjusted EBITDA improved to $18.1M (+25% QoQ) while GAAP net loss was driven by a non‑cash $424.6M mark‑to‑market on warrants/derivatives .
- Commercial momentum accelerated: ~594 MW of contracted HPC capacity (Core42 and Fluidstack/Google) at Lake Mariner plus a 168 MW Abernathy JV, with ~$17B+ in long‑term customer contracts and ~$5B+ long‑term financings including $3.2B senior secured notes and $1.025B 2032 converts .
- Guidance/narrative upgrades: annual HPC signing target raised to 250–500 MW; HPC leasing margins expected to normalize toward ~85% in Q4 after a 3Q stub period; mining operating capacity guided to ~7.2 EH/s in Q4 as site power is reconfigured to HPC .
- Liquidity strengthened: quarter‑end cash, cash equivalents, and restricted cash reached $712.8M; total debt ~$1.5B (primarily convertibles); pro forma liquidity >$1B after October transactions .
- Catalysts: execution milestones (CB‑2 delivery around year‑end), Abernathy financing “before year‑end,” and visible multi‑year HPC ramp with Google credit enhancement underpin institutional capital access .
What Went Well and What Went Wrong
What Went Well
- HPC commercialization began: “We recorded our first HPC revenues with lease commencement at Wolf Den and CV1” (CEO) as gross HPC lease revenue reached $7.2M in Q3 .
- Strategic financings at scale: Closed $3.2B senior secured notes backed by Google lease support to fund Lake Mariner HPC buildout, plus $1.025B 2032 convert to support Abernathy JV; management highlighted a repeatable development/financing model .
- High‑quality counterparties: “We have two world‑class credits…Core42 backed by G42, and FluidStack Google” (CEO), with ~$6.7B contracted revenue at Lake Mariner and $1.3B Google credit support for Abernathy .
What Went Wrong
- GAAP loss inflated by non‑cash fair‑value charges: $424.6M loss from change in fair value of warrants/derivatives drove net loss to $(455.1)M; depreciation rose on accelerated miner building depreciation ($7.8M) tied to HPC reconfiguration .
- Mining output down QoQ: BTC mined fell to 377 from 485 in Q2 as operating hash rate was repositioned for HPC; mining capacity expected to target 7.2 EH/s in Q4 .
- HPC margin below long‑term guide in start‑up quarter: HPC leasing segment realized ~72% profit margin due to partial‑quarter revenue and ~$0.7M Cayuga development expense; management expects normalization toward ~85% in Q4 .
Financial Results
Consolidated Results vs Prior Quarters
Segment Performance and Margins
KPIs
Guidance Changes
Earnings Call Themes & Trends
Management Commentary
- CEO: “The third quarter was truly transformational… signing approximately 360 megawatts… backstopped by Google… and closing $3.2 billion in senior secured financing” .
- CEO on demand/priorities: “We recently increased our annual target for new HPC signings from 100–150 megawatts per year to 250–500 megawatts per year… reflecting the strength of customer demand” .
- CFO on profitability: “GAAP revenues increased 6% QoQ to $50.6 million… Adjusted EBITDA improved 25% QoQ to $18.1 million… inclusive of significant increases in operating expenses and SG&A… invested heavily in our HPC business” .
- CFO on non‑cash loss: “Change in fair value of warrant and derivative liabilities in 3Q 2025 was a loss of $424.6 million related to the Google warrants and the conversion feature of the 2031 convertible notes” .
Q&A Highlights
- HPC margins: Actual ~72% in Q3 due to partial revenue and ~$0.7M Cayuga expense; management expects ~85% in Q4 (normalization with full quarter) .
- Customer diversification/credit quality: Focus remains on tier‑one credits (Core42/G42, FluidStack/Google), with ongoing dialogues; credit quality is key to financing terms .
- JV financing & penalties: Abernathy financing expected before year‑end under similar structure; leases have grace periods and scaled penalties; termination cannot occur until >180 days late .
- Mining reconfiguration: Hash rate reduced as site power is reallocated to HPC; accelerated depreciation and miner sales reflect repositioning while maintaining profitability .
- Build costs & procurement: Capex/MW higher than Core42 due to scale and customized designs; rolling procurement schedules with equipment vendors to ensure delivery timelines .
Estimates Context
- S&P Global consensus for Q3 2025 revenue, EPS, and EBITDA was unavailable via our data feed at this time; therefore, comparisons to Street estimates cannot be provided. Analysts will need to incorporate the new HPC segment (straight‑line lease accounting, power passthrough) and Q4 normalization into models based on disclosed mechanics .
Key Takeaways for Investors
- The core thesis is shifting from pure mining to contracted, credit‑enhanced HPC infrastructure with multi‑year, high‑margin, infrastructure‑style cash flows; execution risk is mitigated by Google backstops and agency‑rated debt .
- Near‑term EBITDA trajectory should benefit from full‑quarter HPC revenue in Q4 and target margin normalization to ~85% as Core42 ramps and Akela/La Lupa progress .
- GAAP results will remain noisy due to non‑cash fair‑value items (warrants/derivatives) and depreciation tied to asset repurposing; focus on non‑GAAP segment margins and lease economics .
- Liquidity and capital access are robust (> $1B pro forma), supporting accelerated build schedules and new site acquisitions amid a power‑constrained environment .
- Watch delivery milestones (CB‑2 year‑end), Abernathy financing timeline, and additional site announcements (management hinted at 1–2 sites by year‑end) as stock catalysts .
- Mining remains opportunistic and power‑price sensitive; guidance to 7.2 EH/s reflects strategic prioritization of HPC revenue density over nameplate hash rate .
- The raised 250–500 MW annual signing target suggests sustained demand tailwinds; diligence on grid capacity and permitting remains the gating factor—management’s disciplined funnel is a differentiator .