PowerCompute (PWCM) Q2 2026 Earnings Call Transcript

PowerCompute reported Q2 2026 revenue of $2.1M (+9.8% YoY), but posted a $4.6M net loss and a $2.8M core EBITDA loss, driven by ~$3.0M fair-value losses on digital assets and higher interest costs. Mining performance improved—27.9 BTC mined (+51% YoY) and mining margin rose to 29% from 24.1%—while average BTC price fell to ~$72k. The company refinanced $18M of debt into a 2% APR Arch Lending Bitcoin-backed revolving facility, replacing debt with ~13% blended rates, and is marketing AI/HPC capacity tied to its 26MW power footprint with an illustrative $20M–$50M annual revenue opportunity contingent on execution and customer contracts.

Analysis

The market should treat this as an option on stranded power, not evidence of a scalable AI franchise. The real read-through is that micro-cap miners with energized sites are trying to re-rate into infrastructure assets, but the equity value only holds if they can sign customer-backed MW commitments; absent that, the story is just a cheap-duration capital structure stapled to volatile BTC exposure. The first-order beneficiary is not PWCM’s equity so much as adjacent equipment and GPU supply chain names if even a fraction of these conversions happen; the loser is any lender financing these pivots, because collateral coverage becomes highly procyclical when BTC and equity both move against them.

The refinancing is a mixed signal: lower cash interest helps near-term survival, but a 30-day rolling facility turns liquidity into a weekly negotiation. That matters over days to months because the stock will likely trade on refinance headlines and Bitcoin volatility, while over 6-18 months the key risk is dilution or forced asset sales if renewal terms tighten or BTC falls. The most fragile variable is not mining margin; it is the ability to keep enough unencumbered BTC and cash to avoid becoming a forced seller into weakness.

Contrarian view: the consensus is probably overrating the AI narrative and underweighting execution risk. Big hyperscalers can source power through established operators with cleaner balance sheets, so a small miner’s “power advantage” may be necessary but not sufficient. Tradeable setup favors fading strength in PWCM until there is a signed colocated load or multi-MW contract; the stronger structural long is the GPU/picks-and-shovels complex, not the speculative converter of power into compute.

Key Decisions for Investors

• Short PWCM on post-call strength or buy put spreads for 1-3 months; thesis is capped unless management announces signed AI/HPC contracts or materially longer debt tenor. Falsifier: first recurring customer revenue from colocation or a renewed facility on better terms.
• Pair trade: long NVDA / short PWCM for 3-6 months as a way to express AI capex growth without taking balance-sheet execution risk. Risk: if PWCM secures a credible multi-MW tenant, the short can squeeze hard.
• Avoid adding to GLXY here; the near-term takeaway is that Bitcoin-backed lending remains a refinancing treadmill, and any BTC drawdown can quickly reprice lender risk. Reassess only if PWCM demonstrates stable collateral coverage through a full BTC volatility cycle.
• Watch-item, not trade yet: any announcement of a signed 4 MW+ colocation contract or containerized deployment timeline. That would be the first real catalyst to justify a long, but until then the equity is narrative-driven.