HIVE expands Canadian AI data center capacity to 16.6 MW

HIVE announced a four-fold expansion of liquid-cooled AI data center capacity to 16.6 MW (from 4 MW) across two Canadian provinces via a Bell Canada partnership, with the BC site providing 5 MW immediately and optional +7.6 MW. Q3 fiscal 2026 revenue was $93.1M, up 219% YoY, but the company carries negative LTM free cash flow of -$138M and remains unprofitable despite $257M LTM revenue (+113%). Management expects to deploy 4,000 GPUs for contracted revenue within six months and targets $200M of contracted annualized run-rate revenue by March 31, 2027, with no additional capex required to secure colocation capacity. Analysts have trimmed price targets (Cantor Fitzgerald to $5, Rosenblatt to $4.50, H.C. Wainwright to $7), reflecting cautious sentiment despite the expansion and record revenue.

Analysis

HIVE’s strategy of leaning on third‑party colocation and staged upgrades shifts the business risk away from heavy upfront capex toward execution and contract conversion. That creates a winner-takes-most dynamic: firms that can rapidly aggregate power, cooling and long-term GPU commitments will exert pricing pressure on smaller, vertically integrated miners and force colo providers to compete on SLA and power terms, not just rack rates.

The primary near-term catalyst is conversion of contracted capacity into billable HPC revenue; failure to hit conversion cadence or margin targets will reprice the equity sharply. Medium‑term risks include GPU spot pricing volatility, large cloud providers undercutting short‑term GPU rental rates, and country‑level tax or regulatory outcomes that can retroactively impair project economics — any of which could reverse optimism within months.

From a supply‑chain angle, sustained demand for dense GPU deployments favors firms providing high-efficiency cooling, transformers, and grid upgrades — an underappreciated source of alpha for suppliers and utilities serving the provinces where deployments cluster. Conversely, legacy ASIC incumbents and ASIC‑dependent jurisdictions face accelerating asset stranding risk as compute economics bifurcate toward contracted HPC revenue streams with higher margin visibility.

Key Decisions for Investors

• Long HIVE via a 12–24 month call spread (LEAP buy funded by selling a nearer‑dated call) to capture asymmetric upside if contracted capacity converts to stable HPC revenue; position size 1–3% NAV, max loss = premium, target >2x return if execution milestones hit.
• Hedge operational execution risk by pairing HIVE long equity (or LEAP) with a short NVDA 3–6 month call spread sized to offset GPU‑price driven rerating; this reduces exposure to semiconductor cyclical moves while keeping upside from contract conversion intact.
• Short ASIC‑centric public miners (e.g., marquee ASIC miners) via 6–12 month equity shorts or buy‑puts — rationale: accelerate downside if ASIC economics deteriorate with GPU migration; size modest (1–2% NAV) given idiosyncratic volatility.
• Event rules: add to HIVE longs on confirmed milestone beats (contracted run‑rate achieved or major colo expansion delivering capacity), cut 50% on two consecutive missed milestones or if GPU spot rental rates decline >30% over a rolling 3‑month window; use Swedish tax/court updates as binary catalyst watch.