Hive Digital’s $297.8M Revenue Jump Signals a Calculated Pivot From Bitcoin Hashrate to AI Compute

Hive Digital is leaning into a simple idea: turn cheap, green power into higher-yield compute. The numbers show traction in Bitcoin; the strategy points squarely at AI.

For the fiscal year ended March 31, revenue climbed 158% to $297.8 million, powered by a fourfold increase in mining capacity and a stronger Bitcoin tape. Hive mined 2,885 BTC, more than double the prior year’s 1,414 BTC, with an average Bitcoin price around $98,000 versus roughly $75,900 a year earlier. Installed hashrate closed at 25.1 exahashes per second across Canada, Sweden, and Paraguay, all running on green energy.

Even with better unit economics, Hive exited the year holding 150 BTC—about $10 million—down from 481 BTC as of December 31. That drawdown likely reflects disciplined liquidity management to fund growth and operations rather than a directional call on price.

The company’s higher-growth narrative sits in its computing arm. BUZZ HPC posted $19.5 million in revenue, up 94% year-over-year, and management is positioning it as the engine to smooth cyclicality and expand margins. The plan: a 320-megawatt AI data center in the Greater Toronto Area designed to accommodate more than 100,000 Nvidia GPUs at full buildout—pitched as Canada’s largest privately owned AI infrastructure project. Hive is targeting $660 million in annualized recurring revenue from computing by the end of 2028.

Here’s the bet worth scrutinizing: translating interruptible, commodity-like Bitcoin mining into contracted, premium AI compute without losing the core advantage—power arbitrage. Mining thrives on flexibility; you can curtail, relocate, and upgrade as economics shift. AI clusters demand uptime, dense interconnects, and capacity reserved years in advance. That flips the risk stack: - Procurement and deployment: securing 100,000+ GPUs and building 320MW on time requires exceptional vendor access and construction execution. Slippage can erase pricing windows fast. - Power and cooling: unlike miners, AI workloads need consistent baseload power and advanced cooling. Hive’s green energy footprint helps, but reliability and Power Usage Effectiveness will drive unit economics. - Revenue quality: AI ARR depends on utilization and contract length. Pre-sold capacity with credible counterparties can de-risk cash flows; spot exposure could be lucrative in a squeeze but volatile if GPU supply loosens. - Asset durability: GPUs depreciate differently than ASICs. If software efficiency, new architectures, or pricing pressure bite by 2028, payback periods could stretch unless contracts are locked and indexed.

Financially, the year was not clean. Hive reported a GAAP net loss of $148.4 million, with approximately $221 million of non-cash items, including depreciation, flowing through the P&L. That mix hints at a heavy investment cycle and accounting headwinds common to capital-intensive infrastructure builds.

Market reaction was muted. HIVE shares were recently down about 2.6% at $4.63, even after touching a year-to-date high of $4.97 earlier in the session, per Yahoo Finance. Investors often fade miner-to-AI narratives until they see firm offtakes, power contracts, and a credible capex cadence.

What I’m watching next: - Concrete milestones on the Toronto campus: power interconnection timelines, cooling design, and phased capacity dates. - GPU procurement and networking architecture: supply commitments, cluster topology, and interoperability choices that dictate workload mix—and pricing power. - Contracting strategy: length, pricing floors, and counterparty quality for AI compute that can bridge from aspiration to the $660 million ARR target. - Balance-sheet discipline: how Hive funds buildout while maintaining optionality in Bitcoin operations and managing BTC treasury prudently.

If Hive can port its power-first DNA into a high-availability AI stack—without surrendering cost leadership—it could turn a cyclical mining business into a more diversified compute platform. The ingredients are there: scale power, green energy, and a miner’s bias for efficiency. Execution will decide whether this is a smart rotation or just a narrative hedge.