The central story in crypto mining today is the widening gap between how Wall Street values a megawatt leased to artificial intelligence tenants and a megawatt reserved for traditional Bitcoin mining. According to VanEck’s latest valuation framework for publicly traded miners, contracted AI and high-performance computing capacity attracts a markedly higher multiple than uncontracted power tied to mining pipelines, a divergence that now shapes pricing across the sector.

VanEck finds that miners with signed AI or HPC leases trade at well above 10 times gross energized power, while peers with little or no contracted capacity trade nearer 2 to 6 times that same metric. In practical terms, investors are separating a megawatt backed by a lease from a megawatt pointed at mining or waiting to be sold, and they are paying up for the contractual certainty that AI customers can provide. The leased megawatt, in this view, is a distinct and more valuable asset class than mined Bitcoin or unsold power capacity.

Market Impact

The premium is arriving ahead of the actual buildout. VanEck estimates that only about a quarter of leased AI and HPC capacity has been delivered across the peer group. That means public equity markets are already capitalizing contracts and future construction outcomes that many operators have yet to execute, supporting valuations before the infrastructure is fully energized and generating recurring revenue.

The funding math underscores this timing mismatch. VanEck points to a near-term construction financing shortfall approaching $50 billion across the group. If the broader pipeline of announced projects is ultimately converted into completed sites, total capital needs climb toward roughly $221 billion. The industry has therefore entered a phase in which the market’s willingness to ascribe data-center style multiples to miners is contingent on closing a sizable financing gap and translating contracts into delivered megawatts.

Valuation Mechanics

VanEck’s framework anchors value in a data-center landlord model. It assumes a baseline net operating income of about $1.5 million per megawatt for AI and colocation sites and applies a 15x enterprise value multiple to that cash flow. On a gross basis, that implies roughly $22.5 million in enterprise value per megawatt.

Construction costs are then deducted to approximate pre-financing equity value. VanEck uses a greenfield estimate near $10 million per megawatt, rising toward about $12 million for later-phase projects as inflation works through equipment, labor, and interconnection. That math implies an initial pre-financing value around $12.5 million per megawatt at the lower capex level, before any probability discounts for construction risk or financing costs are applied.

Sensitivity to execution is significant. A few million dollars of additional capex per megawatt, or a delivery schedule that slips by a year, can quickly reshape the equity value attached to a given site. Tenant credit quality also feeds directly into discount rates: VanEck’s framework suggests that leases with investment-grade hyperscalers can be underwritten with an effective cost of capital in the 6% to 10% range, whereas smaller GPU cloud tenants may warrant rates above 10%, compressing the premium investors are willing to pay for those megawatts.

AI Integration

For miners, the shift toward AI and HPC leasing reframes core assets. Power-dense facilities, land positions with expansion rights, and interconnection queues—historically optimized for hash rate—are being repurposed as inputs for AI infrastructure. A signed lease and an energized megawatt are no longer interchangeable: once counterparty strength is considered, the same unit of power can carry meaningfully different valuation outcomes depending on who is on the other side of the contract.

This repositioning aligns with broader trends that highlight the scale of demand for compute-ready power. Industry projections cited alongside VanEck’s analysis include an International Energy Agency view that global data center electricity consumption could roughly double by 2030, with AI-specific usage tripling, and a McKinsey estimate that overall data center spending could reach about $7 trillion by the end of the decade, with the majority earmarked for AI-capable facilities. Recent initiatives, such as KKR’s $10 billion AI infrastructure venture with Nvidia and Vistra, likewise reflect institutional appetite for power-backed AI capacity as a standalone asset class.

Financing the Gap

Bridging the estimated $50 billion near-term shortfall is already pulling miners toward infrastructure-style financing. Project finance and traditional debt can match long-dated lease revenues but also introduce fixed obligations onto balance sheets historically exposed to volatile mining margins. Selling down Bitcoin treasuries converts a core strategic asset into construction capital and may dilute the original investment thesis that attracted Bitcoin-focused shareholders. Strategic partnerships and tenant prepayments can reduce upfront equity needs, yet often shift a portion of eventual upside toward the capital providers that make delivery possible.

Each pathway carries trade-offs that feed back into valuation. Structures that lower the cost of capital and preserve equity participation tilt outcomes toward the premiums now embedded in AI leases. Conversely, higher-cost debt, rising project budgets, or heavy equity issuance can absorb a larger share of future cash flow before it reaches existing shareholders.

Bitcoin Linkage Persists

Despite the AI pivot, miner equities continue to track Bitcoin closely. VanEck notes that the peer group’s average one-year weekly beta to Bitcoin sits near 1.05, indicating moves that still mirror Bitcoin’s price action. Yet balance-sheet exposure to Bitcoin is concentrated. MARA’s holdings are estimated at roughly 51% of its market cap, CLSK around 24%, RIOT near 11%, and HUT about 7%, while most other peers hold Bitcoin equal to about 1% or less of market capitalization. The result is a pricing dynamic in which AI-focused operators can look undervalued during Bitcoin selloffs, while pipeline-heavy names can appear expensive in rallies, regardless of where their leased capacity or construction progress stands.

Governance and Execution Risk

VanEck’s governance scorecard examines insider ownership, KPIs, compensation, leadership tenure, and related-party transactions and finds no miner approaching a perfect mark, with HIVE and BTDR ranking lower on a relative basis. This matters because funding tens of billions for AI infrastructure requires confidence in management teams to allocate capital at scales far beyond those of a pure mining era. Weaknesses that carried less weight in a hash-rate business become material when operators are selling power to hyperscalers under multi-year agreements.

Two Paths Ahead

VanEck outlines a bull and a bear case that both hinge on delivery. In the constructive scenario, miner valuations converge toward the models used for data-center REITs and infrastructure landlords. Hyperscaler demand for power-dense, interconnection-ready sites remains firm, financing opens for creditworthy projects, and the leaders report delivered megawatts and recurring lease revenue. If delivered capacity continues to command double-digit multiples of energized power, today’s premium could be validated by tomorrow’s cash flow.

The more cautious path centers on dilution and cost escalation. If capex per megawatt rises above the baseline due to labor, equipment, or grid expenses, and if debt is priced for a sector with limited operating history as an AI landlord, miners may lean on equity issuance or Bitcoin monetization to bridge the gap before AI revenue arrives. In that case, lenders, strategic partners, and new equity buyers who entered after dilution could capture a larger share of the eventual upside.

Ultimately, the test is straightforward: delivered megawatts relative to leased megawatts, the credit profile of each tenant, the actual capex required when projects break ground, the financing structures chosen, and the strength of governance to allocate capital at infrastructure scale. Wall Street has already signaled that miners are worth more as AI infrastructure providers than as Bitcoin producers. What remains unresolved is whether investors are paying for AI cash flow that has yet to appear, or for a construction pipeline that still requires tens of billions of dollars before it can become AI revenue at all.