Bitcoin miners trying to morph into AI infrastructure operators are running into a hard math problem. In a research note, VanEck analysts Griffin MacMaster and Matthew Sigel argue the market is starting to separate companies that are actually delivering capacity from those still selling plans.
The headline number is a roughly $50 billion near-term funding gap. VanEck frames it as the spread between sector pipeline ambitions and what miners can fund and build with their current cash positions.
“Energized power” beats megawatts on paper
VanEck calls out a valuation problem created by the sector’s blur. More companies now sit somewhere between Bitcoin mining and AI data center hosting. VanEck says the most reliable metric it can use today is gross energized power, essentially how many megawatts a company has switched on, not just announced.
In VanEck’s comparison, firms with physical leases in hand trade at much higher multiples of gross energized power. The note cites Cipher Mining, Hut 8, and TeraWulf, each trading above 10x gross energized power.
Other miners that remain more tightly tied to Bitcoin mining and show limited contracted AI capacity trade lower. VanEck points to Marathon Digital and CleanSpark at roughly 2–6x gross energized power.
The analysts summarize the pattern bluntly. “For now, we find that the market is paying for contracted and energized capacity, while discounting everything still in the pipeline.”
That means signed contracts are not enough by themselves. VanEck warns the market is already pricing the gap between paperwork and execution.
Execution gap becomes the valuation driver
VanEck says across the peer group, miners have delivered only about 25% of their leased capacity. It expects that delivery figure to fall further before improving, as large-scale construction begins to ramp in 2027 and 2028.
As that construction schedule hits, VanEck argues the dominant valuation driver shifts from capacity announcements to construction milestones. Companies that miss milestones face what the firm calls “structural de-ratings.”
VanEck also flags a less-talked-about risk. Very few of these companies have prior experience building the kind of infrastructure AI customers require. In the note, project management credentials matter almost as much as megawatt counts, because delays and cost overruns can punish valuation even when power eventually gets energized.
Lease activity and near-term deadlines readers should watch
VanEck’s deal tracker implies a busy second half of 2026, with multiple miners in active or advanced lease negotiations. The note names Bitdeer, HIVE Digital, Riot Platforms, and Core Scientific.
It also describes TeraWulf as in advanced negotiations on a 480MW site in Kentucky, with the expectation that a customer lands in the second quarter.
Here is the capacity and funding snapshot VanEck lays out in the research note.
| Theme | What VanEck says | Examples named |
|---|---|---|
| Near-term funding gap | Roughly $50 billion above current cash positions | Applies to the sector overall |
| Long-term capex needs | Approaches $221 billion | Sector-wide estimate |
| Energized power valuation | Market pays for contracted and energized capacity | >10x: Cipher Mining, Hut 8, TeraWulf. 2–6x: Marathon Digital, CleanSpark |
| Delivery of leased capacity | ~25% delivered across the peer group | Applies to the group as a whole |
| Construction ramp | Large-scale construction starts in 2027–2028, delivery may worsen first | Execution risk emphasized |
| Deal activity | Multiple companies in active or advanced lease negotiations in 2H 2026 | Bitdeer, HIVE Digital, Riot, Core Scientific |
Where the money comes from, and where it doesn’t
VanEck estimates capital demands are large enough to force hard trade-offs. It pegs long-term capital expenditure needs at about $221 billion. The near-term version of that story is the roughly $50 billion funding shortfall.
The pressure is not evenly distributed. VanEck says HIVE faces the most acute strain relative to its market cap, driven by its AI Gigafactory ambitions targeting more than 100,000 GPUs. The note also flags IREN and KEEL as having the next heaviest near-term funding loads.
By contrast, VanEck describes WULF and Cipher Mining as relatively better positioned. The analysts attribute that to contracted anchor deals that help de-risk capital raises.
Funding routes differ, too. VanEck highlights that miners with Bitcoin treasury holdings can lean on Bitcoin monetization strategies. It lists MARA at 35,303 BTC, CleanSpark at 13,561 BTC, and Hut 8 at 13,696 BTC.
One of the stark contrasts in the note is REN, which VanEck says has a large near-term funding need but no BTC treasury. For those companies, VanEck points to fewer options, mainly dilutive equity issuance or incremental debt.
Betting on Bitcoin is no longer a clean proxy
VanEck also challenges the way the market treats the sector as a single Bitcoin proxy. The analysts cite an average daily-return correlation to BTC around 0.55 year-to-date and an average one-year beta around 1.05.
But VanEck argues that figure overstates true Bitcoin sensitivity for companies whose value has started to drift away from mining-only dynamics.
In the note, only a few companies carry meaningful balance-sheet exposure to Bitcoin price swings. VanEck says MARA has BTC-sensitive value equal to about 98% of market cap, CleanSpark around 53%, and Riot around 23%.
Other names, including Core Scientific, TeraWulf, Applied Digital, and IREN, have effectively decoupled in VanEck’s framing.
VanEck illustrates the mismatch using its own scenario. It says a drop in Bitcoin to $50,000 would erase roughly 45% of MARA’s equity value and nearly 50% of HIVE’s. It would reduce Hut 8’s by about 4%.
That’s the point. A single-BTC-trade framing can miss the increasingly divergent asset story.
What comes next in the “AI miner” valuation model
VanEck expects the sector’s valuation approach to move away from megawatt counts. In its research note, the firm says the market should eventually shift toward delivery ratios, unit economics, and discounted cash flow models.
At that stage, VanEck expects miners to resemble data center REITs more than pure miners. The note even suggests companies could be sold or converted into REITs once AI revenue matures.
For the near term, VanEck says the biggest re-rating potential sits where the ambition-market pricing gap is widest. It names HIVE, KEEL, IREN, and Bitdeer, while stressing that those same companies carry the highest execution risk.
More conservative positioning, in VanEck’s view, belongs to names with anchor deals already in hand, including WULF, Cipher Mining, and Hut 8.
All of this comes with asset risk. Capacity can be delayed. Cash can run hot. AI infrastructure demand may not line up with build schedules. VanEck’s framework simply makes those risks easier to price than vague “pipeline” claims.