Mentioned in This Article
1. VanEck Says the October Sell-Off Went Too Far
In bitcoin mining news on Wednesday, November 4, asset manager VanEck said shares of bitcoin miners that have pivoted to AI data center infrastructure are oversold after falling an average of 35% in October. The firm described the decline as a dislocation rather than a rejection of the business model, noting that no signed leases were canceled.
The note matters because it separates two ideas that investors had bundled together last month: worry about how expensive it is to fund AI hardware, and the health of companies that supply the electricity and buildings that hardware needs. VanEck argues that the two are very different, and that the selling hit both groups indiscriminately.
According to VanEck, investors treated companies from quite different parts of the AI supply chain as a single trade. When sentiment toward one slice of the sector soured, the others were sold alongside it, regardless of the contracts or cash flows behind each business.
The term oversold has a specific meaning in markets. It describes an asset whose price has fallen so far and so fast that it may have moved beyond what underlying value supports, which can set up a rebound once selling pressure eases. It is a judgment rather than a guarantee, and VanEck's note is an argument that fundamentals and prices have separated, not a forecast of when they will converge.
2. What Are Powered Shell Providers?
VanEck uses the term powered shell providers, or PSPs, for companies that own or control data center sites with secured electricity and deliver them to tenants who bring their own computing equipment. The firm names Hut 8 (HUT) and Core Scientific (CORZ) as examples of miners that have moved into this role.
The label draws a line between the landlord and the tenant. A powered shell provider supplies the site, the building and, above all, the electrical capacity, while the tenant installs and operates the chips. Their revenue comes from leases rather than from the price of any computing service.
That structure is why VanEck thinks the sell-off was misplaced. The customers of these providers are GPU rental companies, often called neoclouds, which rent out access to graphics processors. Those customers carry the heavy financing burden of buying expensive chips. The providers, by contrast, collect rent under long-term agreements.
For an investor, the practical difference is the source of risk. A GPU rental company is exposed to the price of chips, the interest rate on its borrowing and the demand for the computing it resells. A powered shell provider is exposed mainly to whether its tenants pay rent and whether it can bring new capacity online. Lumping the two together, as October's selling did, treats those distinct risk profiles as identical.
3. Why Did Miners Fall 35% in October?
VanEck points to contagion as the main cause. Concerns about debt costs at GPU rental companies such as CoreWeave (CRWV) spread to the power infrastructure providers that serve them. Neoclouds face higher yields on AI-related debt because they finance costly chips, and when lenders demand more, markets begin to question the whole chain of businesses depending on that financing.
Macroeconomic developments added to the pressure. VanEck cites a hotter-than-expected U.S. consumer price index reading for September, which raised doubts about how soon borrowing costs might ease. Uncertainty ahead of the November midterm elections added another layer of caution for investors weighing risk.
The result, in VanEck's account, was a broad retreat from AI infrastructure names across the board. Companies whose economics differ sharply from those of GPU renters were priced as though they faced the same financing strain, even though their income is tied to contracted leases.
The consumer price index is the most widely followed measure of inflation in the United States. When a reading comes in hotter than economists forecast, markets tend to push back their expectations for lower interest rates, which raises the cost of capital for any business that relies on borrowing. For sectors that need large amounts of financing, such as data center construction and chip purchases, that shift can be felt quickly in share prices.
Elections add a separate kind of uncertainty. Investors often reduce exposure to areas that could be affected by shifts in policy until results are known, and AI infrastructure, with its dependence on energy supply, financing and government decisions, is one of those areas. VanEck lists this uncertainty alongside inflation as a contributing factor rather than as the primary cause of the decline.
4. Is Power the Real Constraint on AI Computing?
VanEck says power capacity, not demand, is the physical bottleneck for AI computing. Companies that control large amounts of electricity are therefore holding the resource that appears hardest to obtain, and that scarcity is central to the firm's argument that their shares have been undervalued.
Grid connection timelines reinforce the point. Backlogs in connecting new facilities to the electrical grid mean that capacity already in place carries a scarcity premium. A miner that has already secured a power connection can offer something that a new entrant cannot deliver quickly, because waiting for utilities to approve and build connections can take a long time.
Bitcoin miners are well placed in this respect because their operations have always been built around cheap, large-scale electricity. Securing sites with substantial power supply was a core part of the mining business long before AI demand arrived, and some of those same assets now serve a second purpose.
The distinction between compute demand and power supply is useful for understanding why the firm focuses on electricity. Chips can be manufactured and shipped, and buildings can be constructed, but neither is useful without a dependable supply of power. When utilities cannot connect new facilities quickly, the sites that are already energized become the limiting factor, and whoever controls them holds a position that is hard to replicate.
5. Contracted Cash Flows Versus Share Prices
VanEck says the stocks look significantly dislocated from fundamentals given their long-term contracted cash flows. A contracted lease generates predictable payments for its term, and the value of those payments does not fall simply because market sentiment toward AI financing has worsened, provided the tenant keeps paying.
That is the significance of the observation that no signed leases were canceled. Neoclouds faced high debt service costs, yet the agreements that provide the miners with revenue remained intact. If the tenants had been walking away from contracts, a share price decline of 35% might have been easier to justify.
The firm's view is that the October rout could represent an attractive entry point for investors. It does not interpret the decline as a collapse of the narrative in which miners reposition themselves as AI infrastructure companies.
Investors evaluating a lease-based business generally focus on three questions: how long the contract lasts, how creditworthy the tenant is and how difficult it would be to replace that tenant. VanEck's note emphasizes the first of these by pointing to long-term contracted cash flows, while the tenants' high debt service costs highlight why the second remains a point to monitor.
It is also worth noting what the note does not claim. VanEck does not say that financing strain at neoclouds is imaginary, and it does not say that every AI infrastructure company is cheap. Its argument is narrower: the miners that supply power and sites were sold in sympathy with a group of companies whose circumstances differ, and that mismatch created the opportunity it describes.
6. Why Miners Moved Into AI Infrastructure
The pivot itself rests on a simple observation. Bitcoin miners and AI data centers both need large amounts of electricity, cooling and secure buildings. A miner with a site, a power agreement and operating experience can lease capacity to computing customers, earning income that does not depend on the bitcoin price.
For public companies, the shift can also change how investors value them. Mining revenue fluctuates with bitcoin prices and network difficulty, while lease income resembles that of a data center or real estate operator. That is part of why these shares have traded as AI proxies as well as bitcoin proxies, and why debt worries in the AI sector can feed back into mining stocks.
The October episode shows the downside of that dual identity. When AI financing fears rose, the miners were pulled along with the companies that rent chips, even though their balance sheets and customers differ.
Because leases are denominated in fixed payments over multiple years, the shift also changes the risk profile for lenders and shareholders. Revenue visibility improves, but the companies take on obligations to build and deliver capacity on schedule, which makes execution and access to power central to results. That is one reason the grid connection backlog VanEck highlights matters as much for these firms as the bitcoin price once did.
7. What Investors Should Watch Next
The central test for VanEck's thesis is whether tenants continue honoring their leases. The firm's argument depends on contracted payments remaining intact, so any cancellation or renegotiation of a signed agreement would undermine the case that the sell-off was unrelated to fundamentals.
Credit conditions for neoclouds deserve attention as well, since debt costs at GPU rental firms were the original source of the contagion. Further moves in AI debt yields, along with future inflation data and the post-election policy outlook, could keep influencing how investors value these companies.
Finally, grid connection timelines will continue to shape the value of existing power capacity. If connection backlogs persist, miners that already hold secured power may retain the scarcity premium VanEck describes, though the shares' recent moves show how quickly sentiment can override that logic in the short term.
For bitcoin itself, the topic remains relevant because many of these companies continue to hold or produce the asset while expanding into AI hosting. How they balance the two businesses will affect how investors classify them, and with it how closely their shares track the price of bitcoin versus the fortunes of AI infrastructure.