Bitcoin miners suffer 35% drop in October amid AI sector sell-off

Authors: Matthew Sigel, Griffin MacMaster (VanEck Digital Assets Research Team)

Compiled by Deep潮 TechFlow

Shenchao Overview: As the generative AI arms race enters a more advanced phase, Wall Street is experiencing a paradigm shift in its understanding of capital expenditures for computing power. In October 2024, as new GPU cloud providers faced high bond issuance costs and a correction in tech stocks, a selling wave targeting the AI supply chain rapidly spread to the cryptocurrency mining sector.

However, global leading asset management firm VanEck explicitly stated in its latest research report that the market has conflated the balance sheet pressures of hash rate leasing providers with the owners of underlying power infrastructure, leading to significant valuation mispricing. Wall Street has overlooked the most fundamental physical reality: the ultimate bottleneck limiting global AI computing capacity is no longer the GPU chips themselves, but the extremely scarce grid connectivity capacity and high-load power access points.

For publicly traded mining companies that already own high-voltage substations, grid interconnection permits, and energized assets, their long-term leases have not only remained untouched but have established an irreplaceable monopoly rent premium amid nationwide grid queue delays extending to 2030. The irrational 35% pullback in October created an exceptionally rare valuation recovery window for long-term capital.

In the just-concluded month of October, the U.S. stock market witnessed a remarkable case of asset mispricing.

Publicly traded Bitcoin mining companies that actively transitioned their high-load sites into AI data center infrastructure experienced an average single-month stock price plunge of up to 35%.

From Hut 8 and Core Scientific to TeraWulf and Cipher Mining, these former pure-play cryptocurrency mining companies have recently undergone strategic transitions to become Powered Shell Providers (PSPs).

However, during the broader downturn in U.S. tech stocks in October, Wall Street investors did not differentiate between the underlying business models of these companies; instead, they treated the entire sector as a homogeneous asset and engaged in indiscriminate selling.

VanEck’s digital assets research team, a globally recognized asset management institution, stated in its latest report that this sharp decline is completely disconnected from the core business fundamentals of the companies, and that the current valuations of these mining firms have entered an extremely oversold range.

VanEck emphasizes that the current market sell-off is essentially a classic "dislocation," not a genuine rejection by capital markets of miners' long-term business model transition to AI infrastructure.

The core trigger for this indiscriminate sell-off was panic spreading from Wall Street over the debt crisis affecting new GPU cloud rental providers (Neoclouds).

Neocloud GPU rental companies, represented by CoreWeave, have generally pursued aggressive financing through highly leveraged asset-backed debt to rapidly purchase hundreds of thousands of extremely expensive NVIDIA GPU chips.

Amid the continued high-interest-rate environment from the Federal Reserve and the stronger-than-expected rebound in the U.S. CPI for September, AI-related debt yields have surged significantly, substantially increasing the debt repayment cash flow pressure on these computing power operators.

Adding to the macro political uncertainty ahead of the U.S. midterm elections in November, pessimistic expectations of "hashrate leasing providers defaulting" began to spread across the secondary market.

Unfortunately, Wall Street has labeled the "powered shell providers" that offer space and power to these mining operators as "guilty by association."

During investors' panic selling, GPU operators, who bear the high risk of chip depreciation and debt costs, have been mistakenly conflated with infrastructure landlords who simply collect stable electricity and data center rental fees.

This risk contagion involves a significant cognitive disconnect: mining companies that provide power and data center infrastructure do not bear the hardware depreciation risk associated with chip obsolescence; their business role is purely that of a utility and infrastructure lessor.

In stark contrast to the panic sentiment prevailing in the secondary market, the actual business operations of listed mining companies have demonstrated strong resilience.

VanEck disclosed a highly compelling key empirical finding in its research report: despite increasing debt service costs facing Neoclouds, none of the long-term infrastructure lease agreements formally signed by major mining companies have been canceled or terminated.

Taking Core Scientific as an example, its massive 12-year, up to 500-megawatt hosting lease agreement with CoreWeave is still progressing as scheduled.

The expansion project of Lake Mariner campus, owned by TeraWulf, is also being delivered systematically, continuously providing stable clean power and data center space to external AI customers.

Meanwhile, Hut 8 is collaborating with renowned investment firm Coatue to advance its digital infrastructure platform, while Cipher Mining and IREN are on track to secure large-scale power capacity.

Tech giants and hyperscalers are not slowing down their capital expenditures on generative AI model training and inference; instead, they are continuously increasing them.

Even if individual hash rate lessees encounter extreme liquidity crises in the future, the underlying data centers and power capacity are almost entirely free from idle risk.

In today’s market, where demand for electricity is extremely high, top tech giants like Microsoft, Amazon, Google, or Oracle are ready to fully acquire any high-quality assets that are already powered.

VanEck's research report sharply identifies the core truth behind today's computing power competition: the electricity capacity firmly controlled by these mining companies is the ultimate binding constraint on global AI computing power.

In the early stages of the generative AI boom, global tech giants competed fiercely for NVIDIA’s high-performance GPU chip production capacity.

However, as chip supply chain bottlenecks gradually ease, a more harsh physical barrier stands before everyone: the entire U.S. power grid has no spare immediate electricity to allocate.

The grid interconnection queue times for major regional transmission organizations across the U.S.—such as PJM, ERCOT, and MISO—have been extended to 4 to 7 years, with many new projects delayed until after 2030.

Moreover, the lead time for core power equipment such as large high-voltage transformers has stretched to three to four years from order to delivery, with severe supply chain shortages.

Industry estimates show that by 2028, the U.S. data center electricity demand gap could reach tens of gigawatts (GW).

Under precisely this extremely severe shortage of power grid infrastructure, the powered assets built by Bitcoin mining companies over the past several years have been granted an unprecedented "scarcity premium."

These companies already possess compliant and legally approved grid connection permits, dedicated independent substations, and real-time available live capacity reaching hundreds of megawatts.

Tech giants cannot buy time in the short term with their abundant U.S. dollar capital, while mining companies hold ready-to-use powered sites—the only shortcut to bypass the long grid connection waitlist.

Based on the non-fungible power asset barrier and stable lease cash flows, VanEck believes that current mining company stock valuations have experienced a severe structural disconnect.

In the traditional U.S. stock market data center sector, established real estate investment trusts (REITs), such as Equinix and Digital Realty, have consistently maintained enterprise value-to-EBITDA multiples in the high range of 20 to 25 times due to their stable rental income.

More critically, the vast majority of established campuses operated by these traditional data center giants are already operating at full capacity and generally lack the massive, newly available power sites needed to support the training of next-generation large models.

In contrast, publicly traded Bitcoin mining companies holding gigawatt-scale undeveloped power reserves are trading at forward EV/EBITDA multiples of only 8x to 12x, following a 35% plunge in October—a valuation level just half that of traditional data centers.

This significant valuation discount clearly fails to account for the long-term, fixed-income cash flows from contracts already locked in by mining companies, spanning 10 to 20 years.

As mining companies gradually deliver and begin contributing high-certainty recurring EBITDA from their powered shells, capital markets will inevitably reassess the value of these enterprises.

VanEck concludes that the sharp downturn in October, triggered by market panic, was not a collapse of the AI transformation narrative, but rather an attractive, historic buying opportunity for visionary long-term investors.