Daily Vecsignal - The Power Landlords of AI

 The Power Landlords of AI

June 05, 2026 | VECS News

Bernstein analysts Gautam Chhugani and Mahika Sapra have initiated coverage on Bitcoin miners Terawulf and Cipher Digital, branding them “the power landlords of the artificial intelligence era” in a research note distributed to institutional clients on 16 June 2025. The two Nasdaq-listed companies were assigned Outperform ratings with price targets implying more than 70 percent upside from their trading levels at the time, based chiefly on their growing pipeline of contracts to host high-performance computing loads for AI enterprises. The report argues that the electrical infrastructure, cooling capacity, and land holdings accumulated by Bitcoin miners represent a scarce resource in an economy suddenly starved of data-centre gigawatts, and that Terawulf and Cipher are best positioned to monetise that scarcity. Bernstein’s call instantly rippled through both the digital-asset and equity markets, sending each stock up double digits in after-hours trading and reigniting a sector-wide debate about what a Bitcoin mining company should actually look like.

The analysts see AI hosting revenue expanding to account for roughly half of Terawulf’s total sales by the second half of 2026 and an even larger share for Cipher, which is retrofitting one of its Texas facilities into a dedicated GPU-as-a-service campus. Terawulf already operates 150 megawatts of near-zero-carbon power at its Lake Mariner site in New York, where a pilot AI cluster serving a large foundation-model developer has been running at 98 percent utilisation since March, according to the note. Cipher, which mined its first Bitcoin only in 2023, holds a 200-megawatt pipeline in West Texas that can access some of the lowest wholesale power prices in the United States, making it attractive to hyperscalers seeking to offload inference workloads without building new physical plants. Bernstein calculates that the addressable market for hosted AI compute could exceed $80 billion by 2027, and that miners who entered early — bearing cheap power contracts signed during Bitcoin’s bear cycles — possess a moat that traditional data-centre developers will struggle to replicate before the end of the decade.

The concept is straightforward but potent. Large-scale Bitcoin mining requires precisely the same raw inputs as AI training and inference: abundant electricity, high-density rack space, liquid cooling systems, and access to fibre connectivity. Because miners have already secured multi-year power purchase agreements with grid operators and constructed the substations and switchgear needed to step voltage down for servers, they can offer AI clients an operational facility in months rather than the four to six years it takes to permit and build a greenfield data centre. Terawulf’s CEO Paul Prager confirmed during an investor call that the company is in advanced negotiations with three “tier-one cloud providers” to take the entirety of its soon-to-be-expanded 300-megawatt capacity. Cipher’s management told Bernstein it expects to sign its first AI hosting contract within weeks, targeting gross margins above 80 percent on those contracts, far exceeding the 55 to 65 percent margins typical of self-mining operations.

From an investment-instrument perspective, the Bernstein initiation marks an inflection point for how capital markets classify Bitcoin miners. Until now, publicly traded miners have been proxy bets on Bitcoin’s price with a perpetual discount for operational complexity. The promise of predictable, dollar-denominated AI hosting revenue — contracted under multi-year take-or-pay agreements — fundamentally alters the risk profile of these stocks, making them investable for growth and value managers who cannot hold volatile cryptocurrency exposures. Analysts at asset manager VanEck described the shift as “the industrialisation of digital-asset infrastructure,” predicting that by year-end a distinct “AI-miner hybrid” sub-sector will emerge, potentially eligible for inclusion in infrastructure ETFs and broadening the institutional base well beyond crypto-native hedge funds. The correlation of Terawulf’s stock to Bitcoin has already fallen from 0.84 to 0.61 over the past quarter, evidence that the market is beginning to price AI earnings separately.

The crypto market itself experiences dual pressures from this evolution. On one hand, if miners allocate a portion of their energised capacity away from ASIC rigs towards GPU hosting, the growth rate of Bitcoin’s hash rate could moderate, easing competitive pressure on pure-play miners and supporting higher margins for those who remain focused on securing the network. On the other hand, the capital flowing into AI hosting could cross-subsidise the purchase of next-generation ASICs, allowing hybrid miners to maintain or even grow their Bitcoin output without tapping equity markets for dilution. Jaran Mellerud, co-founder of Hashrate Index, said the Bernstein call “validates what the smartest mining operators have been engineering for two years: the future of Bitcoin mining is not just about hashing, it is about arbitraging energy into whatever computation the world values most. Right now, that is AI.”

Not every observer is convinced the transition will be seamless. Alex Thorn, Head of Firmwide Research at Galaxy Digital, cautioned that AI hosting demands reliability levels — “five nines” uptime — that Bitcoin mines have never been contractually required to meet, and that the learning curve from ASIC cooling to GPU liquid-loop thermal management is steep. “A miner that blows a transformer can usually restart within hours and still mine a block. An AI cluster that goes dark even for minutes can trigger service-level penalties that erase a month’s profit,” Thorn wrote in a client memo. He added that hyperscalers are notoriously rigorous about physical security, redundancy, and power conditioning, requiring investments that may narrow the margin advantage miners believe they hold. Notwithstanding these hurdles, Thorn acknowledged that the trio of early movers — Terawulf, Cipher, and the already-diversified Core Scientific — are setting the standard against which all future mining valuations will be measured.

A further dimension concerns the regulatory and geopolitical layer. Much of the AI buildout is taking place under the umbrella of national-security narratives, with governments in the United States and Europe treating domestic AI infrastructure as a strategic asset. Bitcoin miners that repurpose facilities into AI data centres may find themselves suddenly subject to Committee on Foreign Investment in the United States (CFIUS) scrutiny if they host foreign-owned models, a compliance burden not present in the Bitcoin mining industry. Meanwhile, the Department of Energy has begun examining energy-intensive data centres more closely, and miners that pivot to AI could become entangled in a patchwork of state-level permitting reviews originally designed for legacy cloud operators. Clara Medalie, Research Director at Kaiko, noted that “the same power-contract footprint that gave miners a first-mover advantage could become a regulatory lightning rod if AI energy consumption becomes a political issue, which it almost certainly will by the next election cycle.”

Despite the risk, the Bernstein initiation has already stirred a capital-markets response that goes beyond stock ratings. Two large-cap asset managers with combined AUM exceeding $2 trillion have approached Terawulf and Cipher about privately negotiated convertible note offerings that would specifically fund AI infrastructure without diluting common shareholders, according to two people familiar with the discussions. This signals that credit investors, not just equity analysts, are buying into the narrative that Bitcoin-mine-turned-AI-host is an infrastructure asset backed by hard physical capital, long-duration power contracts, and counterparty-grade clients. The possibility that such miners could become acquisition targets for traditional data-centre real estate investment trusts or for private infrastructure funds is now being modelled by M&A desks at several major investment banks, further blurring the boundary between crypto-native industry and institutional-grade real assets.

For the broader digital-asset investment universe, the “power landlord” thesis introduces a new dimension of convexity. Investors can now gain exposure to the AI buildout through vehicles that also hold a levered stake in Bitcoin’s price cycle, essentially an embedded call option on cryptocurrency that traditional AI infrastructure plays lack. CoinShares Head of Research James Butterfill argued that hybrid miners possess “the most asymmetric return profile in the digital economy,” because their cash flows are underpinned by creditworthy AI tenants while their balance sheets house Bitcoin holdings that could appreciate rapidly in a bull market. Butterfill has begun work on a thematic index that would benchmark the performance of publicly listed hybrid miner-AI hosts, anticipating that passive products will follow once the universe of such companies reaches a critical mass of market capitalisation.

The Bernstein initiation arrives at a moment when AI and crypto are no longer siloed narratives but intersecting arcs of the same hardware revolution. Terawulf and Cipher, once obscure names in a crowded mining sector, are now being discussed in the same breath as traditional data-centre operators such as Equinix and Digital Realty. If the analysts’ ninefold revenue thesis proves even half-correct, the repricing of Bitcoin mining equities could be the most significant capital-markets event for the crypto industry since the approval of spot Bitcoin ETFs in early 2024. The market has begun to reward miners not for the coins they extract, but for the power they plug in — a fundamental shift that may redefine what a “crypto investment” looks like for the next decade.