BlackRock is telling its customers to stop looking at artificial intelligence as software and start treating it as energy.
In its 2026 World Outlook, BlackRock Investment Institute argues that the rise in AI is pushing the boundaries of physics and highlighted electricity as a constraint that is underpricing investors.
The report’s headline warns that AI-driven data centers could consume 24% of U.S. electricity by 2030, a scale that could reposition everything from public capital investment to industrial land.
This type of prediction raises obvious follow-on questions in cryptocurrencies. If grid access becomes a scarce asset, what will happen to an industry that has built its business model around turning cheap, interruptible electricity into Bitcoin?
In 2025, stories about the potential synergy between cryptocurrencies and AI arrive with the theory that AI agents will want to use cryptocurrencies for payments rather than traditional finance. However, a power war could worsen this relationship in the future.
For many years, the mining industry has lived in a political debate over energy waste. Industry counter-arguments are always valid. Miners become flexible loads, able to turn off power when the grid is stressed and absorb excess generation when prices collapse.
In Texas, the Electric Reliability Council of Texas (ERCOT) has specifically designed its program for “large, flexible customers such as Bitcoin mining facilities” to encourage curtailment during peak demand.
However, AI data centers come with different consumption profiles, different contract terms, and different levels of political support. They never want to power down. They want baseload.
Power issues hidden in the technology boom
BlackRock’s broader argument is that the AI boom is unusually capital-intensive. The company says its total capital spending plan for building AI by 2030 is in the range of $5 trillion to $8 trillion, with significant spending on computing, data centers, and energy infrastructure.
What started as a race for chips quickly became a race for megawatts.
Even as analysts debate the limits, it is widely agreed that data center power demands are increasing rapidly. Data center load growth in the U.S. has tripled over the past decade, according to a Department of Energy announcement related to Lawrence Berkeley National Laboratory’s Data Center Report.
Furthermore, it is predicted to double or triple by 2028. According to 2024 EPRI modeling cited by Utility Dive, U.S. data centers are expected to account for 4.6% to 9.1% of U.S. electricity generation by 2030, depending on AI penetration and efficiency gains.
A World Resources Institute explainer, citing a Berkeley Lab study, points out that by 2030, U.S. electricity consumption will increase from 6.7% to 12% (wri.org)
BlackRock’s “up to 25%” framing falls on the offensive end of the spectrum and is intended to be provocative. But even a low-end scenario would be enough to tighten electricity markets and intensify grid politics over who gets power first.
Reuters reports that power companies and grid operators are already adjusting rates and rules as hyperscalers and colocation companies compete for capacity, especially in hotspots like Texas and Northern Virginia.
That is the environment that Bitcoin miners are stepping into. They are large mobile power users and prioritize regions with abundant power generation and attractive pricing. Until now, those characteristics seemed like an advantage.
Miners are built on flexibility. AI operates on certainty
Bitcoin mining is very simple at the physical layer. A dedicated computer performs the hashing and secures the network, but the main input cost is electricity. When electricity is cheap compared to the price of Bitcoin and network difficulties, miners print cash. When electricity becomes expensive, businesses close, relocate, or go bankrupt.
As public scrutiny increases, operational flexibility has become a top topic in the industry. The U.S. Energy Information Administration estimates that crypto mining will likely account for about 0.6% to 2.3% of U.S. electricity consumption in 2024, a small share in percentage terms but large enough to factor into local politics and grid planning.
Texas is the cleanest case study because its competitive electricity market translates that flexibility into revenue. In a 2023 SEC filing, Riot Platforms said it reduced power usage by more than 95% during peak demand periods in August 2023, choosing to forgo mining revenue to support ERCOT reliability.
crypto slate ERCOT reported the same month that it paid out $31.7 million in energy credits to miners for power outages during the heat wave, a detail that speaks both to the value of flexibility and why politics can get ugly quickly.
Then place that model next to the AI. Training and servicing large models requires constant power and short uptime. Hyperscalers with long-term leases want predictable deliveries rather than voluntary reductions.
If miners are shock absorbers, AI is shock generators.
And BlackRock’s full-year outlook effectively says a shock is coming and there’s no stopping it.
Grid constraints make cheap power a moving target
In a mining scenario, “cheap power” could mean stranded hydropower, surplus wind at night, or friendly industrial rates. But as data centers scale, access to the power grid itself becomes a bottleneck, making cheap power a moving target.
Interconnection queues and transmission delays are new issues. Even if there is local generation, there may not be the wires, transformers, or permitting routes to power a new 500-megawatt campus.
NERC warns of reliability threats from rapid load growth associated with AI, data centers, EVs, and electrification, colliding with generator retirement and slow additions. (Financial Times)
This is important for miners because their advantage is speed.
You can drop a container on site, energize it, and start hashing faster than traditional industrial plants. But when the gate items become substation capacity and interconnection approvals, that speed turns into a regulatory battle.
Political perspectives are also changing
When power markets tighten, lawmakers start looking for villains. Mining is often useful because it feels like an option even for people who don’t understand anything about it. In contrast, AI is now serving as a national competitive advantage for both citizens and legislators.
That asymmetry will shape policy. It is easier to impose reporting requirements and additional duties on miners than on data centers solicited by local chambers of commerce. It’s also easy to frame mining as a speculative luxury and AI as the backbone of defense, productivity, and medicine.
If BlackRock is right that AI’s energy footprint poses a macro risk, political coalitions supporting grid investments could grow, but pressure to prioritize “productive” loads could grow as well.
Miners may respond more to the flexibility story. A Duke University report cited by Utility Dive argues that the existing U.S. power grid can handle significant new loads if curtailed during stress events, and mining can do that. Many AI workloads, especially consumer product inference, typically cannot do this.
This creates a potential wedge between miners as controllable loads to support renewable energy integration and data centers as inflexible loads. This debate is already gaining momentum in policy circles and at public hearings at the Electric Power Commission.
However, winning depends on local economic conditions and lobbying, not internet debates.
Hedge: Turn your mining site into an AI site
Another adaptation path is already underway: converting from hashing to hosting.
The logic is simple. If you already own land, power rights, and substations, you have what AI developers need most. Additionally, if your legacy business is unstable, the contract cash flow prospects from compute hosting are attractive.
crypto slate reported in October that some companies initially focused on Bitcoin mining are pivoting to AI infrastructure, with deals related to cloud and AI workloads, just as power access has become at a premium in places like Texas. The message of this article is not that all miners will become AI masters, but that the industry’s primary assets are moving from machines to megawatts.
This pivot is harder than you might think. AI data centers require different cooling, different network architectures, and different uptime guarantees. Mining can tolerate interruptions, but many AI customers do not.
Renovation costs can be significant, and competition includes specialized data center operators with deep relationships and financial advantages.
However, the direction of progress is clear. When power is scarce, the highest value uses of megawatts tend to be prioritized.
Where Bitcoin mining takes place
BlackRock’s predictions are not specific to Bitcoin, but about the end of cheap abundance. As AI propels the U.S. into a world of rapidly increasing power demand and slow transmission, businesses built on marginal power economics will be under pressure.
Of course, miners will not disappear. Bitcoin’s incentive structure is designed to keep hashing power somewhere online, and the industry’s maneuverability means it can chase new pockets of energy. However, the center of gravity may shift.
Regions with surplus generation-friendly policies are likely to view miners as a stabilizing industrial load, especially if miners can reliably propose cuts. Regions that favor hyperscalers will undoubtedly treat miners as a second priority.
Perhaps the result is a barbell.
On the one hand, miners integrate with the grid, sign structured demand response agreements, and become part of public works plans.
On the other side are miners who transform their energy positions into broader computing infrastructure and arbitrage early entry into the power market into essentially new business areas.
Either way, the easy days are coming to an end. BlackRock’s warning that AI data centers could grow to a huge share of U.S. electricity demand is a reminder that the next phase of digital infrastructure will not be constrained by code, but by the messy physical world of wires, permits, turbines, and heat.