BlackRock CEO Larry Fink Warns AI Boom Could Require More Than 70 Gigawatts of Power

BlackRock CEO Larry Fink has highlighted the enormous energy requirements behind the rapid expansion of artificial intelligence, warning that the AI boom will demand a massive increase in electricity generation and data center infrastructure.

His comments underscore a growing challenge facing the global technology industry. As companies race to develop increasingly powerful AI systems, demand for computing capacity is rising at a pace that could place significant pressure on electricity grids worldwide.

The issue is becoming increasingly important for investors because the AI boom is no longer simply a story about semiconductors, software, and cloud computing. It has evolved into an infrastructure story involving electricity generation, data centers, transmission networks, cooling systems, and long-term capital investment.

Fink's remarks add to a growing discussion about whether the world's existing energy infrastructure can keep pace with the extraordinary amount of computing power required by next-generation AI systems. The International Energy Agency estimated in 2024 that global data center electricity consumption was approximately 460 terawatt-hours and could exceed 1,000 terawatt-hours by 2026, driven in significant part by AI workloads.

Source: XPost

AI Is Becoming an Energy-Intensive Industry

Artificial intelligence requires enormous amounts of computing power. Each time an advanced AI model is trained, thousands—or even hundreds of thousands—of specialized processors can operate simultaneously inside large data centers. These processors require electricity not only to perform calculations but also to power the cooling systems, networking equipment, storage infrastructure, and other components necessary to keep data centers running continuously.

As AI models grow larger and more capable, the infrastructure required to support them is expanding rapidly. This has forged a new relationship between the technology industry and the energy sector.

For decades, investors primarily viewed data centers as part of the technology infrastructure supporting the internet. The rapid development of generative AI has fundamentally changed that perception. Data centers are increasingly being treated as major industrial facilities with electricity requirements comparable to those of large manufacturing operations.

The 70-Gigawatt Figure Highlights the Scale

The reference to more than 70 gigawatts of power illustrates the magnitude of the infrastructure challenge. A gigawatt represents one billion watts of power; seventy gigawatts therefore represents an enormous amount of electricity demand—roughly comparable to the total installed electricity generating capacity of a country such as the United Kingdom.

The precise amount of power required by AI will vary depending on how quickly companies build new data centers, the efficiency of future chips, improvements in cooling technology, and how much computing occurs during model training versus everyday AI inference. Even so, the broader trend is clear: AI infrastructure is becoming a major source of future electricity demand.

Research into the future of AI infrastructure has similarly pointed to gigawatt-scale computing clusters and substantial growth in electricity requirements over the coming years. One analysis estimated that global AI data center power demand could increase by more than 130 gigawatts by 2030 under certain scenarios.

Data Centers Are Becoming the New Infrastructure Frontier

BlackRock has already positioned data center infrastructure as a major investment opportunity. In a partnership involving BlackRock, Microsoft, and Abu Dhabi-based MGX, announced in 2024 under the name Global AI Infrastructure Investment Partnership (GAIIP), the companies sought to mobilize an initial $30 billion in private capital for AI infrastructure, including data centers and connectivity.

BlackRock described data centers as foundational infrastructure for the digital economy and argued that investment in the sector could support economic growth and technological innovation.

The investment thesis is straightforward: AI companies need computing capacity; computing capacity requires data centers; data centers require electricity. As a result, the expansion of AI could create opportunities far beyond traditional technology companies. Utilities, power producers, grid operators, construction companies, equipment manufacturers, and infrastructure investors could all benefit from the buildout.

Electricity Could Become a Critical AI Bottleneck

The biggest challenge may not be the availability of AI chips—it could be electricity. Companies can order processors and construct buildings, but those facilities cannot operate without reliable power.

This creates a potential bottleneck for the AI industry. A data center can be completed physically but remain unable to operate at full capacity if the surrounding electrical grid lacks sufficient generation or transmission capacity. That is why energy availability is becoming an increasingly important consideration when technology companies select locations for new data centers.

Areas with abundant electricity, reliable transmission infrastructure, and favorable energy prices could become major centers for AI development.

AI Infrastructure Is Driving a New Investment Cycle

The rapid growth of artificial intelligence is creating what could become one of the largest infrastructure investment cycles in decades. Companies are investing billions of dollars in data centers, high-performance computing systems, and networking infrastructure. At the same time, governments and utilities are considering how to increase electricity generation.

The investment requirements extend across multiple industries. Power plants need to be built or upgraded. Transmission lines need to be expanded. Data centers need to be constructed. Semiconductor production needs to scale. Cooling technology must become more efficient. All of these investments require capital—which is one reason major asset managers such as BlackRock are increasingly focused on AI infrastructure.

Why BlackRock Is Interested

For an asset manager of BlackRock's scale, AI infrastructure represents more than a technology trend. It represents a potentially long-term infrastructure investment opportunity. Data centers can generate recurring revenue through long-term contracts with technology companies. Power infrastructure can also provide long-duration investment opportunities.

This fits the broader investment strategy pursued by large institutional investors seeking assets capable of generating returns over many years. The AI boom could therefore create a bridge between Silicon Valley and traditional infrastructure investing: technology companies provide the demand, infrastructure investors provide the capital, and energy companies provide the electricity.

The Power Grid Faces a New Challenge

Electricity grids were not originally designed around clusters of massive AI data centers. Traditional electricity demand was distributed across households, offices, factories, and commercial facilities. AI data centers, by contrast, create extremely concentrated demand. A single large facility can require hundreds of megawatts, while future AI campuses could potentially require even more.

This creates challenges for utilities. They must determine where new generation should be built, how electricity will reach data centers, and how the grid can remain stable as demand increases. Transmission infrastructure can take years to plan, permit, and construct—creating a potential mismatch between the speed of AI development and the speed of energy infrastructure development.

Nuclear Power Returns to the Conversation

The growth of AI has revived interest in nuclear power. Nuclear plants can provide large quantities of electricity around the clock without depending on weather conditions. That reliability is attractive to data center operators that need continuous power.

Several technology companies have already moved beyond exploration into concrete agreements. In 2024, Microsoft announced a power purchase agreement with Constellation Energy to restart the Three Mile Island Unit 1 nuclear plant, while Amazon disclosed investments in small modular reactor projects with companies including X-energy. Google separately announced an agreement with Kairos Power to purchase electricity from small modular reactors. Small modular reactors have also attracted broader attention as a potential future source of power for large industrial facilities. However, nuclear projects generally face long development timelines, substantial capital requirements, and regulatory hurdles.

For the immediate AI expansion, companies may therefore need to rely on a combination of existing nuclear plants, natural gas, renewable generation, battery storage, and grid upgrades.

Natural Gas Could Benefit From AI Demand

The AI power boom could also strengthen demand for natural gas. Natural gas power plants can provide dispatchable electricity and can often be developed faster than large nuclear projects, making them attractive for regions seeking to support rapidly growing electricity demand.

However, increased reliance on natural gas creates another debate around emissions and climate policy. The AI industry could therefore face a difficult balancing act. Companies want reliable, affordable electricity, but investors and policymakers are also increasingly focused on carbon emissions. The result could be a more diversified energy mix involving natural gas, nuclear power, renewable energy, and emerging technologies.

Renewable Energy Has a Role Too

Renewable energy will remain an important part of the AI infrastructure discussion. Solar and wind generation can provide large amounts of relatively low-cost electricity. The challenge is that AI data centers typically require reliable power around the clock. Solar generation varies during the day, while wind production can fluctuate depending on weather conditions.

That means renewable power may need to be combined with energy storage, grid connectivity, or other firm sources of electricity. Some technology companies are already signing long-term renewable energy agreements to support their growing electricity consumption.

Efficiency Could Change the Equation

The amount of electricity required by AI is not fixed. Technological improvements could significantly reduce the energy required for individual AI tasks. More efficient processors, better software, improved cooling systems, and advances in model architecture could all reduce power consumption.

At the same time, efficiency improvements could encourage greater AI usage—a phenomenon sometimes described as a rebound effect. If AI becomes cheaper and more efficient, businesses may use it more frequently, potentially offsetting some of the energy savings. Improvements in efficiency may therefore reduce the amount of electricity required per AI task without necessarily reducing total electricity demand.

The AI Economy Is Becoming Physical

One of the most important implications of the energy discussion is that artificial intelligence is no longer purely a digital phenomenon. AI may be delivered through software, but the infrastructure supporting it is physical. It requires buildings, electricity, cooling systems, fiber networks, and semiconductor factories.

This means the future of AI will depend partly on industries that traditionally had little connection to software development. Energy companies could become increasingly important to technology investors. Construction companies could benefit from data center expansion. Equipment manufacturers could see increased demand for electrical systems and cooling technology. Infrastructure funds could finance new facilities. The AI economy is therefore expanding into the physical economy.

Investors Are Watching the AI Power Trade

The growing electricity requirements of AI are creating a new investment theme. Investors are increasingly looking beyond companies developing AI models and chips, examining companies that provide the infrastructure needed to operate those systems as well.

Utilities could benefit from rising electricity consumption. Power generators could benefit from long-term contracts. Data center operators could gain from strong demand for computing capacity. Infrastructure investors could participate in financing large-scale projects. The result is a broader AI investment ecosystem.

A Potential Constraint on AI Growth

If electricity infrastructure fails to expand quickly enough, energy availability could become a constraint on AI development. Companies may have the capital to purchase chips and build data centers but still face delays because they cannot secure enough power.

This could influence where future AI facilities are constructed. Regions with excess electricity generation may become increasingly attractive. Countries with faster permitting processes could also gain an advantage. In the United States, the ability to expand power generation and transmission could become an important factor in maintaining technological leadership.

Competition for Electricity Could Increase

AI data centers will not be the only source of growing electricity demand. Electric vehicles, manufacturing, industrial electrification, and other technologies are also increasing electricity consumption. This could create competition for available power.

Utilities may need to balance the needs of data centers against residential and industrial customers. That could create political debates over electricity prices and infrastructure investment. The question of who pays for grid expansion could become particularly important.

AI Could Reshape the Energy Industry

The relationship between AI and energy could ultimately become mutually reinforcing. AI companies need electricity, and energy companies can use AI to improve operations. Artificial intelligence can help utilities forecast electricity demand, manage renewable generation, optimize power plants, and identify maintenance problems.

AI could therefore simultaneously increase electricity demand while helping the energy industry operate more efficiently. The technology could become both a major consumer of power and an important tool for improving the power system.

The Bigger Picture

Larry Fink's warning about the enormous power requirements associated with artificial intelligence highlights one of the most important challenges facing the technology industry. The AI revolution will require much more than better models and faster chips. It will require a massive physical infrastructure capable of supporting increasingly powerful computing systems.

Data centers will need electricity. Electricity systems will need new generation and transmission capacity. Investors will need to provide capital. Governments will need to accelerate permitting and infrastructure development. Technology companies will need to find ways to make AI systems increasingly energy efficient.

The scale of the challenge is enormous, but so is the potential economic opportunity. BlackRock and other major investors are already treating AI infrastructure as a major long-term investment theme. The firm's previous infrastructure initiatives have emphasized the growing importance of data centers and the enormous capital requirements associated with the AI buildout.

For the technology industry, the next phase of the AI race may be determined not only by who builds the most advanced models, but also by who can secure the electricity required to run them. As AI continues expanding into businesses, consumer applications, and government services, energy could become one of the most important strategic resources in the global technology race.

The companies and countries capable of building reliable, affordable, and scalable power infrastructure may ultimately have a significant advantage in determining how quickly artificial intelligence can develop.

Source: Hokanews