Data center power is being taken care of by big tech companies.
Order to continue operating nuclear and coal-fired power plants to ensure stable power supply
Natural gas power generation is also expected to expand as investment becomes more active.
The US Department of Energy is urging Big Tech companies to generate their own power for their data centers.
On January 16th, the Trump administration abruptly announced a $15 billion "Emergency Power Auction" to address the surge in electricity demand resulting from the expansion of artificial intelligence (AI) data centers. This auction, targeting the PJM Interconnection region (13 jurisdictions in the Northeast, including Pennsylvania, Ohio, and Virginia), the largest power grid operator in the United States, will operate on a market-based basis, with power consumers procuring power directly from the market. Key bidders include hyperscale Big Tech companies such as Amazon, Google, Microsoft, and Meta, as well as large-scale data center developers. Notably, this auction, unlike the typical one-year contract, presumes a 15-year Power Purchase Agreement (PPA) and employs a "take-or-pay" model, requiring payment for electricity regardless of actual usage. This is interpreted as a policy intention to induce private capital investment in power generation facilities by ensuring the recovery of the enormous initial costs invested in building new power plants, and to accelerate the expansion of energy infrastructure in the mid- to long-term.
In a statement, U.S. Energy Secretary Chris Wright said, “The additional electricity demand resulting from the surge in data centers will be met by investing capital from tech companies to build new power plants. We will lead the AI era and revitalize the manufacturing boom based on a low-cost and stable power supply.” This plan aims to prevent the costs of expanding energy infrastructure from being passed on to the general public in the form of taxes or higher electricity rates by having electricity users directly bear the costs. In fact, in Virginia, the “world’s largest hub” with 561 data centers, there has been criticism that the costs of building the power grid and infrastructure are being passed on to residents, and complaints have been persistent that citizens are bearing the cost of AI innovation. The Energy Department’s latest measure is interpreted as a policy intention to prevent the spread of such social conflict while clarifying the “beneficiary pays principle” in the electricity procurement process.
Big tech is ahead of policy, securing power with long-term PPAs.
Even before the Trump administration's latest move, hyperscalers* had been proactively investing massive amounts of capital in the energy sector to address power supply uncertainties. In particular, they are actively pursuing a strategy of "direct energy transactions," signing power purchase agreements (PPAs) directly with energy suppliers to secure a stable supply of electricity exclusively for their data centers. Google signed a renewable energy power purchase agreement with NextEra Energy in 2024 and has agreed to receive power generated from the restart of the Duane Arnold Energy Center nuclear power plant in Iowa for the next 25 years, starting in October 2025. Furthermore, the company is pursuing a plan with Kairos Power, a small modular reactor (SMR) developer, to supply SMR-based power to its Tennessee data center starting in 2030.
* A global big tech company that directly builds and operates ultra-large data center infrastructure and can expand IT, power, and network resources almost infinitely as needed.
The competition among big tech companies to monopolize energy is becoming increasingly fierce. Meta has been receiving solar power from NextEra Energy for its Tennessee data center since 2024, and plans to procure power from Constellation's restarted Crane Clean Energy Center (Three Mile Island Unit 1 nuclear power plant) starting in 2027. Meta is also known to have built the largest energy portfolio among big tech companies, securing a total of 6.6 GW of power by 2034 through collaborations with Vistra, Oklo, and TerraPower. Amazon, following its 2024 contract to develop SMRs with Dominion Energy, plans to build up to 12 SMR complexes in Washington state in collaboration with Energy Northwest. Microsoft is also making all-out efforts to secure next-generation energy sources, including signing a 20-year power purchase agreement (PPA) with Constellation and agreeing to receive 50MW of electricity from Helion Energy, a nuclear fusion power startup, starting in 2028.
A particularly noteworthy aspect of Big Tech's energy supply and demand strategies is their bold, preemptive investments in next-generation energy technologies, such as SMRs, micro-modular reactors (MMRs), and nuclear fusion power generation, which have yet to see commercial operation. While these technologies still face technical and economic verification challenges before they can be commercially deployed, Big Tech is embracing a short-term, risk-taking strategy. While they aim to secure immediate power demand quickly through restarting existing power plants, they believe that securing the cheapest and most stable energy source for mid- to long-term needs will come from securing next-generation nuclear power technology. This can be interpreted as a long-term strategy by hyperscalers to secure a stable power supply, a key element of AI competitiveness, regardless of external environmental changes.
Trump declares national energy emergency, signaling shift in energy mix
Early in his term, President Trump declared a national energy emergency and announced his intention to proactively address future increases in electricity demand by immediately reviewing and eliminating various regulations that impede energy production through executive orders. Furthermore, he established the National Energy Dominance Council, spearheaded by the White House, to establish a control tower to oversee policy decisions for all available energy sources, including oil, gas, and nuclear power. He also reduced subsidies for renewable energy sources like solar and wind power and reallocated those resources to natural gas infrastructure and nuclear power (including SMRs), thereby initiating a full-scale transformation of the US energy mix.
Natural gas is a key energy source, accounting for approximately 40% of total power generation in the United States by 2025. While the rapid expansion of renewable energy facilities such as solar and wind power projects a decline in the share of power generation, the absolute volume of natural gas power generation is projected to remain stable or even increase due to the surge in electricity demand stemming from the spread of artificial intelligence (AI). Gas power generation is considered the optimal alternative as a "flexible backup power source" with immediate output adjustments, complementing the volatility of renewable energy and responding to peak demand, such as during heat waves. In response to this trend, Blackstone, the world's largest private equity fund, is aggressively pursuing natural gas infrastructure. Following its recent acquisition of Pennsylvania's Hilltop Energy Center for approximately $10 billion, Blackstone has also finalized a $12 billion investment in West Virginia to build the state's first combined-cycle gas-fired power plant, the Wolf Summit Energy Center. This is a prime example of how fossil fuel-based power generation and advanced tech capital combine to address the power shortage caused by AI data centers.
Coal-fired power generation, which was on the verge of being phased out due to environmental pollution risks, declining price competitiveness compared to gas-fired power generation, and the burden of maintaining aging facilities, is facing a new turning point with President Trump's re-election. With the surge in electricity demand driven by the spread of artificial intelligence (AI), concerns about the stability of power supply are growing, leading to delays in the planned shutdowns of coal-fired power plants across the board. In particular, the Brandon Shores coal-fired power plant in Maryland, originally scheduled to shut down in 2025, has been extended to May 2029 under a government "continued operations order" to ensure grid stability. This is interpreted as an emergency life extension measure to allow aging coal-fired plants to function as baseload power sources until transmission grid upgrades are completed, as part of the Trump administration's executive order on "energy sovereignty and coal industry revitalization."
implication
The US energy infrastructure market is entering an unprecedented phase of transformation as the Trump administration's "Energy Dominance" policy and Big Tech companies' "electricity independence" strategies converge. In particular, the $15 billion "Emergency Power Auction" and the 15-year Power Purchase Agreement (PPA) terms are expected to provide domestic nuclear power and power equipment companies with opportunities to enter the US energy supply chain and expand their business opportunities in the mid- to long-term. In an interview with KOTRA's New York Trade Center, Mr. A, who works for an investment firm, stated, "The biggest challenges facing SMR projects currently underway in the US are construction delays and enormous construction costs." He added, "Korean companies possess experience in manufacturing key components and supply chain capabilities capable of responding to rapid delivery schedules. If they can exploit these areas, they will likely be able to enter the US market." He added, "The natural gas power generation sector is also expected to benefit." "Just like the SMR market, domestic gas turbine companies boast shorter lead times and superior compliance with deadlines compared to their global competitors," he said. "In particular, the 'B2B direct transaction market,' where companies bypass utilities (power agencies) and directly contract with hyperscalers like Amazon and Meta, which urgently need power, is expected to emerge as a high-value-added new market for domestic companies." In line with the U.S. government's energy independence initiative, it is time to establish a foundation for entering the U.S. AI infrastructure market by leveraging compliance with deadlines and a flexible supply chain.