Artificial intelligence is advancing rapidly — and behind the scenes, a critical infrastructure challenge is emerging: energy supply. In 2025, data centers powering generative AI and machine learning require exponentially more electricity. Traditional power grids, especially in the US and Europe, are struggling to keep up.
According to the International Energy Agency, by 2030, global data centers will consume 945 terawatt-hours (TWh) of electricity annually — more than twice the 415 TWh consumed in 2024. This volume exceeds 20% of the electricity currently produced in the United States each year.
To bridge this gap, technology companies and energy strategists are turning to battery-powered microgrids — decentralized energy systems that can operate independently from national grids. These systems depend heavily on batteries, transformers, and integrated storage solutions — much of which are sourced from China.
CATL and Sungrow: Leaders in Export-Driven Profit
Chinese energy companies are experiencing a historic surge in demand. Two firms stand out:
-
CATL, the world’s largest battery maker, saw its shares rise 45% in 2025.
-
Sungrow, the second-largest integrated energy storage provider after Tesla, saw a 130% increase in share value.
Together, they anchor the CSI New Energy Index, which rose 38% in 2025 — a reflection of global investor confidence in China’s role in the energy-AI nexus.
Although Chinese firms do not disclose U.S. sales figures, official U.S. import data shows that China accounts for the majority of imported batteries and energy storage systems.
And profits are largely coming from exports. As domestic competition in China compresses margins, foreign markets offer far greater profitability:
-
For energy storage systems, profit margins are 3–5 times higher on exports than domestic sales.
-
For transformers, gross margins are 40–50% in the U.S. and Europe, compared to just 10–20% in China.
Despite tariffs — including a 30.9% duty on Chinese batteries — companies continue exporting. As one market strategist notes, “They would rather eat the tariff than lose the contract.”
Microgrids Are the New Infrastructure
U.S. data center developers are moving fast. Traditional grid upgrades can take years. But microgrids, supported by imported Chinese battery banks and transformers, can be installed in a fraction of the time.
In the first nine months of 2025:
-
60% of U.S. lithium-ion battery imports came from China — up from 43% in 2020.
-
The value of these imports reached $15 billion, triple the total value in 2020.
The U.S. Department of Energy acknowledges that microgrids will make up a “significant majority” of distributed energy resources in the coming decade. For companies racing to deploy generative AI models, energy speed is now strategic.
Supply Chain Dependency: A Strategic Vulnerability
The Council on Foreign Relations flagged this dependency in a 2025 report, warning that the AI race is being shaped by energy-related supply chains. Many U.S. firms use:
-
Chinese-made optical transceivers (for converting data signals).
-
Printed circuit boards sourced from Chinese manufacturers.
Despite ongoing political rhetoric about decoupling, trade volumes suggest otherwise. As Raymond Yeung, ANZ’s Greater China economist, notes: “China and the U.S. have basically not decoupled — they are one economy in two jurisdictions.”
Chinese firms also hold technological advantages:
-
CATL leads in lithium iron phosphate (LFP) batteries, which are safer and longer-lasting.
-
Chinese manufacturers offer faster delivery times — often measured in months, compared to 2–3 years from Korean suppliers.
What’s Changing in 2026?
However, the situation is evolving:
-
The U.S. plans to increase tariffs to 48.4% on Chinese batteries in 2026.
-
New rules may disqualify energy projects using high Chinese content from receiving U.S. federal tax credits.
As a result, HSBC reported a wave of “frontloaded installations” in 2025 — companies racing to complete projects before new restrictions come into force.
Whether this will shift the supply chain is uncertain. What is clear: China remains the core supplier for the energy systems that power artificial intelligence. Without it, the AI boom could slow — not from a lack of computing power, but from a lack of electricity.