While headlines focus on chip wars and AI model breakthroughs, a less visible but equally critical story is unfolding in the global race for artificial intelligence dominance? Chinese manufacturers of batteries, transformers, and energy storage systems are experiencing unprecedented demand as data centers worldwide scramble to secure power for their AI infrastructure? This surge comes despite ongoing geopolitical tensions and trade barriers, revealing a complex interdependence that defies simple narratives of decoupling?
The Power Behind the Processing
Shares in Chinese equipment makers have skyrocketed this year, with CATL, the world’s largest battery manufacturer, seeing its stock rise 45% and Sungrow, a leading energy storage system supplier, jumping 130%? These companies form the backbone of the Shenzhen Stock Exchange’s CSI New Energy index, which has climbed 38% in 2025 alone? “Suddenly there’s a scramble for this power equipment,” observes Brian Ho, an equity research analyst at Bernstein covering the global energy storage sector?
The numbers tell a compelling story? In the first nine months of this year, 60% of U?S? lithium ion battery imports came from China, up from 43% in 2020, according to U?S? Census Bureau data? The total value of these imports reached $15 billion through September�more than triple the value for all of 2020? This growth occurs against a backdrop of U?S? efforts to reduce dependence on Chinese supply chains, highlighting what the Council on Foreign Relations calls “the biggest threats to the U?S? in its race with China to develop AI?”
Why China Holds the Cards
Chinese manufacturers enjoy several structural advantages that make them difficult to replace? Companies like CATL lead in producing lithium iron phosphate batteries, which offer superior safety and longevity compared to alternatives? “Despite all these tariffs and the decoupling, the demand for [lithium iron phosphate] batteries is strong,” notes Ho, adding that there are “just no other suppliers outside China?”
The speed advantage is equally significant? For transformers�critical components that ensure data center equipment receives proper power�Chinese suppliers can deliver in months while competitors require years? “If you have to urgently build out your grid for a data center, you cannot wait for two years,” explains Matty Zhao, co-head of China equity strategy at BofA Global Research?
Profit margins tell another part of the story? Zhao estimates that for energy storage systems, export margins are three to five times greater than domestic sales? For transformers, domestic sales yield 10-20% gross margins compared to 40-50% for exports to the U?S? and Europe? “They would rather continue to export and eat up the tariff,” she says?
The Parallel Chip Story
This power equipment dependence mirrors another critical AI supply chain dynamic playing out in Washington? Nvidia CEO Jensen Huang recently secured White House approval to export H200 AI chips to China with a 25% U?S? revenue cut, following an intensive lobbying campaign that included at least six private meetings with President Donald Trump? Huang argued that restricting Nvidia would accelerate Chinese domestic chip development and erode U?S? dominance, framing the economic case directly to policymakers?
However, this decision faces bipartisan scrutiny? Republican John Moolenaar, chair of the House China committee, questions the basis for allowing these exports, arguing that Huawei’s chips aren’t true competitors to Nvidia’s and warning that exports could undermine U?S? AI leadership? A bipartisan group of six senators has introduced legislation to bar H200 export licenses for 30 months, reflecting ongoing tensions between economic interests and national security concerns?
The Regulatory Landscape Complicates
Meanwhile, the domestic regulatory environment adds another layer of complexity? President Trump recently issued an executive order attempting to block state AI laws after Congress declined to include such provisions in federal legislation? The order directs federal agencies to challenge state AI laws, claiming they force AI makers to embed “ideological bias” in models and threaten innovation?
This move has sparked debate about its impact on startups? Legal experts warn that creating a Department of Justice task force to challenge state laws could prolong uncertainty and create legal limbo for young companies? “Big Tech and the big AI startups have the funds to hire lawyers to help them figure out what to do,” notes Andrew Gamino-Cheong, CTO and co-founder of AI governance company Trustible? “The uncertainty does hurt startups the most, especially those that can’t get billions of funding almost at will?”
Investor Concerns and Economic Realities
As companies race to build AI infrastructure, investors are growing increasingly cautious? Trading volumes in credit default swaps tied to U?S? tech groups have surged 90% since early September, with particular focus on companies like Oracle and CoreWeave that are raising billions in debt for data centers? Major tech firms including Meta, Amazon, Alphabet, and Oracle have raised $88 billion this autumn for AI projects, with JPMorgan predicting investment-grade companies could raise $1?5 trillion by 2030?
This investor unease reflects broader questions about the sustainability of the AI boom? The International Energy Agency forecasts that by 2030, data centers will consume 945 terawatt hours of electricity�more than a fifth of all electricity currently generated annually in the U?S? This massive power demand is driving the rush for Chinese equipment, even as political rhetoric emphasizes independence?
A Complex Interdependence
The reality, as ANZ chief economist Raymond Yeung puts it, is that “China and the U?S? have basically not decoupled? They’re a single economy of two different jurisdictions?” This interdependence extends beyond power equipment to include optical transceivers from companies like Zhongji Innolight and printed circuit boards produced in China�all critical components for U?S? data centers?
Looking ahead, the Trump administration plans to raise tariffs on Chinese batteries from 30?9% to 48?4% next year and make it harder for equipment with high Chinese content to receive federal tax credits? However, as HSBC notes, there has been “frontloaded installation in the U?S? ahead of the implementation of the foreign entity of concern requirements,” suggesting companies are racing to secure supplies before new restrictions take effect?
The global AI infrastructure build-out reveals a fundamental truth: technological advancement depends on complex, interconnected supply chains that transcend political boundaries? As companies and nations navigate this landscape, the tension between economic pragmatism and strategic independence will continue to shape the future of artificial intelligence development?