A major new AI computing center in Wuxi, China, using 1,536 domestically produced Huawei processors, shows Beijing’s accelerating push for technological self-sufficiency while fueling U.S. concerns over the security of global technology supply chains. The project, announced May 15, is part of a national strategy to create a "computing power grid," treating AI processing capability as a state-controlled utility and deepening the technological rift with Washington.
The move comes as American researchers warn of the long-term dangers of integrating Chinese technology into critical infrastructure. "It is untenable for our much-needed wave of expansion and modernization to depend so deeply on a strategic competitor for the digitally-active components that increasingly govern how power is generated, stored, and directed across the grid," wrote researchers from the Carnegie Mellon Institute for Strategy and Technology (CMIST) in a recent paper on the topic.
The Wuxi facility, dubbed a "Token Factory," is a partnership between A-share listed Hongxin Electronics and the Wuxi High-tech zone. According to the company, the first phase will deploy four Huawei Ascend 384 super-node servers, creating a 1,536-GPU cluster. The project is a key node in China’s ambitious "East-West Computing" initiative, designed to balance computing resources across the country. Underscoring the commercial goal, China Telecom launched trial "Token Plans" on May 17, allowing customers to buy AI processing power much like a mobile data plan.
The project crystallizes a central dilemma facing Western nations: how to secure critical infrastructure without stalling its development. China’s effort to build a vertically integrated AI ecosystem, from chips to cloud services, is seen in Washington as a systemic long-term threat. This has prompted calls for more disciplined security policies that distinguish between commodity hardware and high-risk digital control systems.
A National 'Computing Power Grid' Takes Shape
The "Token Factory" is a tangible step toward what Chinese state media describes as a "computing power grid," an analogy to the national electricity grid. The strategy, endorsed by China's State Council, aims to have state-controlled entities build and manage massive computing centers, with users purchasing "compute" on demand. This approach could lower costs and accelerate AI adoption for Chinese firms by turning immense capital expenditure into a simple operating expense.
This vision of state-managed compute capacity is backed by significant national investment. In May, China's National Development and Reform Commission noted that total investment in the "six networks"—including the computing power grid, new electric grids, and communication networks—is expected to exceed 7 trillion yuan this year. This state-led industrial policy aims to create compounding advantages across multiple sectors, from energy and defense to transportation and autonomous systems, by controlling the underlying "electrotech stack."
'Electrotech Moneyball': US Warns of Systemic Risk
As China builds, U.S. experts are mapping out the corresponding risks. The CMIST paper, titled "Electrotech Moneyball," argues that the greatest danger lies not in commodity hardware but in "digitally active control layers" like firmware, orchestration software, and cloud-connected management platforms. These systems, often remotely updatable, could become vectors for surveillance or disruption. The researchers point to China's National Intelligence Law and the discovery of China-linked actors like Volt Typhoon inside U.S. critical infrastructure networks as evidence of the threat.
However, the paper cautions against "self-inflicted paralysis" from imposing blanket restrictions on all Chinese-made components, which would slow the very build-out the U.S. needs. Instead, it proposes a "Moneyball" framework to strategically prioritize the highest-risk technologies for domestic control while allowing managed global procurement for lower-risk hardware. "Treating the entire electricity ecosystem as if everything is an emergency means that nothing will be defended effectively," the authors noted.
From Academic Warnings to Public Accusations
The strategic concerns outlined by academics are being echoed in more confrontational terms by public figures. Investor Kevin O'Leary recently alleged that a spike in online misinformation targeting U.S. grid and AI projects is linked to Chinese actors. "Who would want us to stop building our electrical grid? Who would wanna stop us from having compute capacity to develop AI? Which adversary would want that? There’s only one. It’s China," O'Leary said, claiming the activity was the work of bots and proxies for the Chinese Communist Party.
This escalating rhetoric shows the U.S.-China tech rivalry is moving beyond policy debates into the public domain, framing it as a direct conflict. While China frames its infrastructure projects as part of its national development, as seen in planning for the Beijing Energy Congress 2027 which lists AI and data centers as key demand drivers, some in the West see a coordinated strategy to simultaneously build its own capabilities while undermining its rivals. The Wuxi project, therefore, is not just a domestic industrial development but a significant event in a deepening global technological competition.
This article is for informational purposes only and does not constitute investment advice.
