Can America lag China in the AI race? A new threat is cropping up

New York's decision to impose a one-year pause on new hyperscale data centres may prove to be more than a state-level policy dispute. It is the clearest sign yet of a growing clash between America's AI ambitions and the infrastructure required to sustain them.

Across the country, communities are increasingly resisting data-centre construction over concerns about electricity consumption, water use and environmental impact. At the same time, AI companies are demanding unprecedented amounts of computing power to train models and deploy AI services at scale. If public opposition and infrastructure bottlenecks make it harder for America to build the power plants, transmission networks and data centres that AI requires, the country could find itself facing an uncomfortable reality. It may continue to lead in AI innovation while gradually losing ground to China in the race to deploy AI across the broader economy.

If deploying AI at scale increasingly depends on access to electricity, land and computing infrastructure, China could gain a decisive edge in translating AI into economic and industrial power.


America's new data centre backlash

The most visible sign of the emerging conflict comes with New York Governor Kathy Hochul imposing a one-year pause on new hyperscale data centres while the state studies their impact on electricity costs, water consumption, grid reliability and local communities.

The move reflects a broader national trend rather than an isolated state decision. A recent Gallup survey found that 71 percent of Americans oppose having a data centre built near where they live, nearly 20 percentage points higher than those who oppose building a nuclear power plant nearby. Across the country, local communities are increasingly objecting to new facilities because of concerns about noise, water use, land consumption and rising electricity demand.

The political significance of this opposition is becoming clear. Data centres have emerged as one of the few issues capable of bringing together figures as ideologically different as Senator Bernie Sanders and former Trump adviser Steve Bannon. Counties and municipalities in multiple states have already adopted temporary restrictions or moratoriums.

The debate has become important enough to attract foreign attention. A recent report by The New York Times documented how Chinese, Russian and Iranian state-linked actors have sought to amplify American concerns about data centres. According to threat intelligence firm Alethea, state media outlets from those countries mentioned data centres roughly 700 times during the first half of 2026. The objective, researchers told the newspaper, was not to invent concerns but to exploit existing divisions and turn them into a political fault line. That foreign actors see data centres as a strategic vulnerability is itself revealing. It suggests that computing infrastructure is increasingly viewed as a geopolitical asset comparable to energy, manufacturing capacity or semiconductor production.

Also Read | US-China tech rivalry explained: Why the AI and chip war matters

Why America needs more data centres

At first glance, the backlash may seem surprising because the United States already enjoys an overwhelming lead in AI infrastructure. According to Stanford University's 2026 AI Index, America hosts more than 5,000 data centres, over ten times the number in China. It remains home to the leading frontier AI developers including OpenAI, Anthropic, Google DeepMind, Meta and xAI. American firms continue to attract the majority of global AI investment.

Yet AI's rapid growth is changing the equation. Traditional cloud computing required enormous amounts of computing power. Generative AI requires far more. Training frontier models demands massive clusters of advanced chips. Serving hundreds of millions of AI users requires additional infrastructure. The next phase of AI development, involving autonomous agents, industrial automation and robotics, will require even larger computing footprints.

Technology companies are responding accordingly. Amazon, Microsoft, Google, Meta and Oracle are collectively planning hundreds of billions of dollars in infrastructure spending. Analysts increasingly speak of gigawatt-scale AI campuses, facilities whose electricity requirements can rival those of major cities. This is why industry leaders view restrictions on data centres with growing concern. The issue is not whether America has enough capacity today but whether it can build enough capacity to meet future demand.

Also Read | US panic is missing the point on Chinese AI

The electricity challenge

The deeper challenge is not data centres themselves but power. Every credible forecast points to a sharp increase in electricity demand driven by AI. Utilities that for years expected relatively flat consumption are suddenly revising projections upward. Large power companies now routinely identify data centres as the single biggest source of future demand growth. Recognising the scale of the challenge, Washington has begun treating electricity infrastructure as a strategic priority.

The Federal Energy Regulatory Commission has launched proceedings aimed at accelerating grid connections for large electricity users such as AI data centres. The Department of Energy has introduced initiatives designed to speed transmission construction and improve grid capacity. The Trump administration has encouraged technology companies to fund their own generation capacity and cover the costs of infrastructure upgrades associated with their projects.

Technology firms themselves are pursuing increasingly ambitious solutions. Microsoft and other companies have supported nuclear projects. Major developers are exploring small modular reactors. Others are investing in dedicated natural gas generation located alongside data centres. Industry executives increasingly refer to this as a "bring your own power" model. These efforts demonstrate how seriously the problem is being taken while also revealing the scale.

Why building enough power can be tough

America's challenge is not a shortage of capital or engineering talent. It is the difficulty of building large infrastructure projects quickly. Transmission lines often take a decade or longer to move from proposal to completion. Environmental reviews, permitting disputes, litigation and local opposition can delay projects for years. New power plants face similar obstacles. Even when projects receive approval, grid interconnection queues can become bottlenecks. Analysts at RAND have argued that reforms to permitting, transmission planning and generation incentives could unlock enormous additional electricity capacity by the end of the decade. The fact that such reforms are necessary shows how constrained the existing system has become.

This is where comparisons with China become uncomfortable for American policymakers.

China is not simply adding data centres. It is simultaneously building power plants, transmission networks and industrial infrastructure. It generates more than twice as much electricity as the US and continues to expand capacity at a remarkable pace. While America debates individual projects county by county, China can coordinate national infrastructure priorities through a centralised system. That does not mean every Chinese project succeeds or that the country's approach is without costs. But when the challenge is scaling energy-intensive industries, speed itself becomes a competitive advantage.

Why China could gain the upper hand

The greatest risk to the US is not that China suddenly overtakes OpenAI or Anthropic in model quality. The more plausible scenario is that AI gradually evolves into an infrastructure-intensive industry where access to power matters as much as access to algorithms. If that happens, China could benefit in several ways.

Lower infrastructure constraints could reduce the cost of operating AI systems. If Chinese firms can access electricity more cheaply and expand capacity more rapidly, they may be able to deploy AI services at lower cost. Larger computing capacity would allow more experimentation and training runs. Even if American firms maintain a technological lead, Chinese competitors could narrow the gap by running more models and serving larger user bases.

China could gain an advantage in industrial deployment. This may be the most consequential area of all. China already leads in many segments of advanced manufacturing and is rapidly expanding its position in robotics. If AI becomes deeply integrated into factories, logistics networks and industrial systems, abundant computing infrastructure could amplify strengths China already possesses.

This possibility is increasingly visible in robotics. The US still leads many areas of frontier AI research, but China is investing heavily in applying AI to physical systems. An environment where electricity, data centres and industrial scale become decisive factors could favour the country that can deploy infrastructure most rapidly.

America may still lead innovation and yet lag China

The US retains formidable advantages in AI. It remains the centre of frontier AI development. It leads in advanced semiconductor design, venture capital, research talent and entrepreneurial ecosystems. No serious analyst believes those advantages will disappear overnight. The more realistic concern is a divergence between innovation and deployment.

America may continue to build many of the world's most powerful AI models. Yet if infrastructure constraints slow the deployment of those models across the economy, the country could struggle to capture their full economic benefits. China, meanwhile, may be able to move more quickly in embedding AI throughout manufacturing, logistics, energy systems and robotics.

The AI race was once viewed primarily as a contest between software companies and research laboratories. In near future, it may look like a contest between power grids, transmission networks and construction schedules. If that shift continues, the question will not be who invents the smartest model but who can build the infrastructure needed to turn AI into national economic power. On that front, the warning signs for America are becoming difficult to dismiss.