The Red Cell Project
The Red Cell series is published in collaboration with The National Interest. Drawing upon the legacy of the CIA’s Red Cell—established following the September 11 attacks to avoid similar analytic failures in the future—the project works to challenge assumptions, misperceptions, and groupthink with a view to encouraging alternative approaches to America’s foreign and national security policy challenges.
Editor’s Note: Lisa Klaassen is a Writer and Producer for CNN and a Doctoral Researcher at the Oxford Internet Institute. Broderick McDonald is an Academic Researcher at Oxford and Kings College London and a Visiting Fellow at the Alan Turing Institute’s Centre for Emerging Technology & Security.
By Mathew Burrows, Program Lead, Strategic Foresight Hub
In recent years, Western leaders have reframed US-China competition as a winner-takes-all race for AI supremacy: who can train the most powerful large-language model, edge closest to artificial general intelligence, and build the next trillion-parameter system. From successive U.S. presidents to Silicon Valley executives, a single refrain has been chanted like a dystopian prayer: Democracies must outpace China in frontier AI innovation — or face existential strategic defeat.
Yet this narrative overlooks the real tiebreaker. In practice, the ability to deploy AI at scale matters far more than the pursuit of ever-larger models for most sectors. First-mover advantage will not be won by the country that produces marginally superior models, but by the one that embeds AI — efficiently, safely, and ubiquitously — across factories, transportation systems, and public services. The most powerful engine in the world is of little value if it cannot be integrated into a functioning vehicle. AI is no different. Raw computational horsepower is not a national strategy, and it will not determine who ultimately prevails. But deploying AI at scale across our economies will require significant public trust in the technology. To achieve this, safety must be seen as the tracks upon which innovation progresses, rather than an impediment.
On both deployment and public trust in AI, China may be years ahead.
China’s Early Advantage
Beijing’s focus on applied AI reflects a long tradition of “grand steering”: combining industrial policy with top-down state direction to reshape entire sectors. As Dan Wang argues in “Breakneck,” China’s advantage lies in its character as an engineering state with deeply embedded “process knowledge” — a capability that determines how new technologies are deployed at scale. While the U.S. and UK debate frontier models, China’s approach extends far beyond generative AI labs and deep into its industrial base, consumer markets, and public services.
This agenda spans factory-floor applications such as robotic arms, machine vision, and automation software; AI-powered consumer products ranging from autonomous vehicles and intelligent robots to adaptive wearables; and the large-scale deployment of AI in logistics, transport, eldercare, childcare, and domestic services. China’s advantage lies not in keeping pace on AI research and development, but in its proven capacity to rapidly deploy emerging technologies at scale — built on decades of experience automating its manufacturing base and digitizing its infrastructure.
The scale of adoption is staggering. China has leapfrogged both Germany and Japan in robot density and now deploys more industrial robots than the rest of the world combined. Across the maritime sector, Beijing operates 18 fully automated port terminals, with a further 27 under construction, slashing turnaround times and tightening supply-chain efficiency. In renewable energy, AI-driven grid management has cut power outage durations from 10 hours to just three seconds.
AI is also revolutionizing China’s healthcare system. In 2024, Tsinghua University launched Agent Hospital, the world’s first AI-powered medical facility, where virtual doctors diagnose and treat thousands of patients daily with 93% accuracy. By 2025, the model had moved into public deployment in Beijing’s hospitals, with AI supporting everything from digital admissions to radiology, infusion management, and nursing care. Meanwhile, DeepSeek has expanded to 260 hospitals spanning nearly every Chinese province. Together, Agent Hospital and DeepSeek are building scalable healthcare in a nation that, like much of the world, is facing critical physician shortages and an aging population.
For Beijing, AI is not just a research frontier. It is industry and infrastructure — central to productivity, national competitiveness, and the state’s broader project of economic transformation.
The West, by comparison, lags far behind in economy-wide adoption. Only 40% of firms in the U.S. and Europe have integrated AI into their workflows. Where deployments do exist, they are often superficial, producing negligible gains in profit, or productivity — the West’s Achilles’ heel. This disparity reflects differences in labor markets: Deploying AI in white-collar industries, where Western implementations are concentrated, presents distinct challenges compared to manufacturing settings, where automation pathways are more established. One recent MIT report found that 95% of AI deployments in the U.S. generated no measurable impact on profit or loss.
It was not always this way. The twentieth century became the American century precisely because the U.S. paired scientific breakthroughs with domestic manufacturing. Learning-by-doing — across automobiles, aerospace, and eventually microelectronics — powered America’s technological leadership. Silicon Valley earned its name from the silicon chips it once produced domestically, before fabrication was offshored to Asia. By decoupling design from production, the U.S. severed the iterative feedback loop that once drove applied innovation.
China, by contrast, already holds a decisive manufacturing advantage, employing roughly 105 million manufacturing workers, compared with just 13 million in the U.S. As Dan Wang has observed, China is run by engineers, while America is run by lawyers — a contrast that helps explain China’s superior capacity to integrate design and production within a single industrial ecosystem. This integration has enabled a rapid shift from copying Western products to pioneering world-leading innovations in drones, solar technologies, batteries, and consumer electronics. After decades of sustained investment, China is no longer merely the world’s workshop; it has overtaken the U.S. and Europe across multiple strategic technologies.
Semiconductors remain China’s principal constraint. The U.S. still controls access to leading-edge chips, and despite the Trump administration easing certain export controls, restrictions on the most advanced technologies persist. Yet even this advantage is narrowing as Beijing invests billions of dollars into indigenous chip production.
China’s other advantage is raw industrial power. If Beijing wanted to build the world’s largest datacenters, it could do so faster and at lower cost than the U.S. In 2024, China generated over 10,000 terawatt-hours of electricity — more than the combined output of the U.S., European Union, and India — while adding roughly 600 terawatt-hours of new electricity demand in a single year, compared with around 130 terawatt-hours in the U.S. Sam Altman’s audacious target of 250 gigawatts of datacenter capacity by 2033 could be multiplied many times over in an economy where individual steel companies outproduce the total annual U.S. output, and where renewable energy capacity in 2024 exceeded that of the U.S. by approximately 1,400.
This represents perhaps the most critical strategic vulnerability for Western AI leadership: With unrivaled industrial capacity and centralized political will, Beijing could deploy infrastructure at a speed and scale that Western competitors, even acting collectively, would find difficult to match. The combination of China’s manufacturing dominance, energy surplus, and ability to coordinate state resources towards singular objectives creates an asymmetric advantage that could prove decisive in any race to build the physical infrastructure that AI supremacy requires.
The Trust Gap
Deploying AI across our economies will also require public trust. If citizens view a new technology as risky, extractive, or misaligned with their values, they will not adopt it willingly — and may actively oppose it.
Here, Western democracies face a serious handicap. The 2025 Edelman Trust Barometer finds that 72% of people in China say they trust AI, compared with just 32% in the U.S. and other Western democracies. Polling by the UN Development Programme shows a similar pattern: Trust in AI is highest in low-income countries and weakest in wealthy nations with “very high” Human Development Index (HDI) scores.
Some of this divergence is structural. Democracies cultivate skepticism towards concentrated power — technological as well as political — while authoritarian regimes emphasize cohesion and alignment with state-led technological goals. But the gap also reflects profoundly different lived experiences.
Decades of hype cycles have promised transformative change but delivered sluggish productivity and stagnant wages. These cycles, along with offshoring, have produced a historically grounded disillusionment with “technological progress.” Once heralded as engines of shared prosperity, many digital innovations have instead produced a proliferation of online gambling platforms, gig-economy intermediaries, and disposable mobile apps — yielding limited public benefit and, in many cases, new social harms.
As a result, many Western citizens now view AI — and digital technology more broadly — as a threat to job security, privacy, and public safety rather than as a source of empowerment.
China offers almost the inverse picture. There, technological progress is woven into a national story of revival following a “century of humiliation”— the period from the mid-19th to mid-20th century when China suffered military defeats, colonial exploitation, and internal collapse at the hands of foreign powers.
Over the past two decades, many digital technologies have tangibly improved the daily lives of hundreds of millions of Chinese citizens, even as many have restricted civil liberties, privacy, and human rights. Payments, healthcare, transportation, and public services have digitized far faster than in most Western economies. Simultaneously, China’s late but rapid industrialization — driven by export-led growth and state planning — lifted roughly 800 million people out of extreme poverty within a single generation.
Technological innovation did not single-handedly produce this transformation, but it extended the gains of industrial growth by dramatically improving efficiency and scale.
Researchers call this an “experiential prior”: a baseline expectation that new technologies will deliver empowerment, mobility, and efficiency. When elite Chinese university students were surveyed about existential risks, misaligned AI ranked last — far behind nuclear war, pandemics, and climate change. Meanwhile, more than half of Americans rate the societal risks of AI as high.
How the West Can Win the AI Deployment Race
To catch up, the West must pivot from an innovation-only mindset to a deployment-first strategy built on public trust. As Jeffrey Ding has argued in his book “Technology and the Rise of Great Powers,” the scale of technological deployment typically delivers greater returns than chasing ever-smaller performance improvements. In certain military domains, shaving milliseconds off inference time may be decisive. But for the vast majority of industrial applications — manufacturing, logistics, healthcare, finance — models that are simply close to the frontier can unlock substantial productivity gains when deployed widely and integrated deeply into existing workflows.
Yet deployment at scale cannot occur without public trust, and here lies the West’s central challenge. Workforces and consumers who doubt the safety, fairness, or purpose of AI will resist integrating it into their jobs and daily lives. Restoring trust requires more than reassuring rhetoric. It demands rigorous third-party evaluation of AI systems, extensive red-teaming to identify failure modes before they occur in the wild, transparent governance frameworks that give citizens meaningful input into how AI is deployed, and credible plans — backed by substantial investment — to mitigate labor-market disruption and address distributional impacts. These measures will require time, resources, and political courage. But they are not optional extras; they are the foundation for unlocking AI’s economic potential in democratic societies.
This should prompt a rethink of the familiar framing of trade-offs between progress and safety. In reality, reasonable safeguards and public reassurance are not obstacles to technological progress but preconditions for it. With them, economies can capture the broad-based productivity gains that AI promises; without them, deployment stalls as public resistance, workforce hesitation, and institutional caution introduce friction that prevents adoption at scale. The bottleneck is not the technology itself, but the societal infrastructure required
Western governments should therefore resist the temptation to simply emulate China’s model. Beijing’s rapid AI adoption has been enabled by a distinctive political system that tolerates far weaker protections for privacy, civil liberties, and human rights. Such a strategy is neither politically legitimate nor practically transferable to democratic societies, where circumventing public trust would ensure backlash, litigation, and ultimately, failure. The very features that appear to slow Western deployment — public debate, oversight, labor protections — are not bugs to be eliminated but the foundation upon which durable, economy-wide adoption must be built.
Durable leadership in AI will not come from cutting corners on safety standards. It will come from constructing an ecosystem that is transparent, accountable, and capable of earning broad social acceptance. This path may appear slower in the short term, but it is the only one capable of delivering the deep, sustained integration that transforms economic productivity for generations to come. The race is not to deploy fastest, but to deploy best — and in doing so, to chart a model of AI development that others will want to follow.
