China’s power surge tests U.S. AI comeback

By late 2025, the US-China AI chip supply dynamic became a focal point of intense debate and strategic maneuvering. Nvidia’s implementation of location verification technology on its Blackwell-generation AI chips marked a significant step to curb chip smuggling and enhance security, yet the White House’s controversial decision to permit sales of Nvidia’s H200 chips to China sparked fears of accelerating China’s AI capabilities. Despite Chinese demand outstripping supply and Nvidia weighing the tradeoff of diverting fab capacity from cutting-edge Blackwell chips, critics like former UK PM Rishi Sunak warned this could hasten China’s overtaking of Western AI leadership. Meanwhile, skepticism mounted over the White House’s justifications, with experts like James Sanders suggesting the decision was based on flawed assumptions about China’s domestic chip production, possibly crafted to rationalize the sale rather than reflect reality.

Throughout 2025, US export controls on AI chips oscillated amid geopolitical tensions, reflecting the complex balancing act in the AI race. Biden’s AI Diffusion Rule initially imposed strict restrictions, only for Trump to rescind it months later and broker a deal allowing H200 sales to China with a 25% revenue cut to the US government. In response, China intensified efforts to reduce Nvidia chip dependence, banning its tech giants from purchasing these chips by September and unveiling competitive offerings from Huawei. Yet, Chinese AI companies faced severe compute constraints, with leaders like Alibaba’s Justin Lin estimating less than a 20% chance of surpassing US AI frontrunners within five years, underscoring how export controls, while slowing China’s near-term progress, simultaneously fuel its push for semiconductor self-reliance.

Energy infrastructure emerged as a critical and often underappreciated battleground shaping the AI competition’s trajectory. By early 2026, US data center electricity demand was projected to more than double to 426 TWh by 2030, outpacing China’s expected 277 TWh, yet the US faced a bottleneck due to regulatory delays, local opposition, and a stagnated grid since the 1980s. Conversely, China’s rapid addition of over 500 gigawatts of mostly renewable power capacity in 2025, coupled with a coordinated Western Data, Eastern Compute strategy, allowed it to swiftly align energy supply with AI demand. This structural advantage, amplified by China’s dominance in renewable supply chains and centralized planning, contrasts starkly with the US’s fragmented approach, exemplified by legislative proposals like Senator Tom Cotton’s DATA Act aiming to circumvent regulatory hurdles but risking grid isolation.

The evolving US-China AI rivalry reveals a nuanced interplay between hardware bottlenecks, strategic pragmatism, and differing political economies rather than a simple race. While US export controls constrain China’s access to cutting-edge chips, completely severing sales risks accelerating China’s independent chip development, effectively building a 'golden bridge' for its semiconductor ambitions. China’s AI development remains largely pragmatic and techno-deterministic, focused on rapid adoption to avoid falling behind, often at the expense of safety and alignment concerns emphasized in the US. Industry observers predict that the next pivotal moment in this competition will hinge less on AI models themselves and more on hardware supply dynamics, particularly memory chip availability and pricing, underscoring how infrastructure and industrial policy—centralized in China but fragmented in the US—will shape long-term AI leadership.

By early 2026, the United States is grappling with a critical energy bottleneck that threatens to throttle its AI data center expansion despite leading in advanced chip technology. US data centers demand is projected to more than double to 426 TWh by 2030, yet the grid—stagnant since the 1980s—faces regulatory delays, local opposition, and slow infrastructure growth, prompting legislative responses like Senator Tom Cotton's DATA Act to enable isolated power plants. Meanwhile, China’s aggressive power infrastructure buildout, adding over 500 gigawatts in 2025 alone, largely from renewables, has created a vast and cheaper electricity supply that underpins its rapid AI and industrial scaling, giving it a decisive strategic advantage.

China’s dominant position in electricity generation—reaching nearly 3.9 terawatts by the end of 2025, roughly triple the US capacity—combined with centralized long-term infrastructure planning and energy subsidies, fuels its manufacturing and AI ambitions. This coordinated approach, exemplified by strategies like 'Western Data, Eastern Compute,' aligns regional energy resources with AI workloads, optimizing power-intensive training in resource-rich interior provinces and inference near coastal centers. In contrast, the US faces rising electricity prices and prolonged grid connection delays, which threaten to increase operational costs and constrain AI infrastructure growth just as demand surges exponentially.

While China’s energy abundance propels its AI and industrial capacity, its chip production remains a critical bottleneck due to export controls and technological hurdles in extreme ultraviolet lithography (EUV). Many Chinese data centers sit idle, hampered by limited access to cutting-edge Nvidia H200 chips and domestic manufacturing challenges, with industry leaders like Alibaba’s Justin Lin estimating less than a 20% chance of surpassing US compute power in the next 3-5 years. The recent easing of export bans offers some relief, but supply constraints and strict licensing keep chip shortages a persistent strategic constraint for China.

The US faces a staggered bottleneck across its AI supply chain: although chip production may recover relatively quickly, critical shortages in electricity supply, memory components like RAM, and power transformers threaten to stall AI infrastructure growth. As Ben Horowitz highlights, servers sometimes arrive without RAM due to manufacturing shortfalls, and the electricity grid cannot keep pace with the vertical surge in AI compute demand. This unprecedented infrastructure challenge demands massive investment and innovation to rebuild America’s industrial base and energy systems, or risk ceding long-term AI and manufacturing leadership to China.

By early 2026, it became clear that the primary bottleneck in revitalizing American manufacturing was not machinery but the scarcity of tacit workforce knowledge—expertise accumulated over years and difficult to scale. Companies like CloudNC demonstrated that AI tailored to think like seasoned machinists could capture and amplify this tribal knowledge, enabling factories to absorb complexity and boost productivity with their existing skilled workers, a critical factor for reshoring and defense manufacturing efforts.

Despite widespread adoption of robotics and AI, the skilled labor shortage remains acute, with 69% of manufacturers struggling to fill technical roles as retiring engineers take decades of institutional knowledge with them. Industry leaders like Mike Hughes of Peak International Group emphasize that closing this expertise gap demands smarter knowledge capture, targeted AI deployment, and prioritizing use cases such as remote diagnostics to sustain operational performance amid this workforce crisis.

The aging American manufacturing workforce holds invaluable tacit knowledge that is rapidly being lost, driving companies to seek AI and robotics solutions to encode and preserve this expertise. Squint’s AI, backed by $40 million in Series B funding, exemplifies this approach by integrating tribal knowledge with work orders and regulations, outperforming major AI competitors in complex industrial problem-solving and underscoring the strategic imperative to safeguard and scale workforce expertise for national competitiveness.

Addressing the projected shortfall of 1.9 million manufacturing workers over the next decade requires overcoming entrenched perceptions of manufacturing as dirty and dangerous, despite its transformation through robotics, AI, and sustainability. Industry leaders call for coordinated action among educators, policymakers, and businesses to realign training with modern manufacturing needs and invest in talent pipelines, recognizing that true innovation only materializes when a skilled workforce can scale production and solve real-world challenges.

By mid-2026, experts emphasized that durable, long-term industrial policy frameworks are essential to galvanize private capital and sustain confidence in U.S. manufacturing and energy sectors. Proposals such as establishing a federal highway trust fund for the electricity grid aim to create master plans for linear infrastructure, connecting collocated energy and manufacturing buildout zones to improve resilience and reduce costs. This coordinated federal-state approach seeks to streamline project approvals and foster integrated supply chains, addressing the patchwork challenges that have historically hindered industrial growth.

The Biden administration's industrial policy has demonstrated tangible success by catalyzing irreversible investments exceeding hundreds of billions of dollars in clean energy supply chains, which have generated jobs, raised wages, and advanced climate objectives. Analysts argue that rather than discarding this framework, incremental improvements should be pursued to preserve its momentum, especially given the critical role of industrial policy in securing domestic supply chains amid the energy system's shift toward highly manufactured components. Political stability, notably preventing administrations hostile to such policies, remains a pivotal factor in sustaining these gains.

The 2026 Reindustrialize Summit in Detroit underscored a bipartisan consensus on accelerating American industrial dominance through collaboration among over 1,500 leaders spanning government, business, and investment sectors. Despite widespread adoption of robotics and AI by more than half of manufacturers, 69% report difficulties filling skilled technical roles, spotlighting workforce development as a critical bottleneck. This platform fosters durable frameworks by aligning policymakers, technologists, and industrial leaders to rebuild manufacturing capacity with cutting-edge technologies, emphasizing that workforce readiness is as vital as technological advancement.

Federal low-cost financing mechanisms have become a cornerstone of U.S. industrial strategy, with the Department of Defense’s Office of Strategic Capital authorized to commit up to $100 billion in loans and the Department of Energy’s Office of Energy Dominance Financing holding over $339 billion in loan authority. These programs aim to lower capital expenditure hurdles and mitigate risks for pioneering industrial projects, effectively retraining the sector’s 'muscle memory' to attract private investment. However, sustaining these initiatives requires urgent congressional reauthorization of entities like the Export-Import Bank and the Office of Energy Dominance Financing to provide long-term certainty for decade-spanning projects, while policy discussions increasingly favor debt financing over production tax credits for greater durability.

Critics caution that tariffs alone are insufficient for restoring U.S. industrial leadership; without reinvestment in the industrial base and workforce, protectionist measures risk becoming symbolic walls around an empty lot. The Boeing-China aircraft deal illustrates China's strategic push to develop its own aerospace industry, underscoring the need for the U.S. to build resilient domestic capabilities rather than rely solely on trade barriers. A comprehensive industrial strategy must therefore integrate durable financing, infrastructure, and workforce development to rebuild manufacturing and maintain global competitiveness.

Addressing the looming manufacturing workforce shortage—projected at 3.8 million jobs over the next decade with half potentially unfilled—requires a coordinated effort among educators, policymakers, and business leaders to align training with industry needs and invest in talent pipelines. Modern manufacturing offers high-paying, technologically advanced careers that defy outdated stereotypes, with average salaries exceeding $106,000 and workplaces powered by robotics, AI, and sustainable practices. Companies like Saint-Gobain North America exemplify this approach, having invested nearly $7 billion supported by an 18,000-strong workforce that innovates through problem-solving rather than mere strategic planning, highlighting that investing in people is fundamental to rebuilding American industrial capacity.

America’s industrial golden age was defined by a powerful synergy between innovation and manufacturing capacity, a cycle that fueled economic strength and global leadership for over two centuries. This era was anchored by strategic federal policies, such as tariffs protecting infant industries in the 19th century, and robust partnerships between universities and government that catalyzed breakthroughs from the Internet to lifesaving medical treatments. However, since the 1950s peak when the U.S. commanded 45% of global manufacturing output—a figure now diminished to 11%—this vital connection has frayed, undermining the nation’s ability to translate discovery into production and economic power.

The erosion of America’s manufacturing base, accelerated from the 1970s onward by a shift in focus from fabrication to finance, has left the country vulnerable as global competitors like China aggressively build their own industrial capacities, notably in aerospace and semiconductor manufacturing. This retreat, described by Atomic Industries CEO as a six-decade reliance on globalization and service economies, has hollowed out essential trades, factories, and engineering talent, creating a strategic imperative to rebuild physical production capabilities alongside innovation ecosystems to sustain national power in the decades ahead.

Renewing American industrial leadership demands a modernized, strategic partnership between academia, government, and industry that shortens the distance between discovery and deployment. As Farnam Jahanian emphasizes, this requires rethinking funding and incentives to seamlessly connect basic science with real-world production, while expanding educational access to cultivate the next generation of scientists and engineers. Without sustained federal investment across the full innovation continuum, the U.S. risks ceding technological and economic dominance to rivals who are more adept at integrating innovation with manufacturing infrastructure.

Underlying the decline of America’s industrial might is a loss of the long-term optimism and commitment to worker welfare that once defined its manufacturing ethos. Visionaries like Bob Noyce warned of a 'death spiral' leading to a 'Build-Nothing Country' where short-sighted profit motives erode the middle class and community fabric. Reviving this spirit of redemption and expansive optimism—rooted in decades-long horizons and inclusive growth—is essential to reforge an industrial base that not only drives innovation but also uplifts workers and rebuilds communities, setting the stage for a new American century of purposeful industrial renewal.

By mid-2026, technology had unequivocally transitioned from a mere tool of diplomacy or IT concern to the very battleground of national power, where control over supply chains for critical components like semiconductors and pharmaceuticals became foundational to economic security and geopolitical influence. This evolution redefined deterrence beyond traditional military might, emphasizing instead the velocity of innovation and rapid response to emerging technological threats, including those posed by adversaries leveraging inexpensive software-based tools. As one analysis put it, “Deterrence is no longer just the size of a military... it’s the pace of innovation in reacting to cheap software-built tech that adversaries may be using whether they’re a country, criminal or a terrorist group.”

The private sector emerged as the primary engine behind cutting-edge technologies such as AI, autonomous systems, and cyber defense, underscoring the necessity of robust public-private partnerships to harness these innovations for national economic growth and strategic advantage. Governments recognized that these technologies were not government-built but depended on cultivating symbiotic relationships with companies driving innovation, ensuring that technological advancements serve as a force for good on the global stage. This integration became critical as software vulnerabilities in essential infrastructure—power grids, water systems—transformed from IT glitches into potent geopolitical leverage points, amplifying the stakes of technological control.

Beyond hardware and software, the embedded values within AI systems—shaped by their cultural, ethical, and historical contexts—became a strategic front in the contest for global influence. As AI models inherently project perspectives on history, culture, and ethics, promoting American and Western values through these technologies was seen as vital to shaping international norms and countering the diffusion of competing value systems. This recognition positioned technology not only as an economic and security asset but also as a vessel for ideological leadership in the 21st century.