Red Star Rising: How America Could Lose the AI Race to China
by TWR. Editorial Team | Wednesday, September 9, 2026 for The Weekend Read.
The AI race is no longer about who has the smartest chatbot. It is becoming a contest over who controls the economic, industrial, military and technological infrastructure for the remainder of the 21st century. China understands that. America is only beginning to act like it does.
Artificial intelligence is still treated by much of the American public as a software story. Most encounter it through ChatGPT, Gemini, Claude, coding assistants, image generators and increasingly capable agents. China sees something much larger.
Beijing treats AI as a general-purpose technology capable of reshaping manufacturing, science, logistics, transportation, government, military operations and the physical economy itself.
That difference matters because the gap at the technological frontier is already smaller than many Americans assume.
U.S. private AI investment will reach $725 billion+ in 2026, more than 9 times China's $80 billion, and America still leads in many of the world's most important frontier companies. But capital is only one measure of power. China installs more industrial robots than the rest of the world combined, possesses the world's largest manufacturing ecosystem, is rapidly expanding electricity consumption and has spent years building AI into an explicit national industrial strategy.
The real contest is therefore not simply who invents the smartest model. It is who converts intelligence into factories, energy, productivity, military capability and global dependence first.
The AI race will not be won by the country with the smartest chatbot. It will be won by the country that builds intelligence into the greatest share of its economy.
China remembers what happens when you miss a revolution
Beijing's urgency is rooted in history. Chinese leadership has long viewed major technological revolutions as moments that determine whether nations rise or fall. The experience of Qing China, which fell behind Western industrial powers and subsequently suffered military defeats, foreign intervention and the period remembered domestically as the "Century of Humiliation," remains deeply embedded in China's national story.
The modern lesson Beijing draws is straightforward: never again become strategically dependent on another country for foundational technology. AI now sits directly inside that project. China's "AI Plus" strategy seeks to push artificial intelligence across industry, science, transportation, consumer services and government, with national targets for widespread adoption of intelligent terminals and AI agents during this decade.
While America debates whether Tilly Norwood has acting chops, China has been designing for an economy in which avoiding AI increasingly becomes the exception.
The factory floor may matter more than the leaderboard
The starkest comparison is not ChatGPT versus DeepSeek. It is what is happening inside factories. Of the 542,000 industrial robots installed worldwide in 2024, approximately 295,000 went into China. That is 54 percent of the entire global market. China now operates roughly 2 million industrial robots, about 4.5 times Japan's fleet, the second largest in the world. Chinese manufacturers also captured 57 percent of their own domestic robot market, compared with roughly 28 percent a decade earlier.
The comparison with America is revealing. The United States is actually more automated relative to manufacturing employment, with 307 industrial robots per 10,000 manufacturing workers versus 166 in China. But China's annual installations are estimated at roughly ten times the U.S. level because its industrial base is so much larger.
That distinction is critical. America may have greater automation density while China has far greater deployment volume. Factories generate real-world data. That data improves industrial AI. Better AI improves robots. Better robots reduce costs and enable more deployment. If that feedback loop compounds, China becomes an enormous training environment for physical intelligence.
America could invent the future and still lose the deployment war.
The old American trap: invent here, scale there
The United States has seen versions of this story before. America pioneers a technology, captures the highest-value intellectual property and then watches significant portions of the manufacturing ecosystem grow somewhere else.
AI makes that pattern much more dangerous because intelligence itself becomes part of the production machinery. American companies could continue building better foundation models while Chinese companies combine slightly weaker models with cheaper robots, batteries, manufacturing scale and increasingly capable domestic chips.
China doesn't need the entire hardware stack to be best in class. They simply need their system to become good enough and cheap enough to deploy widely.
Semiconductors remain one of America's strongest advantages. The United States and its allies retain commanding positions across chip design, advanced fabrication, lithography, high-bandwidth memory and semiconductor equipment. China remains behind at the leading edge. But Beijing has responded by investing in Huawei Ascend processors, domestic fabrication, packaging, interconnects, compute clusters and model efficiency.
The strategic question is therefore not whether every Chinese processor can beat Nvidia's best chip. It is whether China can assemble enough usable compute to continue improving competitive AI systems.
Export controls buy America time; they do not determine what America does with it.
$725 billion means little without enough electricity
America's capital advantage remains extraordinary. U.S. private AI investment will reach $725 billion in 2026, while China attracted just $80 billion in measured private investment. The U.S. also produced 2,839 newly funded AI companies, more than ten times the next closest country. Stanford cautions, however, that conventional venture statistics understate China's effort because government guidance funds operate outside ordinary private-capital measurements.
But even $725 billion cannot train a model without electricity.
The International Energy Agency expects U.S. electricity consumption to increase nearly 2 percent annually through 2030, with data centers responsible for roughly half of the total increase. China is operating on another scale entirely: it is expected to account for nearly 50 percent of the increase in global electricity demand through 2030 and add roughly 2,600 terawatt-hours of demand over five years, about the equivalent of today's entire European Union.
AI competition is therefore becoming a power-generation contest as much as a model contest.
Every transformer matters. Every transmission line matters. Every gigawatt matters.
America has a building problem, and Beijing noticed
The United States possesses enormous data-center infrastructure, but proposed facilities increasingly collide with opposition over electricity prices, water consumption, land use, tax incentives and environmental effects. Many of those concerns are legitimate. The danger comes when a country declares AI infrastructure a national-security priority while becoming politically incapable of constructing it.
That vulnerability is no longer theoretical.
In June 2026, OpenAI disclosed that it had disrupted accounts likely originating in China that supported covert influence activity around U.S. AI policy. One cluster generated social-media content claiming AI data centers were increasing electricity prices for ordinary Americans. The operation attempted to manipulate a real American political debate rather than invent one from scratch.
That does not mean people opposing data centers are Chinese agents. It means genuine grievances can become geopolitical attack surfaces.
The answer is not to silence communities. It is to remove the conditions that make opposition easy to exploit. Require developers to fund grid upgrades. Protect residential ratepayers. Demand transparent water plans. Build generation and transmission alongside compute. America should be capable of moving quickly without asking communities to absorb every cost.
The AI race eventually reaches the battlefield
The stakes extend beyond GDP. Modern warfare runs on information from satellites, drones, sensors, cyber networks and communications systems. AI can analyze those streams, identify patterns, coordinate autonomous systems and shorten the time between detecting something and acting on it.
China has spent years developing what its strategists call "intelligentized warfare," combining AI, networked sensors, autonomous systems and data-driven command structures. The United States retains significant military advantages, but AI changes the value of speed. A force that can interpret a battlefield and coordinate decisions faster may gain an advantage even when the underlying weapons are similar.
That makes adoption-speed a military variable.
The same dynamic applies economically. A country that deploys AI across engineering, science, logistics and manufacturing faster can reduce development cycles, increase productivity and compound those gains year after year. Eventually productivity becomes GDP. GDP becomes tax revenue. Tax revenue becomes research spending, industrial capacity and military power.
That is the serious meaning of "adopt AI or die." It is not literal extinction. It is the possibility of relative decline.
The contest is already going global
America and China are not merely building technologies for themselves. They are competing to determine whose infrastructure the rest of the world uses.
Washington increasingly promotes full-stack AI exports incorporating chips, cloud infrastructure, models, cybersecurity and applications. China offers its own ecosystem through manufacturing capacity, infrastructure, financing and increasingly capable open models.
Once a country builds businesses and public systems around one supplier's chips, cloud, models, APIs and industrial software, switching becomes expensive. Technology turns into dependency.
That means an AI model does not need to dominate Americans to matter strategically. A sufficiently capable Chinese model that is cheap, customizable and easy to deploy can become important across Southeast Asia, Africa, Latin America and the Middle East without ever beating the best American system on a benchmark.
What is already on your phone?
The geopolitical competition also reaches much closer to home than most consumers realize. People routinely grant applications access to contacts, photographs, precise location, microphones, files, payment information and browsing behavior without knowing who owns the software, where the company is incorporated or where the data may ultimately travel.
The issue is not "Chinese app equals dangerous app." That is neither useful nor defensible.
The issue is knowing what you are agreeing to regardless of where a company is based.
A user should be able to answer a basic question before granting sensitive permissions: Who exactly am I giving this data to?
Make the terms readable before accepting them
Most users will never manually read a 10,000-word privacy policy. AI can at least make that information easier to interrogate.
Paste the Terms of Service and Privacy Policy above and click enter or into your trusted AI system and ask:
“Explain what data this company collects, where it may be stored, whether it can be shared with affiliates or governments, whether uploaded content may train AI models, what intellectual-property rights I grant, what survives account deletion, and which provisions create unusual privacy, jurisdictional or security risk.”
Then ask the model to separate ordinary industry-standard language from genuinely unusual provisions.
In the AI era, clicking ‘accept’ can be a data-governance decision.
AI analysis can be wrong and is not legal advice. But using a tool to interrogate the agreement is better than treating "Accept" as a meaningless button.
The advantage is still America's to lose
None of this means China has already won. America still has deeper private capital markets, many of the strongest frontier AI companies, leading research institutions, critical semiconductor capabilities, enormous compute infrastructure and an alliance-network Beijing cannot easily replicate. China faces advanced-chip constraints, demographic pressure, capital inefficiencies and political structures that can inhibit entrepreneurship.
But technological advantages are not permanent assets. They are leads that must continually be converted into deployment.
China's not looking to dominate every category. It needs models that are close enough, chips that are capable enough, robots that are inexpensive enough, factories that are automated enough and electricity abundant enough for each generation of the system to improve the next--supported by distilling American models all along the way.
America therefore faces a much larger challenge than producing the next breakthrough model. We must translate our lead into electricity, manufacturing, robotics, science, defense, workforce productivity and international infrastructure faster than China can.
China remembers what happened when it missed the last great industrial transformation. Its leadership has spent decades ensuring it does not make that mistake twice.
America faces a different danger: inventing the future, confusing invention with victory, and discovering too late that somebody else became better at deploying it.
TWR. Last Word: The question is no longer who invents the next AI model, it's who builds the world runnning on it.
Insightful perspectives and deep dives into the technologies, ideas, and strategies shaping our world. This piece reflects the collective expertise and editorial voice of The Weekend Read — 🗣️Read or Get Rewritten | www.TheWeekendRead.com
Nomenclature
Frontier Model: A highly capable general-purpose AI system operating near the current performance frontier across reasoning, coding, multimodal tasks, research, or agentic work. Frontier leadership matters, but it is only one layer of national AI competitiveness.
Model Distillation: A technique in which a smaller or less expensive model learns from the outputs or behavior of a more capable model. Distillation can accelerate capability transfer and reduce the time advantage created by expensive frontier research.
Compute: The processing capacity used to train and run AI systems. In practice, compute depends on advanced chips, servers, networking, memory, data-center capacity, software efficiency, and the electricity required to operate all of it.
AI Infrastructure: The physical and digital systems required to develop and deploy artificial intelligence at scale, including data centers, semiconductor supply chains, power generation, transmission, cooling, cloud infrastructure, networking, and specialized hardware.
Embodied AI: Artificial intelligence designed to perceive, reason, and act in the physical world through robots, vehicles, industrial equipment, drones, or other machines. Embodied AI is especially important to manufacturing, logistics, defense, and autonomous systems.
Industrial Robot Density: The number of operational industrial robots per 10,000 manufacturing workers. The metric helps compare automation intensity across countries, but it does not measure total deployment scale.
Deployment Scale: The degree to which AI is actually integrated across factories, businesses, government systems, scientific institutions, infrastructure, military operations, and consumer products. A country can lead in model quality while lagging in deployment.
Semiconductor Chokepoint: A strategically important part of the chip supply chain where access is concentrated among a limited number of companies or countries. Advanced fabrication, lithography, high-bandwidth memory, chip design software, and manufacturing equipment can all function as chokepoints.
Export Controls: Government restrictions on the transfer of sensitive technology, hardware, software, or technical knowledge to designated countries or entities. In the AI race, export controls are intended to slow access to advanced compute and semiconductor capability.
AI Plus: China's national policy framework for integrating artificial intelligence throughout industry, science, consumer services, transportation, healthcare, government, and other parts of the economy. The strategy emphasizes economy-wide adoption rather than AI development as a stand-alone technology sector.
Intelligentized Warfare: A term used in Chinese military thinking to describe warfare increasingly shaped by artificial intelligence, big data, autonomous systems, networked sensors, and faster machine-assisted decision-making.
Full-Stack AI: An integrated AI ecosystem spanning chips, servers, cloud infrastructure, models, cybersecurity, software, applications, data, and supporting standards. Both the United States and China increasingly compete to export full-stack systems rather than individual technologies.
Technology Stack: The collection of interconnected technologies that support a digital system, such as chips, operating infrastructure, cloud services, models, APIs, applications, and security tools. Countries that adopt one technology stack deeply can become costly to separate from it later.
Data Jurisdiction: The legal jurisdiction governing how data may be stored, processed, accessed, transferred, or disclosed. A company's headquarters, ownership structure, server locations, and subsidiaries can all affect which laws apply to user information.
Influence Operation: A coordinated effort to shape public opinion, political debate, or behavior through information, social media, covert accounts, propaganda, or amplification of existing grievances. Such operations do not necessarily create controversies; they can intensify ones that already exist.
General-Purpose Technology: A foundational technology capable of improving productivity and enabling innovation across many industries rather than serving only one narrow use. Electricity, computing, and the internet are historical examples; AI is increasingly viewed through the same lens.
Sources
International Energy Agency. (2026). Electricity 2026. IEA. https://www.iea.org/reports/electricity-2026
International Federation of Robotics. (2025). World Robotics 2025. https://ifr.org/worldrobotics/report-2025
International Federation of Robotics. (2025, September 25). Global robot demand in factories doubles over 10 years. https://ifr.org/ifr-press-releases/news/global-robot-demand-in-factories-doubles-over-10-years
International Federation of Robotics. (2026, May 5). China makes AI-powered robots core of national strategy. https://ifr.org/ifr-press-releases/news/world-robotics-report-2024
OpenAI. (2026, June 10). PRC-linked influence operations are targeting AI debates in the US. https://openai.com/index/prc-linked-influence-operations-ai-debates/
Stanford Institute for Human-Centered Artificial Intelligence. (2026). The 2026 AI Index report. Stanford University. https://hai.stanford.edu/ai-index/2026-ai-index-report
The White House. (2025, July 23). Fact sheet: President Donald J. Trump promotes the export of American AI technologies. https://www.whitehouse.gov/fact-sheets/2025/07/fact-sheet-president-donald-j-trump-promotes-the-export-of-american-ai-technologies/
The White House. (2025, July 23). Promoting the export of the American AI technology stack (Executive Order 14320). https://www.whitehouse.gov/presidential-actions/2025/07/promoting-the-export-of-the-american-ai-technology-stack/
Reporting note
This article is a reported analysis based on public government documents, industry data, research institutions, company disclosures, and major news reporting. Figures are presented using the most recent comparable data available and may differ by methodology or reporting period. References to the United States or China “winning” or “losing” the AI race describe relative technological, industrial, economic, and strategic position, not a predetermined outcome.
Claims regarding foreign influence operations refer only to documented campaigns identified by platforms or government authorities and should not be read to imply that Americans who oppose data centers, AI infrastructure, or specific technology policies are acting on behalf of a foreign government. Consumer-security recommendations are informational and are not legal, cybersecurity, or privacy advice.
Comments