Breaking AI Inertia: How the US Must Outmaneuver China’s Centralized Strategy

Rejecting incrementalism while accelerating AI innovation to secure the future.

China’s AI Strategy (2017–2025): A System Designed for Strategic Dominance

China’s formal AI strategy began with the New Generation AI Development Plan (AIDP), issued by the State Council in July 2017. Prior to this, China had been investing in AI through its broader industrial and technology modernization efforts, including the "Made in China 2025" initiative and earlier AI research programs led by the Chinese Academy of Sciences. These efforts laid the foundation for the 2017 AIDP, providing a pathway for China to rapidly scale its AI ambitions with a clear national roadmap. The plan outlined a three-phase roadmap for China to achieve global AI leadership by 2030, focusing on economic, military, and social applications of AI.

• The first phase (2017–2020) focused on achieving parity with global AI leaders by investing in core technologies such as machine learning, computer vision, and natural language processing.

• The second phase (2021–2025) emphasizes AI-enabled industrial transformation and the creation of world-class innovation hubs.

• The final phase (2026–2030) envisions China as the undisputed leader in AI technologies, ethics, and governance.

China is right on track. As the AIDP states in its second phase,

"by 2025, China will achieve major breakthroughs in basic theories for AI, such that some technologies and applications achieve a world-leading level..."

Exactly on January 20, 2025—the day of the US Presidential Inauguration and on the eve of the Chinese New Year—China’s DeepSeek foundation lab released its R1 model and variants, which outperformed OpenAI’s leading O1 model at just 1% of the cost to run.

How: the Role of MOST & NDRC

China’s approach is deeply centralized. Following the State Council’s 2017 AIDP directive, the National Development and Reform Commission (NDRC), which oversees funding and governance, and the Ministry of Science and Technology (MOST) spearheaded implementation. MOST is at the helm of turning AIDP’s vision into reality by closely collaborating with local interests (also party controlled), while the NDRC provides long-term tracking, oversight, and management of funds (to party controlled entities).

MOST has designated leading Chinese technology companies to spearhead national AI innovation platforms, each focusing on specific applications (e.g., health, cities, and manufacturing). For example, Baidu was appointed to lead in autonomous driving, Alibaba in smart city development, and Tencent in medical diagnostics. These companies focus on building nation-state platforms to foster collaboration between industry and academia, inclusive of societal advancement to accelerate the development and deployment of AI technologies across various sectors.

Unlike the US, which funds hundreds of competitors to create technology ecosystems, China creates a rival of equals, ensuring that select firms compete at a high level within a structured and state-supported ecosystem on nation-state priorities. This approach fosters rapid iteration and refinement of AI applications within a political governance structure, allowing breakthroughs to be integrated into national priorities without the inefficiencies of excessive market fragmentation. By contrast, the US system, while fostering competition, often lacks long-term strategic cohesion to translate innovation into unified national advantages. Unlike China’s model of selecting and backing companies through direct government involvement to ensure alignment with state priorities, the US relies on market-driven forces, which can lead to fragmentation and inefficiencies in scaling AI capabilities by creating a rivalry of capabilities.

MOST has also initiated aggressive programs to cultivate AI talent, directly integrating AI courses into university curricula, establishing research institutes, and promoting interdisciplinary studies. The government-driven approach is markedly different from the US University Affiliated Research Centers (UARCs)—China directly shapes coursework and research direction to ensure alignment with national priorities where as the US merely funds affiliated centers (some of which are not even within geographic line of site of the main campus).

MOST facilitates the exchange of knowledge and best practices, through domestic and international forums, as well, positioning China as a key player in the global AI community. We can ascertain that China has figured out how to make conferences, effective. MOST also facilitated research funding, partnerships with universities, and the establishment of AI standards to ensure the cohesive development of the technology. Overall, the ministry has implemented policies to create a favorable environment for AI innovation, including tax incentives, grants, and the establishment of AI-focused industrial parks.

Hangzhou: A Model of State-Driven Innovation

China’s AI strategy integrates local and national government priorities. The NDRC allocated funding to establish AI innovation hubs, such as Hangzhou’s China Artificial Intelligence Town, which became a focal point for integrating AI research, entrepreneurship, and infrastructure. Hangzhou’s main academic institution, Zhejiang University, has become a leading AI organization given both Hangzhou’s designation as an AI town and MOSTs influence at the University. The university is where DS’s founder studied and where much of DS’s talent is drawn.

Hangzhou’s local government, a CCP party aligned organization, provided additional support to AI companies through tax incentives, subsidized real estate, and infrastructure investments, aligning regional initiatives with national goals. Hangzhou is central to China's AIDP strategy and largely missed by analysts. As of 2024, Hangzhou's GDP exceeded 2.18T yuan, approximately $299B, which rivals San Francisco's GDP. Hangzhou has been recognized as a leading city in AI development, ranking second in national assessments. This recognition reflects the city's robust AI ecosystem, characterized by a high number of AI-related patents and significant projects like Alibaba's "City Brain 2.0," which is widely deployed within the city. Hangzhou researchers at Alibaba led the development of and participated in China's pioneering City Brain project starting in 2016, which for Hangzhou saw, "ambulance response time dropped by 50% and the average travel time dropped by 15.3%. Moreover, the accuracy of traffic incident real-time detection reaches 95%."

A Bit About DeepSeek

DeepSeek, founded in 2023 as an offshoot of the quantitative hedge fund High-Flyer founded in 2015, both based in Hangzhou, exemplifies the success of the CCP’s centralized model. High-Flyer’s development of the Fire-Flyer supercomputers in Hangzhou (I in 2020 and II in 2023) provided the foundational infrastructure for DS’s AI research, including its landmark R1 model. The Fire-Flyer systems were likely deployed in Hangzhou’s AI innovation ecosystem, benefiting from government incentives and shared infrastructure.

DeepSeek's founder, Liang Wenfeng, also and uncoincidentally participated in a symposium hosted by Premier Li Qiang on January 20, 2025 (upon the R1 release), where he, along with other experts and entrepreneurs, provided input on the draft Government Work Report. The Government Work Report is a comprehensive annual policy document presented by the Chinese Premier during the National People’s Congress (NPC), outlining the central government’s priorities, achievements, and goals for the coming year. It serves as a key roadmap for national development and governance. This involvement indicates recognition at the national level and directly implicates DeepSeek's development as a direct participant in locally-driven political affairs that saw its development in Hangzhou.

China’s AI ecosystem operates as a centralized, cascading system, where local and national governments align resources and policies with state priorities. Hangzhou’s status as a leading AI hub is directly tied to this model, as its infrastructure and policy support allowed companies like High-Flyer and DeepSeek to scale rapidly. By focusing on AI-enabled industrial applications, talent development, and reducing the costs of AI model training, China has built a cohesive pipeline from innovation to application, tightly controlled by the CCP to ensure alignment with national security and economic goals.

This highly integrated strategy highlights how China’s AI agenda extends beyond technology, embedding AI into its governance and global ambitions. DeepSeek’s cost-effective AGI research contributes to China’s ability to scale AI across military and civilian domains, securing its role as a global leader while limiting dependencies on foreign technologies.

The US AI Approach (2017–2025): A Fragmented Strategy in an Era of Global AI Competition

The US AI ecosystem has long been a powerhouse of innovation, yet its development has been fragmented, reactive, and largely driven by the private sector rather than a cohesive national strategy. Unlike China’s centralized approach, the US has relied on commercial leadership, with government agencies playing a supporting role through programs like DARPA’s AI research, NSF funding, and limited DoD initiatives like the Joint AI Center (JAIC) and the newer Chief Digital and AI Office (CDAO).

Federal involvement in AI has historically been decentralized, with multiple agencies funding and directing AI research in silos. DARPA has led military-focused AI research since the 1960s, the NSF has funded fundamental AI and machine learning advancements, and the Department of Energy (DOE) has driven high-performance computing initiatives. However, these agencies have lacked a unifying framework that points to a national strategy, leading to inefficiencies in translating innovation into national security or economic competitiveness.

While the US remains a leader in commercial AI—thanks to firms like OpenAI, Google, and Microsoft—its lack of a coordinated strategy has led to systemic vulnerabilities. This fragmentation has resulted in redundancy, delayed deployment, and an overreliance on short-term market incentives rather than national AI development.

Attempts at Coordination: The American AI Initiative & the National AI Initiative Act

The American AI Initiative, introduced by President Trump in 2019, marked the first formal attempt to prioritize AI development at the federal level. However, it was largely symbolic, as it lacked dedicated funding and enforceable mandates, leaving much of the responsibility to individual agencies and private-sector actors. This initiative attempted to unify disparate efforts under a broader strategy but failed to provide the necessary resources and coordination to execute on its vision.

The National AI Initiative Act of 2021 sought to address these gaps by establishing the National AI Initiative Office (NAIIO) to coordinate AI research and policy. However, the NAIIO’s authority remained limited, and governance remained fragmented across multiple agencies, leading to inconsistencies in investment priorities and long-term AI planning. Furthermore, shifting political priorities have meant that US AI policy has been subject to frequent changes, further exacerbating uncertainty in both government and industry. A notable recent example of fragmentation includes the establishment of the US Artificial Intelligence Safety Institute (US AISI) within NIST.

Infrastructure Gaps and Strategic Risks

Unlike China’s top-down approach, where the government funds AI-specific supercomputing centers and data infrastructure, the US lacks a comparable state-backed AI infrastructure initiative. In the waning days of the Biden Administration, Executive Order (EO) 14141 attempted to rectify this situation by making Federal land available for such infrastructure. Most AI compute resources remain concentrated in a handful of private companies, leaving national security AI projects dependent on commercial cloud providers like Microsoft Azure, Google Cloud, and AWS. The Biden Administration had attempted to create established national guidelines for AI safety and equity with EO 14110, however, in early 2025, it was rescinded. Instead the new Trump administration has solicited a Request for Information (RFI) on a National AI Action Plan.

Additionally, the US has struggled with talent retention and workforce alignment. While China integrates AI education into its national curriculum and channels talent into state-prioritized AI sectors, the US has largely left workforce development to market forces. The result is an AI talent gap in key national security domains, where expertise is either concentrated in private-sector AI labs or lost to foreign competition.

The Disjointed Approach to Military AI Integration

China’s AI strategy seamlessly integrates military, commercial, and academic innovation, whereas the US military’s approach to AI has been marked by bureaucratic delays and inefficiencies. The Department of Defense (DoD) launched JAIC in 2018 to centralize AI adoption across military branches, but it struggled with execution and bureaucratic resistance. By 2022, the JAIC was merged into the DoD CDAO in an effort to streamline AI and digital transformation, but progress has remained slow compared to China’s rapid military AI adoption.

Moreover, defense-oriented AI programs such as Project Maven—which sought to use AI for intelligence analysis—have faced backlash from industry employees and human rights advocates, leading to tensions between the private sector and national security interests. Although this tension has since subsided, it significantly offset the progress of the program. This contrast sharply with China’s AI ecosystem, where government-aligned firms face no such ethical resistance to military applications.

A Path Forward: Strengthening the US AI Strategy

If the US is to maintain its leadership in AI, it must reject incrementalism and act decisively to close the gaps in its AI strategy. The following reforms, some of which are occuring piecemeal, are essential:

1. National AI Infrastructure – Establish Federally funded AI supercomputing centers and secure access to AI-specific semiconductor supply chains, reducing dependence on commercial cloud providers and foreign chip manufacturers.

2. Strategic Public-Private Partnerships or Compacts – Expand government-industry collaboration to align private-sector AI research with national security and economic goals, ensuring AI breakthroughs translate into tangible national advantages. This includes incentivizing US companies to invest in AI infrastructure through tax breaks, grants, and streamlined procurement mechanisms.

3. AI Workforce Development – Implement a federally supported AI talent pipeline, including AI-focused university programs, national security fellowships, and visa reforms to attract global AI talent.

4. Defense AI Modernization – Accelerate the integration of AI across military and intelligence domains, improving the efficiency of AI adoption through streamlined procurement and innovation hubs within the DoD.

5. Long-Term AI Research Investment – Fund AI research beyond commercial applications, focusing on foundational advancements in artificial general intelligence (AGI), next-generation machine learning, and AI safety research.

6. National AI Doctrine – Establish a unified federal AI authority beyond just NAIIO to drive a coordinated, long-term AI strategy. We should lean in on developing global AI standards rooted in democratic principles, offering an alternative to China’s authoritarian model.

The Future Belongs to Those Who Act

The US has significant talent, venerated companies, and the intellectual firepower to lead AI—but lacks the decisive national strategy to translate innovation into dominance. While China’s centralized model is incompatible with democratic values, that does not mean we should continue sleepwalking through decentralized inefficiency. AI is a national security priority, and if we fail to act with urgency, we will find ourselves following rather than leading. The US must stop managing AI like a private-sector playground and start treating it as a critical pillar of economic and defense power. We don’t need to mimic China—we need to outmaneuver them. That means aligning state incentives with industrial ambition, fast-tracking AI infrastructure investments, and rejecting bureaucratic hesitation. It’s time to break inertia. The future isn’t won by those who wait—it’s won by those who move.