Three Vectors of US AI Dominance: Innovation, Infrastructure, and Security

The global Artificial Intelligence (AI) race could enter a new phase driven by three significant changes: the Trump administration’s assurance to allow Nvidia to resume sales of H20 AI chips to China; increased demand for advanced data centres and semiconductors across the world, and how AI is likely to transform the energy usage landscape globally. As models grow hungrier for data, energy, and compute, countries led by the United States (US) and China are competing to secure dominance in the vectors that power AI: chips, cloud services, data centres, and energy. While the US and China remain locked in a high-stakes competition, emerging regions are increasingly becoming battlegrounds for influence and increasing capacity. The Trump administration is moving in a direction where helming the three sectors of AI development and training – innovation, infrastructure, and security – will likely shape its competitive edge over China and global dominance.

In the AI innovation domain, the Trump administration has assumed a starkly contrasting view from its predecessor. Nvidia was embattled with the US government with restrictions to sell chips to China since the Biden administration’s ‘diffusion rule’, which revised the Export Administration Regulations (EAR) controls on advanced computing integrated circuits (ICs) to China and added new controls on AI model weights for certain advanced closed-weight dual-use AI models. The second Trump administration, building on the previous administration’s restrictions, imposed new restrictions on US  chip exports to China in April 2025. The latest move by the Trump administration to allow Nvidia to sell chips to China upends all previous efforts and could reshape the global AI race.

The Trump administration is keen on developing key partnerships outside of China to advance in the AI race, most notably with Saudi Arabia and the United Arab Emirates (UAE).

These developments come on the eve of the Trump administration’s decision to launch its AI Action Plan. The plan hinges on three key pillars – accelerating innovation, building AI infrastructure, and leading in AI diplomacy and security. Arguably, some of the steps taken by the Trump administration in recent times appear in sync with this thought. First, the decision to allow the sale of Nvidia H20 chips to China is thought to retain the comparative advantage for the US by slowing innovation in China. It seeks to achieve this by reopening supply lines and preventing China from breakthrough innovations in the field. Second, the H20 chips are considered to be the third- or fourth-best amongst US-made chips, with the top-tier models reserved for domestic use. These changes in the US AI policy come against the backdrop of a series of developments, both global and domestic.

At the domestic turf, leading AI companies such as Nvidia and Advanced Micro Devices (AMD) played to the gallery with the Trump administration, and Nvidia’s Chief Executive Officer (CEO) Jensen Huang almost became a Trump whisperer, convincing him that the odds are in favour of the US, provided Washington competes openly rather than imposing restrictions.

Abroad, the Trump administration is keen on developing key partnerships outside of China to advance in the AI race, most notably with Saudi Arabia and the United Arab Emirates (UAE). Strategically, the Middle East has been the only key region that Trump has visited in his second term. As Trump toured the Middle East in May 2025, several key deals were crafted with two Middle East allies—Saudi Arabia and the UAE—that would power America’s AI ambitions. Many US companies, including Nvidia, AMD, and QUALCOMM, announced deals with both countries, with the Trump administration claiming to have secured US$600 billion in investments from Saudi Arabia alone. The agreement with Saudi Arabia is expected to ensure 18,000 of Nvidia’s advanced ‘Blackwell’ chips for Humain, which is a Saudi start-up launched by the country’s sovereign wealth funds. The UAE struck a much bigger deal with Nvidia, slated to receive as many as 500,000 advanced chips from the company annually.

Energy and Infrastructure: Two key flanks of AI

AI’s rapid evolution is expected to have two consequential impacts on the need for unique AI-related infrastructure and energy usage. An Instagram reel, a news feed article, a financial transaction or even a Google search are actions contributing towards growing energy demands that fuel the expansion of data centres. These centres have become the heart of global infrastructure and require access to vast amounts of energy inputs and strategic geographical placements. The growing dominance of cloud-based data centres, hyperscale campuses, and China’s underwater infrastructure illustrates a shift in how location can be leveraged for efficiency and cost. Among the AI infrastructure, the demand for data centres has increased over the past decade, driven by the exponential growth of AI. In 2024, AI accounted for 24 percent of servers and 15 percent of total data centre electricity use. By mid-2025, its share rose to 20 percent, with projections suggesting it could reach nearly 50 percent by the end of the year. At this pace, global data centre power demand could rival Japan’s annual electricity consumption by 2030.

AI’s rapid evolution is expected to have two consequential impacts on the need for unique AI-related infrastructure and energy usage.

This demand, driven by the rise of generative AI, is entering a new phase of high-density development. The expanding material footprint of AI directly correlates with growing power consumption triggered each time a prompt is entered on platforms such as ChatGPT.  Key US data centre hubs – Virginia and Texas – are now witnessing data centres consuming 25 percent and 4.5 percent of their electricity demand, respectively, with growth outpacing local energy capacity expansion.

The rollback of clean energy incentives in the ‘Big Beautiful Bill’ raises concerns and doubts alike about sustainability and whether conventional power alone can meet the evolving digital infrastructure. Despite growing concerns over sustainability and power shortages, the US is expanding its domestic and global AI infrastructure through investments in data centres. A key example is the AI Infrastructure Partnership (AIP) – a global initiative backed by BlackRock, Global Infrastructure Partners (GIP), Microsoft, NVIDIA, and xAI. AIP aims to mobilise up to US$100 billion in capital to invest in next-generation AI infrastructure, including advanced data centres and supporting energy systems. With NVIDIA serving as a key technical advisor, the partnership reflects a growing public-private convergence in the US. Building on this, the US-UAE AI Acceleration Partnership is centred on a major data centre in Abu Dhabi under the Stargate UAE initiative. However, the partnership is not without its own set of frictions/challenges. A major point of concern revolves around G42, an Abu Dhabi-based AI firm central to the Stargate initiative whose previous collaborations with Chinese entities have triggered concerns in the US related to national security, particularly over the possibility of advanced American chips being diverted to benefit Beijing.

While the Biden presidency focused on integrating AI growth with clean energy goals – through the CHIPS and Science Act and executive orders on AI risk – Trump’s approach pivots towards scaling AI infrastructure driven by competition with China and choosing the right partners abroad. This includes revoking Biden’s executive order on AI risk and proposing large-scale ventures such as Stargate—a US$500 billion AI infrastructure initiative supported by OpenAI, SoftBank, and Oracle.

At the core is the projection of American technological leadership. Nvidia chips, OpenAI models, and a tight-knit web of public-private partnerships define the US digital architecture, which is aligned with sustainability commitments and local capacity-building efforts. In contrast, China’s model, led by Alibaba Cloud, Huawei, and Tencent, prioritises rapid deployment and scalability under the aegis of Digital Silk Road. Beijing has used its Digital Silk Road to offer low-cost infrastructure bundled with Huawei’s Ascend AI servers and turnkey cloud solutions.

Among the AI infrastructure, the demand for data centres has increased over the past decade, driven by the exponential growth of AI.

While rapid digital transformation may seem advantageous to governments, it also raises concerns about data sovereignty and long-term data dependency. Washington has been keen to spotlight these concerns in its dialogue with Southeast Asian partners. Notably, the Association of Southeast Asian Nations (ASEAN) have chosen a ‘third way’ as a balancing act between the growing US-China tug-of-war. As they welcome Chinese investments and low-cost solutions, they also focus on the best-in-class technology and regulatory models from the US and Europe. Nonetheless, this balancing act has raised contradictions due to the unequal readiness of AI. Singapore has advanced digital frameworks, while countries such as Myanmar significantly lag, posing issues in setting up coherent regional standards for ethical AI and digital sovereignty Together, these dynamics indicate that the Southeast Asia region has become a strategic theatre for investment and infrastructure, with both China and the US competing for dominance in the global AI and data centre race.

As the AI race deepens, new geopolitical frontiers have emerged. China’s experimentation with underwater data centres signals a willingness to reimagine infrastructure. Concurrently, policy shifts—including those under the ‘Big Beautiful Bill’—could undermine American tech giants’ efforts due to domestic uncertainty around power availability and sustainability. Nevertheless, China’s ambitious AI push also faces structural constraints with respect to increasing energy demands and power supply shortages due to coal-heavy grids. Recent policy shifts towards green data centres reflect Beijing’s recognition of these structural vulnerabilities. As both powers confront these challenges, the race for AI infrastructure is hurtling on. 

Vivek Mishra is the Deputy Director of the Strategic Studies Programme, Observer Research Foundation.

Himanshi Sharma is a Research Intern at the Strategic Studies Programme, Observer Research Foundation.

• International Affairs
• China
• India
• USA and Canada
• Middle East
• artificial intelligence action plan
• global ai dominance
• global ai race
• us ai action plan
• us ai policy
• us dominance in ai
• us role in ai development

The views expressed above belong to the author(s). ORF research and analyses now available on Telegram! Click here to access our curated content — blogs, longforms and interviews.

PREV NEXT