By: Andreas Kuehn
The following piece originally appeared in ORF’s Expert Speaks: Raisina Debates on August 28, 2026.
The race for artificial intelligence is rapidly shifting. AI leadership is often framed narrowly in terms of frontier models, GPUs, and select applications. But there is an underlying development, one that is much more significant: the multi-trillion-dollar buildout of AI infrastructure. The core of AI competition has shifted from model capability to infrastructure buildout, where private capital, not government, is the principal driver. Most policy analysis, however, still focuses on chips, models, and export controls, and in doing so, misses the more consequential race for infrastructure enabled by global capital markets. In this next phase of the AI competition, the decisive question is not who can produce the most capable models, but who can build, finance, power, and deploy the infrastructure that will allow AI to scale.
Others have written about AI infrastructure and geopolitics, but existing analysis tends to lead with the role of states, when in fact governments do not build data centres. And while the sheer cost of the buildout is well documented, the role of capital markets in financing massive next-generation infrastructure, and the strategic imperative to align private investment with states’ interests, remain underexplored. Unquestionably, governments play an important role in shaping this new geography of AI infrastructure, but they are not the principal architects. States can set priorities, create incentives, and define trusted-partner frameworks. The still-nascent Pax Silica initiative, led by the US State Department, points to the strategic value that a fully-fledged technology alliance among like-minded states could eventually deliver. However, it is the technology firms that develop, build, and operate next-generation digital infrastructure, while investors are the real drivers of this future.
AI is viewed through the lens of economic competitiveness and national security, a contest currently playing out between the United States and China. However, the new geography of AI infrastructure is one in which capital is not peripheral to geopolitical competition, but one of its principal instruments.
Technology firms are making large bets on compute, cloud capacity, semiconductors, and regional infrastructure hubs. Infrastructure investors increasingly view AI as a strategic rather than a passive investment, while sovereign wealth funds and public subsidies support AI projects at massive scale. Yet private-sector primacy comes with structural tensions. As governments grow dependent on enormous amounts of private capital and corporate technological capabilities to execute their strategic objectives, there is a risk of misalignment when investor interests diverge significantly from national strategic interests.
Microsoft has announced a US$3 billion AI and cloud investment in India and a US$2.9 billion commitment in Japan, while Amazon plans to invest US$9 billion in cloud infrastructure in Singapore by 2028. Foreign technology firms have also invested in semiconductor fabrication in the United States under the CHIPS and Science Act, reinforcing US domestic capacity through inward investment. These investments will be largely responsible for reshaping regional AI capacity and deepening infrastructure ties among partners across markets.
Even with concerns about an AI bubble, investment in data centre infrastructure is projected to reach US$7 trillion by 2030. The AI Infrastructure Partnership has assembled US$30 billion in private equity capital for US data centre buildouts, while the Stargate Project signalled investment ambitions to the tune of US$500 billion over four years. Adding an international dimension to this effort, Stargate UAE supports the buildout of sovereign AI capacity in Abu Dhabi. It has also underscored the vulnerability of these multi-billion-dollar projects after Iran threatened to attack the facility. Major infrastructure investors such as BlackRock and Brookfield Asset Management, along with sovereign wealth funds including Singapore's GIC and Temasek and the UAE's Mubadala, are no longer considered passive investors. Their capital is influencing where AI capacity will be built, and which ecosystems will gain strategic depth. Lastly, Alphabet's rare 100-year bond issuance, which raised US$31.5 billion, further illustrates the long-term financing strategies now being mobilised.
Driving AI Leadership Through Infrastructure Buildout
Private investment decisions do more than determine where infrastructure is built. They set de facto standards, shape diffusion, and drive deployment at scale. If, in fact, private firms and capital markets are the real decision-makers, alliance coordination for the AI buildout must be designed around their logic, not the government's. As infrastructure expands across trusted-partner networks, it reinforces the global spread of the US AI stack. Infrastructure investment is a mechanism of strategic influence, and it determines whose technologies become embedded in key markets, whose platforms will scale, and which political-economic networks will define the terms of AI adoption. Yet AI infrastructure investments face geopolitical risk that the private sector is traditionally less familiar with managing. Government frameworks like Pax Silica offer private investors a means by which to navigate geopolitical risk more effectively. Conversely, governments are less equipped to price risk and sequence investments efficiently, tasks that capital markets perform as a matter of course.
Coordination among partners, such as under the Pax Silica framework, should be understood as a defensive effort to reduce supply chain risk and strengthen ecosystem resilience, but also as a strategic mission to build aligned networks of trust and interdependence. This is particularly true when viewed as a process to compete with China. Securing AI leadership in this phase of competition requires an infrastructure-first strategy, coupled with a new kind of economic statecraft: one that works with the grain of private capital rather than against it, and one that creates the necessary conditions that make strategic alignment the path of least resistance for private capital.
Without such an approach, private capital will continue to flow according to its own logic, sometimes aligned with strategic interests, often not. For the private sector, the cautionary signal is equally clear: infrastructure built outside trusted-partner frameworks carries growing exposure to regulatory disruption, export control regimes, and economically costly geopolitical fracture. Firms and investors that treat geopolitical alignment as optional rather than structural will find themselves on the wrong side of that divide.
The key question is not who invents the most advanced AI, but where AI infrastructure is built, who finances it, and under whose terms it can be deployed at scale. The power in this next phase of AI competition will rest with those most able to shape the geography of AI infrastructure to their advantage.
Dr. Andreas Kuehn is a Senior Fellow for Technology Policy at Observer Research Foundation America.