The $900 billion question: Redistributing control of AI development

Most people will never step into a data center, write a line of code, or develop a machine learning model. Yet artificial intelligence is already shaping decisions that affect nearly every aspect of our lives. It operates quietly yet decisively, influencing courtrooms, hospitals, border checkpoints, and job applications. As AI becomes embedded in these systems, power concentrates with those who control it, not those who live with its consequences.

By 2027, global private investment in AI startups is projected to reach $900 billion, with more than 80 percent of that capital originating from just three countries: the United States, China, and the United Kingdom. These nations shape not only innovation but also the infrastructure, research priorities, and policy frameworks that influence global outcomes. As capital and computational resources consolidate, power accrues to a shrinking circle of firms that design and deploy the infrastructure of modern decision-making. Algorithms increasingly shape the distribution of essential goods: who receives a loan, who gets hired, and who qualifies for public support. These systems function less like market tools and more like public utilities, but without the oversight, transparency, or accountability that public infrastructure demands.

The real power of AI lies not in its technical sophistication but in the authority it grants over the systems it mediates. A concentrated group of companies, including familiar names such as Alphabet, Microsoft, Amazon, Meta, OpenAI, Baidu, and Tencent, are setting the standards, controlling access, and defining the economic logic of the digital infrastructure billions rely on each day. Others are left to navigate systems they did not design and cannot meaningfully influence. This dynamic evokes an earlier era, when the ability to build and control physical infrastructure, such as railways or electrical grids, determined economic and political power. Today, that leverage is digital, but the pattern remains the same.

It is the ability to imagine and build alternatives, to construct digital infrastructure that reflects the values, histories, and ambitions of the societies it serves. For Europe, the question is whether to claim that role or to drift into dependency. The alternative is not a mirror image of Silicon Valley or Shenzhen. It is a public architecture of AI: open, accountable, and built with the slow confidence of democratic intent.

This perspective is shaped by work at the intersection of technology, policy, and global development. Efforts, such as the United Nations High-Level Advisory Body on Artificial Intelligence, have demonstrated that global coordination is possible, but only when power is distributed, voices from the Global South are included, and investment aligns with the public interest. These lessons are reflected in the Patrick J. McGovern Foundation’s partnership with France’s Current AI initiative, a €400 million collaboration that enables regional leadership, fosters cross-sector engagement, and supports the long-term development of public infrastructure. When capital reaches this scale and is directed with global intention, it opens the door to an AI ecosystem built for social purpose and democratic integrity.

These experiences have sharpened a core belief. When algorithms are used to guide decisions in education, healthcare, finance, or public service delivery, their design and deployment must be transparent, participatory, and subject to public accountability. Performance alone is not enough. Legitimacy stems from whether these systems accurately reflect the values and lived realities of the societies they impact. That includes local customs, regional priorities, and a meaningful sense of ownership over the tools that govern daily life. Without that grounding, even the most sophisticated systems risk deepening exclusion rather than expanding agency.

Some countries are beginning to turn these principles into practice. India provides a clear example of what inclusive ambition can achieve when combined with strategic investment. As co-host of the 2025 Paris AI Action Summit and the upcoming host of the 2026 conference, India is positioning itself as a global leader on public-interest AI. But the signals go deeper than diplomacy. The government has launched a national compute grid that extends affordable access to millions of developers. It has created AIKosha, a platform that curates multilingual datasets and pre-trained models for public use. Paired with frameworks like the Data Empowerment and Protection Architecture, these efforts reflect a broader philosophy: that digital infrastructure should be both widely accessible and shaped by the people it serves. This approach does more than widen participation. It begins to redraw the boundaries of where innovation originates, and who it is intended to include.

But the scale is still off. In 2024 alone, 4 US companies, xAI, Databricks, Anthropic, and OpenAI secured $40 billion in funding (FDI Intelligence, 2024). That kind of capital flow determines which technologies scale, who accesses them, and whose priorities guide their design.

This moment demands more than critique. If the foundational infrastructure of AI continues to be shaped by a small group of actors, the systems that govern education, healthcare, and the economy will reflect their priorities.

Diagnosis is not enough. If the foundational infrastructure of AI remains in the hands of a narrow group of actors, the systems that shape education, healthcare, and the economy will reflect their priorities, rather than the common good. The challenge now is to move from critique to construction and to articulate a credible alternative that aligns technological development with democratic values and long-term public benefit. That shift begins with concrete interventions. The actions below offer a path forward.

• Invest in Public Infrastructure: Governments and multilateral organizations should support open-source models, accessible compute power, and high-quality datasets that lower barriers to participation.

• Create a Global Research Commons: Institutions such as the European Commission or UNESCO can coordinate shared research platforms to ensure global access to knowledge.

• Expand AI Education and Capacity: Technical training, fellowships, and literacy programs must be made available globally to cultivate and retain talent across regions.

• Align Procurement With Public Values: Public sector AI use should reflect democratic principles by demanding transparency, interoperability, and ethical design.

• Build New Governance Institutions: Institutions modeled on initiatives like the Global Digital Compact can offer scientific oversight, monitor risks, and create accountability mechanisms for frontier AI development.

Europe’s advantage is not scale but structure. Its strength lies in a legacy of collaborative governance, clear regulatory frameworks, and the ability to coordinate across public, private, and academic sectors. These institutional assets enable the shaping of AI development around public purpose rather than market expedience. Initiatives like Current AI demonstrate how co-investment can foster infrastructure grounded in transparency, accountability, and long-term value. Europe also brings specific technical leverage. It plays an outsized role in setting standards for B2B data flows, enterprise interoperability, and industrial AI applications-areas often overlooked in consumer-focused debates. Combined with trusted data protection regimes and a strong academic research base, these capabilities position Europe not only to govern responsibly at home but to influence how infrastructure is designed globally.

European businesses have a stake in ensuring AI systems are built to reflect their operational realities and regulatory environments. Financial firms depend on models that meet strict compliance and auditability standards. Hospitals and care networks require systems that align with ethical norms regarding patient data and clinical safety. Industrial AI must integrate with legacy infrastructure and regional supply chains. Models trained and tested elsewhere, under different legal, cultural, and market assumptions, often fall short. What is required now is coordinated investment in systems that are not only interoperable and secure but also designed from the outset to serve public and sectoral priorities across Europe.

But Europe has the tools to offer a different model, one built on democratic trust, institutional competence, and a commitment to long-term public benefit. The opportunity is not simply to regulate AI at home, but to lead globally by demonstrating what it looks like to build digital infrastructure in service of human dignity and collective progress - this could become both a responsibility and an opportunity.

This article is also published on Institut Montaigne. illuminem Voices is a democratic space presenting the thoughts and opinions of leading Sustainability & Energy writers, their opinions do not necessarily represent those of illuminem.

See how the companies in your sector perform on sustainability. On illuminem’s Data Hub™, access emissions data, ESG performance, and climate commitments for thousands of industrial players across the globe.