What the first United Nations Global Dialogue on AI Governance means for investors

“We may be the last generation able to set the terms on which humanity and machines coexist.”

UN Secretary-General António Guterres’ warning set the tone for the inaugural UN Global Dialogue on AI Governance. The conference, which drew senior world leaders, civil society and business executives, demonstrated how far AI has climbed up the global agenda.

AI is no longer only a technology story: it has become a defining geopolitical issue, shaping relations between states, concentrating economic power at remarkable speed and testing whether international institutions can govern a technology that evolves faster than policy. For investors, the question has shifted from whether AI governance matters to how to shape it, capitalising on opportunities whilst guarding against risks.

A new global platform

The Dialogue, established by the UN General Assembly, is the first UN platform for member states to convene on AI governance, informed by the Preliminary Report of the Independent International Scientific Panel on AI. A full report will be published ahead of the next Dialogue in New York in May 2027.

A concentrated technology in a divided world

Concentration of AI capacity ran through the entire conference. 90% of the compute power across the 500 largest known AI clusters sits in just two countries, the United States and China, with 75% in the US alone. Put differently, 191 of 193 UN member states lack independent compute capability.

The benefits and risks of AI are just as unevenly distributed. Wealth creation is concentrated almost entirely where compute is, while the risks – from labour market disruption to information harms – are global. Countries from the Global South were clear that access to tools is not enough; they require technical assistance and genuine participation in AI development, a dual development and investment opportunity.

UN member states also diverge on how AI should be governed. The United States has opposed centralised UN-level AI governance, while China supports it. Geopolitical positioning is intensifying: China was highly visible at the Dialogue and is building multilateral coalitions on AI governance and capacity, including the UN Group of Friends for International Cooperation on AI Capacity Building. Fossil-fuel-rich states, meanwhile, positioned themselves as ideal hosts for AI infrastructure, raising concerns about the energy systems powering data-centre growth and emissions.

Investors are increasingly considering how geopolitical tension will affect portfolios, and how to reduce risk exposure while benefiting from AI-fuelled growth.

The evidence dilemma

The timeline for action is tight, with a poll at the PRI’s 20th anniversary event in London finding that 66% of respondents consider AI risk as posing equal or greater downside potential in the coming five years than climate risk

However, the evidence and regulatory frameworks needed to act are not yet in place. By the time evidence materialises, the risk has usually already occurred, and policy may be too late to help. Given the potential scale and short-time horizon of AI risks, this “evidence dilemma” cannot be an excuse for inaction.

Pragmatism was called for: evidence need not be perfect, but rather timely, directional and useful. As Maria Ressa, co-chair of the Panel and Nobel Peace Prize laureate, put it: “If you don’t act now, AI will run on without you.” Investors can move quickly to supply policy makers with useful data and reinforce other stakeholders’ calls for urgent action.

The regulation gap

While little concrete AI regulation exists or was flagged at the conference, the EU AI Act and Singapore’s Agentic AI Governance Framework were the main policies referenced. These will likely become defining models for many emerging regulatory frameworks globally, with the hope that this can lead to convergence around shared principles and standards.

Appetite for better regulation is strong, particularly across the Global South, and was consistently framed as a driver of trust rather than an innovation brake. One theme recurred across sessions: whatever governance emerges must be highly adaptive, given how fast AI and its risks evolve.

Absent regulation, one plenary panel pointed to demand for AI vendor responsibility as a key driver of better governance, mirroring the World Benchmarking Alliance Collective Impact Coalition for Ethical AI’s approach to developer stewardship. This PRI Award-winning initiative is an excellent example of how investors can leverage their capital and capabilities to advocate for AI risk mitigation.

The investor voice

For the PRI, perhaps the most striking observation from Geneva was who wasn’t in the room.

Across two days spent discussing a technology reshaping the global economy, the investment community (or reference to investment) was largely absent from the stage. This is a remarkable gap. The concentration risk that dominated the conference is itself a largely financial phenomenon, visible in index concentration, valuations and capital flows to large-scale data centre buildout. Investors hold unique levers, perspectives and incentives, from capital allocation to stewardship, that few other actors can match.

There is also a unique opportunity for investors – given their modelling and analytical strengths – to play a central role in shaping effective policy by growing the body of evidence around AI risk materiality. This work can help inform policymaking in the most impactful arenas, while supporting further innovation and growth. .

What this means for investors

Three implications stand out from the conference.

1. Investors are well positioned to build the evidence base on AI risk materiality. The clearest request from policy makers in Geneva was for better evidence, with an understanding that it does not need to be perfect. Investors observe AI risk where it becomes financially material across portfolios, sectors and geographies, and can supply the financially grounded evidence that policy makers need. The PRI is convening signatories across markets and at PRI in Person to support peer sharing that grows this evidence base. Please get in touch if you are interested in becoming involved in the PRI’s AI governance work.
2. Investors can engage on AI policy. With relatively little regulation globally, and many governments actively seeking reference points, early investor input has an unusual opportunity to shape better regulation. The PRI is exploring what supporting signatories in this work would look like – please get in touch if you would like to discuss.
3. Investors can start evaluating concentration risk at both country and developer levels. The concentration of compute power is a geopolitical issue with significant portfolio risk implications. It is also an AI risk we hear about less from signatories. Understanding that exposure, and how it interacts with policy, energy systems and market structure, will become a core part of AI governance for investors. The PRI is partnering with stakeholder organisations to help bring technical expertise to our signatory base that can support this work.

The Dialogue showed AI governance will be shaped in the coming years, with or without investor input. Investors that engage early will not only manage the risks better, they will help define what responsible AI looks like for markets globally.