Geopolitics and Critical National Infrastructure

Report from the 2026 GZERO Summit in London

October 2026

By Stephen Hermanson

Introduction

It was a pleasure to return to the GZERO Summit in London¹ on 8 October 2026. As in 2025, Eurasia Group and LSEG assembled a strong line-up that spanned a broad agenda of geopolitical and policy risks.

A few weeks before, I came across an article in The Times by Gerard Baker (17 September 2026) aimed at the UK Prime Minister, Andy Burnham: “Grin and bear it, Andy, and stick with Trump – The US-UK relationship may no longer be so special but Mark Carney’s idea of a new ‘middle powers’ bloc is a non-starter”.² Carney had set out the idea at Davos in January, arguing that “the middle powers must act together, because if we’re not at the table, we’re on the menu”.³

I anticipated the GZERO Summit would return in part to the Middle Powers question, so in addition to a general update, I decided to use the conference to take a temperature check on the Middle Powers concept – picking up a thread from my earlier report on the Chatham House London Conference 2026.⁴

I sense there’s a secret doubt amongst Middle Powers proponents that concedes Gerard Baker’s premise is possibly quite accurate. I’ve also wrestled with the seemingly intractable list of obstacles that a successful Middle Powers coalition would need to overcome – chief amongst them the prisoner’s dilemma and the game theory of breaking ranks.

The Prisoner’s Dilemma

The logic is simple and unforgiving. Each Middle Power is better off if all of them hold the line together against pressure from Washington or Beijing. But each one, acting alone, can usually secure a better short-term outcome by striking its own deal. So everyone defects, and all end up worse off than if they had cooperated.

We’ve seen this play out. Faced with US tariffs in 2025, the UK, Japan, the European Union (EU) and South Korea each negotiated separate arrangements with Washington rather than a coordinated response, and Canada withdrew its digital services tax under US pressure. Each decision was rational on its own terms; together they confirmed that Middle Powers can be picked off one at a time. I sense this sits at the heart of Baker’s scepticism, and it’s a fair challenge.

Variable Geometry: A Different Kind of Alliance

A useful insight that took shape during the summit was the different modes a Middle Powers alliance would need to adopt. The broad idea, bear with me, is that a Middle Powers alliance doesn’t look and behave like a traditional alliance such as the EU or NATO. Instead, the members form and reform around salient issues depending on their exposure and leverage – especially in relation to the US, China and Russia, where each Middle Power has a slightly different set of problems and expectations. Carney calls this “variable geometry, in other words, different coalitions for different issues based on common values and interests”.⁵

This matters for the prisoner’s dilemma. A one-off game rewards defection; a repeated game changes the calculation. When the same players meet again and again – on trade, defence, technology and energy – reputation and reciprocity start to count, and cooperation becomes the rational strategy. That was Robert Axelrod’s insight in his work on “tit-for-tat”.⁶ Issue-based coalitions, constantly forming and reforming, are in effect a repeated game.

Some traditional institutions such as the UN probably operate in a similar way, where members lean into or away from issues and subjects depending on their capacity and interest. Of course, the UN is so broad that the effect is often diluted, contorted or deadlocked. As a smaller group, Middle Powers could do better.

The key, we heard, was to ensure the alliance focused on the future as opposed to relitigating past issues. For example, if the EU were created today (in the guise of a broader Middle Powers alliance that could include the UK, Canada, Japan, South Korea, Australia and others), then security and cost of living would be fundamental drivers alongside single markets, the rule of law and other principles.

There are early signs of variable geometry in practice. In June 2026, Canada became the first non-European country to join the EU’s €150 billion Security Action for Europe (SAFE) defence loans instrument⁷, while the UK’s own SAFE negotiations collapsed in late 2025 over the size of its financial contribution,⁸ a reminder that the national interest can still trump the collective one. In March 2026, members of the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP), a 12-nation Trans-Pacific trade pact that the UK joined in 2024, and the EU issued a joint ministerial statement committing to cooperate on rules-based trade, supply chain resilience and economic coercion.⁹

Europe as the Anchor

The tone at the summit in 2025 was Europe in irreversible decline, but the discussion in 2026 showed signs of Europe repositioning itself as a trusted and reliable partner in a Middle Powers bloc, with Europe contributing strengths such as infrastructure, relative stability and the rule of law. The concern was fragmentation in Europe – itself an experiment in whether Middle Power coalitions could align sufficiently to make a difference. National interest and competency caveats across a range of dossiers seem to conspire to fragment the European alliance on the precise priorities it’s seeking to grasp, with security and defence as prime examples.

There’s no doubt that flexibility is important in keeping an alliance together and sharing the benefits and costs. It was interesting to hear that perhaps the EU could be more flexible with the UK as it develops a new relationship with the EU – not necessarily re-entry (although Prime Minister Andy Burnham said in late September that rejoining is one of the options),¹⁰ but the makings of a different alliance (and a reformed, modernised EU) that’s perhaps the genesis of the elusive Middle Powers coalition. Brussels, for its part, still regards free movement, the jurisdiction of the European Court of Justice and dynamic alignment as non-negotiable, so the test of that flexibility has some tall asks.

What a Middle Powers Coalition Needs

If variable geometry is the operating model, four things would give it substance: a single market, regulatory coherence, a workable definition of sovereignty, and hard power backed by credible signalling.

1. A single market as leverage

A single market is perhaps the last lever available to a Middle Powers bloc that would have meaningful leverage with the US and China. Assuming technology and energy are firmly dominated by the US and China, Middle Powers will need a single market to exert leverage in other ways. This single market should extend to telecoms, energy, capital, AI and security – much of the territory Enrico Letta mapped out in his 2024 report on the future of the EU Single Market.¹¹

However, the national interest (the politics around jobs, investment, innovation and growth) inserts a protectionist and localisation agenda that undermines collective strategy. This tension (global vs local, protectionist vs open) isn’t new. Globalisation under the liberal international order, which we hear has largely unravelled in recent years, probably peaked a decade ago. Middle Powers will need to find that openness again and ring-fence it with a single market. Until then, the spectre of a GZERO world (Ian Bremmer’s term for a world where no power or group is willing to lead) seems likely to remain.

2. Regulatory coherence, not deregulation

Regulation is often cited as a reason that Middle Powers might struggle to be competitive in areas that the US and China dominate (a case Mario Draghi made forcefully in his 2024 report on European competitiveness).¹² The tension between deregulation and regulation keeps surfacing, and there seems to be a real lack of evidence about where to draw the line. There’s clearly considerable social and political pressure to ensure that regulation is appropriate to keep society and the environment safe, secure and sustainable – and the very topical debates about AI and online safety exemplify the point. At the same time, we hear that Europe is overregulated compared to the US and China – with the ability to build and commission data centres (energy connections, planning, environmental rules etc.) providing another topical example. In the UK, Ofgem has consulted on fees and readiness milestones to clear speculative data centre projects from the grid connection queue.¹³

As a thought experiment, would scrapping the EU GDPR, Digital Services Act, Digital Markets Act and anything else that would seem to create friction in the ability to innovate, start up, scale and operate make any noticeable difference in the short to medium term? Some in the EU say no; it’s anathema, and regulation is a convenient scapegoat for other more nuanced factors. The US and China have regulation too – and the US has its own fragmentation problem, with the White House moving in December 2025 to challenge a growing patchwork of state AI laws.¹⁴ One gets the sense that neither China nor the US would find it easy to advise the EU on precisely how their own regulatory regimes create less friction than EU regulation does. Suffice to say, regulation is a consideration, but by no means a silver bullet.

The issue is more likely the fragmented way in which EU regulation is implemented than the regulation itself. The IMF has estimated that barriers within Europe are equivalent to a tariff of around 44% on manufactured goods and 110% on services.¹⁵ A group of Middle Powers all reserving the right to adjust elements to accommodate the national interest seems to be the culprit. If Middle Powers can agree on a single market, then surely they can agree on common and consistent regulation too. The EU’s Digital Omnibus on AI, provisionally agreed in May 2026, which delays and simplifies parts of the AI Act rather than scrapping it, suggests a pragmatic direction of travel.¹⁶ Perfect shouldn’t be the enemy of good, but all too often the fragmentation risk persists, and part of the reason is possibly that politics isn’t doing enough to explain the why to societies as opposed to the what.

AI will be the next test. There are growing calls for treaties to manage AI proliferation and the pace of the AI frontier, responding to political pressure (stemming from societal concerns) about the safety of AI. Where this translates into more regulation, the arbitrage between low and high regulatory regimes reignites the trade-off between more or less regulation and how that incentivises or undermines competitiveness and investment. It would seem sensible for Middle Powers to have a common framework and approach, bearing down on over-regulation and fragmentation. With AI capabilities underpinning so much of the future economy, it’ll need to be pragmatic to avoid throttling the advantages AI can bring.

3. Sovereignty, dependence and investment

Another obstacle to the Middle Powers coalition that doesn’t appear to have any obvious solution is the heavy US and Chinese technology and industrial integration in Middle Power economies. This leverage, it seems, is always poised to countermand any attempts by individual Middle Powers or groups to hold the line in other areas, such as single market access and the ability to compete in parts of the technology stack and value chain.

Yet we hear that the pace of technology development is being measured in days whilst the pace of government is measured in years. Can Middle Power governments move swiftly enough to move onto the offensive and away from the defensive posture we seem to be trapped in? Arriving at a pragmatic and workable definition of sovereignty and sovereign capability that makes sense across this coalition is a start and should be high on the agenda. A smart and nuanced approach to the sovereignty question (as opposed to the binary protectionism and excessive regulation we know the Middle Powers must avoid) feels critical to unlocking the potential of a Middle Powers alliance.

We hear from the US and China that current European sovereignty requirements are dissuading investment and undermining European competitiveness. Investment from outside the Middle Powers alliance is important, but has to safeguard the ability of the Middle Powers to create and sustain choices (efficient, effective and competitive choices). Middle Powers will need to agree who gets the investment and jobs that US and Chinese investment brings, and reconcile the national interest so that the alliance isn’t a zero-sum system. The EU’s “Made in Europe” preference is a live test: Prime Minister Burnham has asked Brussels not to shut UK firms out of it, warning of “collateral damage” to UK industry.¹⁷

4. Hard power and signalling

In addition to a single market, Middle Powers must rebuild hard power. Russia is probing and testing EU and NATO defences and the appetite to respond. Some will be averse to hard power and find its development and exercise very alien. This is where sub-coalitions will need to form to lead and assure where others might not have the cultural or political space for hard power competition, for example the UK and French-led “coalition of the willing” for Ukraine and the UK-led Joint Expeditionary Force. The premise is that contributions come in various ways and places. This, it seems, is another possible obstacle, as the burden-sharing formula and narrative could be quite complex – especially given the lessons we’re taking from the US challenge to NATO and Europe on defence spending, which produced the 5% of GDP target agreed at The Hague in June 2025 and an immediate argument with Spain over whether it applied.

Something that Middle Powers will need to establish is a reliable mechanism for signalling to adversaries their capabilities and intent in such a way as to avert escalation and maintain the status quo. In the Cold War era, the elaborate systems of signalling between NATO and the Soviet Union (and Warsaw Pact) – from the 1963 Washington–Moscow hotline created after the Cuban Missile Crisis to the 1972 Incidents at Sea Agreement – kept either side from stepping over the brink, albeit there were a number of scares and close calls, Able Archer 83 among them.

That Cold War system of signalling and escalation management seems to be in a state of deep disrepair. New START, the last remaining US–Russia nuclear arms control treaty, expired in February 2026,¹⁸ and Russian drone and aircraft incursions into Polish and Estonian airspace in September 2025 showed how quickly ambiguous incidents can become crises. As Olga Oliker and Alexander Graef argue, the danger now lies as much in misreading incidents as in deliberate escalation.¹⁹ A world with a more complex group of Middle Powers will need something equally robust to accommodate hybrid threats that manifest in a greater variety of ways across the Middle Powers. This system is missing, and without it the hard-power element is only half complete.

The View from the G2: the US and China

It’s interesting to reflect on how China sees all of this. China has increasing leverage over Russia stemming from Russia’s war in Ukraine and its hybrid campaign against Europe. European Middle Powers are increasingly distracted and preoccupied with the war, and China is likely happy to see Europe and NATO contained in this way. Indeed, China’s foreign minister, Wang Yi, reportedly told the EU’s Kaja Kallas in July 2025 that Beijing did not want to see Russia lose, fearing the US would then turn its full attention to China.²⁰ So it doesn’t appear likely that China will take any significant steps to bring an end to a war which is increasing its leverage on Russia and diluting European and NATO capacity to compete. Hard power, robust signalling mechanisms, single market access and stronger regulatory alignment, exercised as a cohesive Middle Powers strategy, may change the calculus for China.

The US and China (the G2) see themselves on a separate (and higher) plane. A regulatory “deal” between them could constrain options and access for Middle Powers, as G2 regulation slows the ability of others to close the gap and effectively bakes in the G2 advantage. Whether the US and China can ever come to such a consensus is uncertain, but the first steps may already be visible. During Xi’s visit to Washington in late September 2026, the two sides agreed to set up a bilateral communication channel for AI incidents.²¹ Both will be looking for their own equilibrium, regulating enough to answer domestic concerns, but no more than their rival.

At the moment, with the US still fractionally ahead of China in AI (Stanford’s 2025 AI Index found the performance gap between leading US and Chinese models had narrowed to under two percentage points²²), the US has more incentive to reach a regulatory deal that in effect puts throttles on China’s ability to catch up and move ahead.

Conclusion: So What?

Gerard Baker may be right that a formal Middle Powers bloc is a non-starter, but he’s arguing against the wrong model. What’s emerging is variable geometry: SAFE opening to Canada, CPTPP talking to Brussels, coalitions of the willing on Ukraine. Whether it holds depends on Middle Powers treating cooperation as a repeated game rather than a one-off deal, and on governments explaining the why, not just the what. For business, it means planning for several overlapping rulebooks rather than one. You can’t wait for clarity, and you shouldn’t bet on a single bloc. Five practical steps follow:

1. Map your exposure by jurisdiction. Identify where your technology stack and supply chain depend on US or Chinese suppliers (cloud, AI models, semiconductors, network equipment, energy) and test what happens if that dependency becomes leverage.
2. Track which coalitions you’re eligible for. Market access and procurement increasingly depend on which coalitions your country belongs to. Build this into horizon scanning and bid strategy.
3. Turn sovereignty into practical procurement terms. Specify data location, operational control and exit or switching rights. The aim is to keep choices open, not to buy local by default.
4. Bring evidence to the regulation debate. Governments need evidence on where regulation and fragmentation create friction. In the UK, the forthcoming Regulating for Growth Bill²³ and government’s call for evidence on AI in the energy system (closing 6 November 2026)²⁴ are live opportunities to contribute, as is the EU’s Digital Omnibus for companies operating in both markets.
5. Treat geopolitical risk as an operational risk. As escalation management weakens, hybrid incidents (sabotage, hostile surveillance, cyber-attacks, economic coercion etc.) land first on businesses and critical national infrastructure. Use threat intelligence in business continuity planning and risk scoring.

---

References

1. LSEG, 2026 GZERO Summit in London, 8 October 2026. lseg.com
2. Gerard Baker, “Grin and bear it, Andy, and stick with Trump”, The Times, 17 September 2026. thetimes.com
3. Mark Carney, address to the World Economic Forum, Davos, 20 January 2026. aol.com
4. Stephen Hermanson, “Report from the Chatham House London Conference 2026: The Weight of Expectation on Middle Powers”, Geopolitics and Critical National Infrastructure (50years.blog), 12 July 2026. 50years.blog
5. See note 3.
6. Robert Axelrod, The Evolution of Cooperation (Basic Books, 1984).
7. Eunews, “Defence: EU countries give final go-ahead for Canada’s participation in SAFE funding”, 15 June 2026. eunews.it
8. Bloomberg, “Talks for UK to Join Flagship EU Defense Fund End in Failure”, 28 November 2025. bloomberg.com
9. GOV.UK, CPTPP–EU Joint Ministerial Statement, 27 March 2026. gov.uk
10. Euronews, “Andy Burnham is rethinking the UK’s ties with the EU. What are his options?”, 2 October 2026. euronews.com
11. Enrico Letta, Much More Than a Market: Speed, Security, Solidarity (April 2024). economie.fgov.be
12. Mario Draghi, The Future of European Competitiveness (September 2024), and the Commission’s one-year review. commission.europa.eu
13. Ofgem, Curate – Demand Connections Reform (consultation on proposed data centre connection reforms), 29 July 2026; responses closed 16 September 2026. ofgem.gov.uk
14. Executive Order 14365, “Ensuring a National Policy Framework for Artificial Intelligence”, 11 December 2025; summary by Wiley. wiley.law
15. International Monetary Fund, Regional Economic Outlook for Europe: A Recovery Short of Europe’s Full Potential (Washington, DC: IMF, October 2024). imf.org
16. Loyens & Loeff, “Digital omnibus on AI agreement: an overview” (provisional agreement, 7 May 2026). loyensloeff.com
17. See note 10.
18. Marion Messmer, “Avoiding a new nuclear arms race”, Chatham House research paper, 13 May 2026. chathamhouse.org
19. Olga Oliker and Alexander Graef, “A Three Body Problem: NATO, Russia and the Politics of Air Incursions”, Deep Cuts Issue Brief, 17 December 2025. deepcuts.org
20. Finbarr Bermingham, “China tells EU it does not want to see Russia lose its war in Ukraine: sources”, South China Morning Post, 4 July 2025. scmp.com
21. Al Jazeera, “China, US to open AI communication channel after summit, White House says”, 26 September 2026. aljazeera.com
22. Stanford HAI, AI Index Report 2025, Technical Performance. hai.stanford.edu
23. Pinsent Masons, “Sandboxes central to Regulating for Growth plans”, Out-Law, 14 May 2026 (Bill announced in the King’s Speech, 13 May 2026). pinsentmasons.com
24. Department for Energy Security and Net Zero, Vision for an AI-enabled clean energy system (call for evidence), 8 September 2026; closes 6 November 2026. gov.uk