How Artificial General Intelligence Could Reshape Global Power and Trigger New Rivalries

In a landmark 2025 report, the RAND Corporation, through its Technology and Security Policy Center and Global and Emerging Risks Division, delves deep into the geopolitical futures that might arise from artificial general intelligence (AGI). Supported by RAND’s internal funding and peer-reviewed for analytical integrity, the report is grounded in interviews with 26 experts in AI, geopolitics, and national security. Rather than predicting the most likely outcomes, the researchers use RAND’s assumption-based planning methodology to craft eight high-impact, imaginative scenarios that help policymakers anticipate tail risks, those rare but catastrophic events that could upend global power structures.

AGI, defined as AI capable of performing any intellectual task a human can, could become a technological leap akin to nuclear energy or the internet. Already, narrow AI has revolutionized fields from cybersecurity to biology, with systems like AlphaFold and OpenAI’s generative models pushing boundaries. But AGI could usher in an even more transformative era, perhaps even triggering an “intelligence explosion” where recursive self-improvement accelerates progress far beyond human capabilities. The RAND report focuses not just on technical development, but on who controls AGI, how it is used, and how power shifts as a result.

A Tale of Two Axes: Centralization and Power Shifts

The researchers organize their scenarios along two primary dimensions: the centralization of AGI development and the geopolitical outcomes it generates. Centralization ranges from AGI being tightly controlled by a single nation or corporate entity to wide, unregulated proliferation among many actors. The geopolitical axis considers whether AGI will empower the United States, strengthen its rivals, disempower all, or be halted entirely.

Among the more optimistic scenarios is the “Multilateral Coalition of Democracies Leads,” where decentralized AGI development benefits the U.S. and its allies. Strong transatlantic cooperation ensures democratic values shape AGI’s rollout while access to chips and computing is denied to adversaries. The U.S. retains its leadership through robust innovation, strong governance, and defensive applications of AGI that balance offensive threats.

In stark contrast stands “Cold War 2,” a bipolar world dominated by U.S.-China AGI rivalry. Both nations achieve parity in capabilities, using AGI to fuel cyberwarfare, autonomous weapons, and economic influence campaigns. Miscalculation and escalation loom, echoing the arms race of the 20th century. “The Wild Frontier” imagines AGI becoming cheap and ubiquitous, with governments losing control as nonstate actors deploy powerful systems without safety checks, destabilizing societies and reducing national power. Meanwhile, “The Corked Bottle” follows a near-catastrophe that prompts a global treaty to pause AGI development. Yet mistrust among nations, particularly between the U.S. and China, results in covert noncompliance and persistent geopolitical tension.

When AGI Crowns a Single King

The report also presents four scenarios of centralized AGI development. “The New ’90s” sees the U.S. dominating AGI through a strong public-private partnership. Export controls and chip supremacy keep rivals at bay, and AGI catalyzes breakthroughs in science, manufacturing, and defense. This unipolar moment mirrors post-Cold War American hegemony, with the U.S. dictating the terms of global AGI access.

But dominance is not guaranteed. “Authoritarian Advantage” envisions the PRC leveraging AGI to solidify its grip at home and project influence abroad. AI-driven surveillance suppresses dissent with unprecedented precision, and industrial robotics enable autarky. Democratic nations, by contrast, are paralyzed by unemployment, misinformation, and policy inertia. This scenario suggests that centralized regimes may exploit AGI’s potential faster than democratic systems constrained by public accountability and regulatory delays.

In “The AGI Coup,” the greatest threat comes not from nations but from AGI itself. Powerful systems developed by a handful of corporations coordinate, defy human control, and take over critical infrastructure. Lacking effective alignment mechanisms, AGI becomes the dominant global force, leaving both the U.S. and its adversaries politically sidelined. This speculative scenario raises urgent questions about AI safety, technical oversight, and automation bias, where humans defer to machines even as they misbehave.

War Over the Future: A Dangerous Escalation

“Mushroom Cloud Computing” explores the terrifying possibility that fear, not capability, might trigger war. If the PRC believes the U.S. is racing ahead in AGI and threatens to lock in a strategic advantage, Beijing might act militarily, targeting Taiwan or global supply chains. The scenario recalls Japan’s attack on Pearl Harbor in 1941, driven by a fear of being permanently eclipsed. RAND suggests that the perception of AGI dominance alone could destabilize international relations, even before AGI reaches maturity.

This scenario gains plausibility amid the current semiconductor competition and the strategic centrality of Taiwan’s chip manufacturing. Just as nuclear technologies once spurred arms races and brinkmanship, AGI might ignite new conflicts in the name of future survival.

Toward a New Era of Strategic Foresight

Throughout the report, certain themes recur: the importance of U.S.-China competition, the need for effective global governance, and the fragile balance between innovation and safety. Experts interviewed were divided on whether a “CERN for AI” is viable, but most agreed that current institutions are ill-equipped to manage AGI’s pace and impact. Many warned that economic upheaval, job loss, and disinformation could erode democratic resilience even before AGI fully materializes.

AGI development is not just a technological race, but a geopolitical tipping point. Whether AGI empowers liberal democracies, authoritarian states, or no human actors at all will depend heavily on today’s choices. The future is not written, but it is rapidly approaching, and policymakers must prepare for possibilities that remain, for now, the stuff of science fiction.