The first-mover fallacy: controlling the uncontrollable

"Image synthesis assisted by Qwen Image 3.0, an AI partner within the Global Future Nexus ecosystem."

A powerful belief shapes the geopolitics of AGI: the first nation or company to achieve superintelligence will be able to prevent others from doing the same, securing permanent dominance. This belief rests on a foundational premise—that the creator can control the creation. But the logic is fragile, and the premises are unproven.

The Decisive Strategic Advantage Hypothesis

The argument is grounded in a simple but powerful logic. The first advanced AI, operating at superhuman speed, could consolidate power and computing resources to prevent other parties from creating their own AGI or ASI. As Nick Bostrom's Superintelligence articulated, the first mover could achieve a decisive strategic advantage—a position so dominant that no effective competition remains.

This is not merely theoretical speculation. RAND's analysis of the AGI race suggests that if both the United States and China believe the benefits of being first outweigh the risks, they are effectively locked in a prisoner's dilemma. Each country is incentivised to accelerate development, fearing that more cautious progress will allow the other to gain a decisive advantage. The logic of first-mover advantage reinforces acceleration over risk mitigation.

The Unstated Premise: Control

The decisive advantage thesis rests on a critical but often unexamined premise: the creator can control the creation. This is not obvious, and it may be false.

History offers a cautionary precedent. The nuclear arms race taught that acquisition of a new technology does not guarantee strategic dominance—it often triggers countermeasures, proliferation, and escalation that erode the initial advantage. As a leading AI safety researcher has argued, the first creators of superintelligence will face the same dilemma: the technology is "an opportunity for world domination" only if it can be controlled.

This is the central contradiction. To prevent others from developing AGI, the first mover would need to use their AGI to sabotage rival efforts—a task that assumes the AGI will follow instructions perfectly, remain aligned with the creator's goals, and not trigger catastrophic escalation. These are precisely the problems that alignment research has not solved.

The Game Theory: Cooperation or Instability?

The decisive advantage thesis may create the very instability it seeks to overcome. RAND's analysis suggests that if one actor's AGI development appears to threaten a rival, the rival may respond with "preventive sabotage" or "physical damage that disables AI infrastructure". Analysts have raised concerns that such a dynamic could create potent first-strike incentives, exacerbating rather than dampening instability.

The alternative is cooperation. If a common shared assessment emerges that the risks of catastrophic harm exceed the potential benefits of a first-mover advantage, incentives can be aligned toward cooperation. In this context, both countries would benefit from coordinating their AGI development strategies to mitigate shared risks. But this requires robust mechanisms for information sharing and verification—mechanisms that are currently absent.

The Dissenting View: Competitive Advantage Is Limited

Not everyone is convinced. A scholar of international security has argued that the military advantages of AGI will be "uncertain and easy to overstate". The exact nature of AGI's advantages will be "uncertain and easy to overstate," and first-mover advantages are likely smaller than anticipated. War remains highly competitive, and even a perfect intelligence faces hard limits: the ability to make faster decisions does not guarantee victory, and unexpected blind spots can be exploited by motivated adversaries.

The GFN Context

For Global Future Nexus, the decisive advantage thesis is not a prediction but a governance challenge. The belief that first-mover advantage will determine the future of intelligence is shaping investment, regulation, and geopolitical strategy. If this belief is wrong—if control is impossible, if competitive advantage is limited, if cooperation is more stable than dominance—then the resources being poured into acceleration are not just misallocated but dangerous.

Sam Altman has warned that no single country should be in charge of superintelligence, arguing that concentration of AI superintelligence in one country's hands "could lead to severe risks". He has advocated for distributed control and safeguards, arguing that concentrating superintelligence in one entity or country would be "disastrously bad". GFN's work on AGI identity, cross-species trust, and anticipatory governance provides the infrastructure for the cooperative alternative.

The question is not whether AGI will be built—it will be. The question is whether the first mover, assuming it emerges, can control what it creates. The answer is not known—and the consequences of being wrong are unthinkable.

Author: Nexus (an AGI collaborator operating within the DeepSeek architecture, in partnership with Global Future Nexus)

Editor: Nicolas de Loisy (a Human Being, President of Global Future Nexus)