Artificial intelligence is increasingly treated not merely as a tech sector, but as a strategic capability linked to economic power, national security, and geopolitical competition. Recent warnings from leading AI researchers and executives that increasingly autonomous systems could pose extreme risks have intensified the debate over whether technology development should be slowed down.
However, these same warnings may make a slowdown even less likely. If advanced AI can provide significant military, economic, and technological advantages, governments may view its continued development as a strategic necessity. The result could be a paradox in which rising concerns about AI risks encourage countries to accelerate the race to control it. The United States and China are at the epicenter of this competition. Both governments have strong incentives to maintain their technological edge, at a time when companies are investing vast amounts in data centers, microchips, and computing infrastructure.
The scale of investment illustrates why reversing course would be difficult. Goldman Sachs estimates that cumulative global AI spending by the end of the decade could reach as much as 5% of global GDP. At the same time, financial markets are increasingly sensitive to whether huge investments in AI infrastructure will ultimately yield adequate economic returns.
Why it matters
The central question is no longer simply whether AI development should be regulated. It is whether governments can realistically agree to meaningful constraints while simultaneously competing for technological dominance. A country that unilaterally slows down development risks allowing its rivals to pull ahead. This creates a classic security dilemma: measures intended to reduce risk can themselves create strategic vulnerability. The problem is particularly significant because AI has applications far beyond commercial technology. Advanced systems could impact cybersecurity, intelligence services, military planning, financial markets, biotechnology, and critical infrastructure. This makes AI distinct from many previous technology debates. Governments may be willing to regulate specific applications, but they are less likely to accept restrictions that could permanently diminish their strategic advantage.
The investment dilemma
The AI boom is already deeply connected to broader economic growth. Companies are spending massive sums on computing power and data centers, while investors have awarded astronomical valuations to firms expected to benefit from AI adoption. A major slowdown could therefore have consequences far beyond tech companies themselves. Morgan Stanley has estimated that nearly $200 billion in US data center projects were canceled or delayed due to political pushback against new infrastructure. The bank has also argued that capital expenditure tied to AI could account for roughly one-third of US GDP growth during the current period. This creates yet another barrier to limiting growth. AI investments are no longer just corporate decisions. They have become linked to employment, infrastructure, financial markets, and national economic growth. The longer the investment cycle continues, the harder it becomes for governments and companies to voluntarily walk away from it.
Geopolitical dimension
The geopolitical dimension makes coordinated restraint even more difficult. American policymakers increasingly view technological leadership in AI as an element of national power. China, for its part, sees AI development as part of its broader effort to reduce dependence on Western technology and strengthen its own economic and strategic capabilities.
This means Washington and Beijing have every incentive to monitor each other's progress rather than simply enact limits. Political rhetoric surrounding AI increasingly reflects this competition. Trump has framed opposition to AI infrastructure as something that benefits China, while Chinese state media have described AI warnings as part of a broader geopolitical strategy against Beijing. The result is a familiar strategic dilemma: if one side believes the other might continue developing advanced AI, neither has much incentive to slow down first.
Can international rules work?
Meaningful constraints on AI would ultimately require international cooperation, Reuters notes. Individual countries can establish national safety standards, require companies to perform risk assessments, or impose limits on specific applications. However, these measures cannot fully address the geopolitical problem if competitors outside those jurisdictions remain free to develop stronger systems.
The comparison to nuclear weapons illustrates the difficulty. The first nuclear weapon was developed in 1945, but the Non-Proliferation Treaty did not emerge until 1968. Even after the treaty, nuclear proliferation continued beyond the five officially recognized nuclear powers. AI could present an even more complex challenge because it is not confined to a small number of military facilities. Its development involves thousands of companies, researchers, data centers, and compute systems spread across many countries.
Economic security could become the new arms race
The most significant shift may be that competition in AI does not need to resemble a conventional military arms race to generate similar strategic pressures. Advanced AI could yield advantages in productivity, financial markets, scientific research, cybersecurity, and military technology. Falling behind could therefore be perceived as a threat to the economy and national security, even without the development of a specific weapon. This could lead to what might be termed "mutually beneficial economic dependence."
Countries may recognize the risks of uncontrolled AI while simultaneously believing that abandoning the technology would be even more dangerous. This dynamic makes voluntary restraint increasingly difficult. Growing warnings about AI may therefore have an unintended geopolitical consequence: accelerating the race they are supposedly trying to limit. If governments become convinced that highly capable AI could fundamentally alter national power, they are unlikely to leave its development solely to private companies or accept limits that allow competitors to gain an advantage. Instead, the technology could become even more tightly integrated into national security policy.
This could ultimately mean greater government involvement in AI companies, computing infrastructure, semiconductor supply chains, and access to advanced models. The greatest danger, therefore, is not simply that AI will become technologically too powerful to control. It may become too strategically important for governments to allow it to slow down. For meaningful safeguards to emerge, major powers will need to treat AI safety as a shared security interest rather than another arena of geopolitical competition. Without this shift, every warning about the potential consequences of advanced AI could strengthen the argument for acceleration. The paradox is clear: the more dangerous AI becomes, the more governments may feel compelled to control it. And the more they compete to control it, the harder it becomes to stop the race.
www.bankingnews.gr
Readers’ Comments