Dr. Muhammad Akram Zaheer

Everyone seems to have a theory about artificial intelligence. Some imagine a future dominated by super intelligent machines, capable of transforming societies in ways unprecedented in history. Others envision a more incremental trajectory, where AI enhances productivity, accelerates scientific discovery and gradually reshapes economic and military landscapes without a sudden leap. The divergence in these expectations underscores the difficulty policymakers face: how does one prepare for a technology whose pace, impact and diffusion remain uncertain?

Central to any discussion of AI strategy are assumptions about the replicability of breakthroughs. Some analysts expect rapid imitation: rivals could soon catch up through reverse engineering, access to leaked models, or innovative use of existing infrastructure. Others argue that the complexity of AI, particularly its reliance on proprietary hardware, vast datasets and specialised expertise, creates durable advantages for those who reach the frontier first. Similar disagreements surround China’s ambitions. Some contend Beijing is determined to surpass the United States in frontier innovation, while others suggest it prioritises deploying existing technologies and adapting American breakthroughs once they appear.

The stakes are high. Strategies built on incorrect assumptions risk wasting resources and losing technological leadership. For Washington, the challenge is not predicting a single future but building resilience across multiple possibilities. AI strategy must ensure that national security, economic competitiveness and democratic values are strengthened, even as the shape of the AI era remains uncertain. A useful framework for understanding the challenge involves three central questions: first, will AI progress accelerate toward superintelligence, or will it plateau in uneven and bounded ways? Second, will breakthroughs be easy to replicate, or will catching up remain difficult and costly? Third, is China racing aggressively toward frontier AI, or is it prioritising diffusion and adaptation? Each question has two plausible answers, yielding a three-dimensional matrix with eight potential scenarios, each with distinct strategic implications.

The first dimension AI progress spans a spectrum from superintelligence to bounded and jagged advancement. Superintelligent AI, capable of recursive self-improvement, would be a transformative force with far-reaching consequences for economics, science and military power. Even a narrow lead could prove decisive, justifying substantial investments in frontier research. In contrast, bounded and uneven AI delivers significant capabilities in areas such as mathematics, coding and logistics, yet remains constrained in creativity, judgment and physical application. In such a world, prioritising broad deployment and diffusion may be more effective than concentrating resources on a few moonshot projects. The second axis considers the ease of replication. In a world where catching up is straightforward, technological advantages are fleeting and strategy centres on rapid adoption and embedding systems across allied networks. Conversely, if replication is difficult, the focus shifts to safeguarding critical infrastructure, datasets and human capital. Here, the first mover retains a lasting edge and the ability to compound advances over time becomes a central concern. Strategy must then balance diffusion with the protection of core capabilities.

The third axis focuses on China’s intentions. At one extreme, Beijing races aggressively toward frontier innovation, funding major training runs and competing laboratories. At the other, it emphasises adoption and diffusion, occasionally producing frontier-scale models to signal capability but relying on replication and adaptation of foreign breakthroughs. While China’s internal strategy may vary across institutions, its overall behaviour determines whether Washington must prepare for direct competition at the frontier or a focus on global diffusion. Across these dimensions, eight worlds emerge, each demanding tailored policy responses. In the most extreme scenario—superintelligence, difficult replication and an aggressive China—the United States could face a contest akin to a technological arms race. Here, securing frontier breakthroughs becomes a matter of national survival. Policymakers may need to mobilise extraordinary resources, coordinate government and private sector efforts and strengthen security across the semiconductor supply chain, training infrastructure and model data. International risk management would be crucial, including potential restraint agreements with China to avoid unintended escalation and maintain human oversight of autonomous systems.

In a variation where superintelligence is achievable and replication remains difficult, but China does not race aggressively, the United States might enjoy a unipolar moment. The challenge here is less about competition and more about the responsible use of transformative capabilities. Securing critical infrastructure and talent, safeguarding intellectual property and pacing development to ensure safety would be priorities. Diffusion of technology would be selective, extending access to trusted allies and partners, while domestic policy could focus on harnessing productivity gains for broad-based prosperity and social resilience.

If breakthroughs are easily copied, the race shifts from frontier research to global diffusion. Here, the advantage lies in who deploys systems fastest, shapes standards and integrates technology into domestic and allied industries. In such a world, export controls and secrecy have limited effect; strategy must emphasise deployment, adoption and embedding systems aligned with democratic values. Cooperation with allies becomes indispensable, not only to extend technological influence but also to manage security risks arising from rapid proliferation.

Even if superintelligence remains out of reach, frontier competition remains significant. A scenario where replication is hard and China races toward the frontier presents a protracted innovation contest. The United States must sustain research and development, invest in industrial capacity and ensure early deployment of capable systems. Diffusion remains important, but the emphasis shifts toward building a durable advantage over time, rather than achieving immediate global dominance.

Across all scenarios, certain constants emerge. Computing power underpins AI capability. Control over chips, data centres and energy resources determines who can train and deploy advanced systems. Robotics and advanced manufacturing translate digital intelligence into tangible economic and military capacity. A strong industrial-scientific base, a robust talent pool and resilient infrastructure remain essential. Risk management—often seen as a constraint—is equally a source of stability, ensuring that technological progress does not outpace safety, oversight, or societal readiness.

Policymakers also shape the trajectory of AI indirectly. Through incentives, regulations, research funding, export controls and strategic signalling, Washington can influence where investment flows, which capabilities mature and how systems diffuse globally. Engaging with allies to offer trusted alternatives to Chinese technology and embedding safe and open systems abroad strengthens both security and global influence. Yet the private sector remains central and its priorities do not always align with national objectives. Managing this tension balancing frontier innovation, deployment and risk is among the United States’ greatest challenges.

The eight-world framework serves not to forecast a single outcome but to discipline thinking, expose hidden assumptions and test strategies against a range of possibilities. Policymakers must select a base case, hedge against adverse outcomes and identify policies resilient across multiple scenarios. Intelligence and national security agencies should monitor signals along each axis, updating strategic assumptions as evidence accumulates. This probabilistic approach allows for flexibility, adaptability and informed decision-making in a rapidly evolving technological landscape. Finally, this framework is instructive beyond the government. Debate over AI often degenerates into advocacy for opposing visions. By clarifying assumptions whether about the pace of progress, replicability of breakthroughs, or China’s intentions participants can engage more productively, distinguishing between differences in assumptions and differences in strategy.

Geopolitics in the AI era will not be simple. Yet, without a disciplined framework, hidden assumptions and untested narratives risk guiding policy, squandering advantage and undermining security. Mapping possible worlds and the decisions they entail enables both policymakers and citizens to navigate uncertainty. The task for the United States is clear: treat AI not as a single story but as a shifting landscape. Success will go to those who understand the multiplicity of futures and act strategically, rather than reacting to the illusion of certainty.

In an era defined by uncertainty, clarity comes not from predicting a single outcome but from preparing for all plausible ones. The AI age is already upon us and the choices made today will define the balance of technological power, economic influence and national security for decades to come.