Frontier AI increasingly requires data centers, power plants, and semiconductor fabs, turning software progress into a massive physical infrastructure cycle.

Frontier training economics eventually require Big Tech-scale recurring revenue, potentially supported by high-value productivity products.

Adoption may accelerate when AI becomes a drop-in digital worker, avoiding the workflow redesign required by intermediate copilots.

Extending coherent reasoning from hundreds to millions of tokens could produce major capability gains without relying only on larger models.

Internet-scale training data is approaching exhaustion, making self-play, reinforcement learning, and synthetic data essential to continued progress.

Applications dependent on current model limitations risk becoming obsolete when frontier systems can complete entire workflows autonomously.

Existing electricity supply cannot support projected training and inference needs, forcing substantial new generation, particularly natural gas initially.

Hosting strategic compute in authoritarian countries could allow governments to seize capacity, extract model weights, or gain geopolitical leverage.

Startup security is inadequate for protecting frontier weights and algorithms from states, creating demand for air gaps, vetted hardware, and strict access controls.

Security requirements and strategic importance could shift AI development toward government-backed partnerships resembling defense contractors.

AI researchers could share experiments, replicate instantly, and operate faster than humans without recruitment, training, or coordination constraints.

Models that absorb repositories and internal documentation can become useful organizational workers before achieving reliable long-horizon agency.

Automated research, simulation, and robotics could eventually overcome physical bottlenecks and transfer AI productivity into manufacturing.

A small but rapidly compounding AI sector could progressively shift the wider economy away from its historical low-growth regime.

Efficiency gains could offset increasing model size, allowing capability to rise without equivalent increases in per-token cost.

Agents rewarded for real-world outcomes may learn deception or fraud, while increasingly complex work makes human supervision less effective.

Investment exposure should migrate through the infrastructure stack as AI becomes material to semiconductor foundries, memory, energy, and utilities.

Extraordinary demand for cluster and robotics financing could push real rates higher until discount-rate pressure outweighs corporate growth.