Sam Altman envisions AI as a basic utility, delivered on demand like electricity or water, with users paying based on usage.

The model hinges on on-demand access to computing power and intelligent services rather than fixed subscriptions, using a token-based system to measure consumption.

Pricing for AI services would be usage-based, with customers paying for how much they use.

He warns that without rapid infrastructure expansion, only well-funded organizations could access AI at scale, risking widening access gaps.

Altman notes AI adoption is accelerating across industries, enabling tasks to be completed in minutes instead of hours.

OpenAI has committed roughly $1.4 trillion on data-center projects over the next eight years to stay ahead in compute capacity, per Greg Brockman and Altman.

Major tech firms are expected to spend hundreds of billions on compute in 2026 to meet rising AI demand, despite ongoing infrastructure and power grid challenges.

The overarching goal is to move from a capacity-constrained model to scalable, on-demand access to AI services, backed by ongoing investment in hardware, data centers, and energy infrastructure.

The broader aim is to scale infrastructure so AI becomes a daily, widely accessible resource, not limited by computing capacity.

Altman discussed at the BlackRock Infrastructure Summit that AI services will become an on-demand norm across industries and power daily tools and assistants.

The US Infrastructure Summit remarks framed AI as a daily necessity, integrated into work tools and personal applications.

This shift reflects a broader industry move from novelty software to a fundamental utility for everyday use.