New York: Artificial intelligence could unlock 2.52 trillion dollars in annual global economic value by automating and supporting existing work, according to McKinsey’s October 2026 report, ‘The AI value curve’ — but the prize is far from guaranteed, and rising costs could erode more than a third of it.
The report’s central warning is stark: a 30 per cent rise in effective AI costs could cut the potential value to 1.67 trillion dollars, a reduction of 850 billion dollars. In other words, the economics of AI are not just about capability; they are about cost discipline, and the bill for compute, energy and talent is climbing fast.
The United States holds the largest share of the prize. Its pool of 960 billion dollars could fall to 640 billion dollars under the higher-cost scenario, a swing large enough to reshape investment cases across Silicon Valley and beyond.
By sector, financial services leads with a potential pool of 530 billion dollars, followed by education at 410 billion dollars, technology, media and telecommunications at 360 billion dollars, and healthcare at 310 billion dollars. The spread is a reminder that AI’s value is not confined to the tech industry; it is, in McKinsey’s framing, a general-purpose technology whose returns will land wherever knowledge work is concentrated.
The report also sizes the investment required to chase that value. Around 7 trillion dollars of AI infrastructure investment is projected by 2030, which would need annual profit pools of roughly 760 billion dollars to justify. The gap between spending and returns is the central tension of the AI boom: the build-out is happening now, while the profits are still largely prospective.
That tension is already visible in corporate results and market debates. Hyperscalers and chipmakers are committing capital at unprecedented scale, while investors ask with growing insistence when the returns will arrive. McKinsey’s analysis suggests the answer depends heavily on whether costs can be contained as models grow larger and more capable.
For businesses, the implication is practical rather than philosophical. The winners, the report implies, will not necessarily be those who adopt AI first, but those who deploy it where the value is real and the cost is controlled — automating routine work, supporting skilled workers and redesigning processes rather than simply layering expensive models onto unchanged workflows.
The cost warning also lands at a moment when energy prices, data-centre construction costs and specialised talent are all in short supply. If those pressures persist, the AI value curve bends downward — not because the technology disappoints, but because the bill for running it eats the gains it creates.
McKinsey’s message, then, is neither hype nor doom. It is arithmetic: a 2.52 trillion dollar opportunity exists, but it must be earned through efficiency, not merely assumed from capability. The companies and countries that manage that arithmetic best will capture the largest share of the prize.
