The Economics of Artificial Intelligence

The US economy is experiencing something unprecedented: AI infrastructure spending is now contributing as much to US GDP growth as consumer spending. For Australian investors, this is enormously significant. The investment phase must eventually give way to an impact phase, and the timeline will shape global markets for the next decade.

A Historic Shift in GDP Composition

AI infrastructure spending is now contributing as much to US GDP growth as consumer spending. This is remarkable, given consumer spending represents roughly 70% of the American economy while AI infrastructure is a tiny fraction by comparison.

The scale of investment is staggering. The top 11 cloud providers are forecast to deploy $392 billion in 2025 alone, more than the previous two years combined. Just four companies (Amazon, Microsoft, Alphabet, and Meta) account for $315 billion of this AI infrastructure spending. Add Oracle’s massive cloud deals and Tesla’s AI infrastructure buildout, and the capital deployment becomes so enormous it’s reshaping the composition of US economic growth itself.

Investment Phase, Not Impact Phase

The critical point is we’re measuring the investment, not the impact. Think of AI’s economic contribution today as construction spending. It boosts GDP through building activity, equipment purchases, and employment. But construction spending isn’t the same as the economic value of the completed building.

Right now, AI is contributing to growth through capital deployment. The actual productivity gains that justify this spending remain largely ahead of us. Goldman Sachs research found no discernible impact yet on aggregate productivity measures. Enterprise AI initiatives currently achieve just 5.9% ROI against 10% capital investment. The gap between spending and economic returns is wide.

We’ve seen this pattern before. During the internet boom, companies invested over $500 billion in just five years, yet IT spending initially created a “productivity J-curve.” The investments made in the 1980s didn’t translate into economy-wide gains until the late 1990s — more than a decade later.

Risks if the Transition Stalls

Several factors could delay or diminish the transition from investment to impact:

Power constraints:

Even if all planned energy plants are delivered on time, the US could face a capacity deficit exceeding 15 gigawatts by 2030.

Slow adoption:

While nearly all companies are investing in AI, only 1% call themselves “mature” in deployment.

Labour market displacement:

Entry-level workers in AI-exposed occupations experienced 6% employment declines. By 2030, up to 30% of hours worked could be automated, requiring 12 million occupational transitions. If displaced workers face extended unemployment, consumption could weaken precisely when the economy is becoming more reliant on AI investment to maintain growth.

The Long-Term Prize

Yet if the transition succeeds, the economic transformation extends well beyond productivity statistics. Projections suggest AI will increase productivity by 1.5% by 2035 and nearly 3% by 2055, with 170 million new jobs created by 2030 alongside displacement.

More profoundly, AI promises to drive down costs and improve access to essential goods and services. In pharmaceuticals, AI could generate $60 billion to $110 billion annually in economic value by accelerating drug discovery, making life-saving medications cheaper and more accessible. These cost reductions could extend across healthcare, education, legal services, and professional services, effectively lowering inflation in sectors that have historically outpaced general price growth.

The result would be improved living standards through both higher productivity and greater accessibility to essential services, benefits that traditional GDP measures struggle to fully capture. The investment phase is real and quantifiable today. The impact phase, with its productivity gains, cost deflation, and improved access to critical services, remains a forecast extending into the 2030s. For global investors, successfully navigating this transition will define portfolio positioning for the next decade.