The AI trillion-dollar question spooking markets: Will hyperscalers justify the biggest spending spree in history?

Goldman Sachs estimates that Amazon, Microsoft, Alphabet and Meta alone will invest around US$5.3 trillion in AI infrastructure between 2025 and 2030. That figure includes data centres, GPUs, networking, power infrastructure and the enormous ecosystems needed to support them.

The obvious question for investors is:

How much revenue must this infrastructure ultimately generate to justify that investment?

A simple back-of-the-envelope calculation provides a useful reality check.

Assume:

• US$5.3 trillion of cumulative investment.
  
• An average asset life of 8 years.
  
• Investors require approximately a 12% annual return on capital.
  
• 40% cash margin
  

Under those assumptions, which are, of course, open to debate, the AI ecosystem would ultimately need to generate approximately US$1 trillion of annual operating cash flow. At a 40% operating cash flow margin, that implies around US$2.7 trillion of incremental annual revenue, equivalent to roughly 2–3% of global GDP or the entire GDP of the UK or India.

To put that figure in perspective, the global software industry generates around US$800 billion to US$1 trillion of annual revenue, global advertising around US$1 trillion, and global pharmaceuticals approximately US$1.7 trillion.

If AI ultimately transforms professional services, software, healthcare, manufacturing, financial services and transportation, it is conceivable that it could become one of the largest industries in the world.

But the bear case is equally compelling.

If enterprise adoption proves slower than expected, or businesses conclude that many AI applications fail to generate an adequate return on investment, these spending assumptions become much harder to sustain. Large enterprises may spend heavily, but smaller businesses could simply rely on low-cost AI features bundled into existing software subscriptions.

In that world, the industry could find itself with enormous computing capacity chasing insufficient demand. Prices would come under pressure, returns on invested capital would disappoint, and today's record-breaking capital expenditure could become tomorrow's case study in capital misallocation.

Worse still, AI may prove extraordinarily valuable for society while capturing relatively little economic rent for the companies providing it. If competition drives AI models towards commodity pricing, much of the productivity gain may accrue to customers rather than shareholders. That's exactly what happened in airlines, PCs and many internet businesses.

Framing the debate this way shifts the focus from the technology itself to the economics. The question is no longer whether AI will change the world. It's whether the world's businesses will ultimately spend an extra ~US$2.5 trillion every year using it.

So how should investors think about the odds? While nobody knows how the AI story will unfold, we do know something about history.

One of the most reliable lessons from the past 30 years is what has happened to companies that dramatically increase capital expenditure.

We show in the chart below the stock price performance of the 2.5% of companies with the highest capex growth each year using a global investment universe of around 10,000 companies per year.

The result is remarkable.

A hypothetical US$1 million invested in 1997 and rebalanced each year into the 2.5% of companies with the fastest capex growth would have shrunk to just US$87,000 today, even before allowing for inflation.