Every few months, Big Tech steps onto the earnings stage and drops another eye-watering number on the table. Tens of billions of dollars earmarked for AI data centres, chips, servers, and everything in between. And each time, the market winces a little. You can almost hear investors whispering: “This better pay off…”
It’s a fair reaction. The numbers have become staggering.
Based on the latest guidance from the Big Four hyperscalers – Microsoft, Meta, Amazon, and Alphabet – AI spending in 2025 is set to hit around $380 billion. By 2026, that balloon is expected to expand past $500 billion.
It’s a tidal wave of capital flooding through the entire AI supply chain.
GPUs. Memory. Networking. Cooling. New data centres. Upgrades to old data centres. Software layers stacked on software layers.
Everything is scaling at once, and at a pace the industry has never witnessed.
So, the natural question everyone is asking is simple: Is it worth it?
Wall Street isn’t fully convinced…
Critics argue that AI isn’t delivering ROI quickly enough to justify these massive capex budgets.
They say AI chatbots are cute, enterprise pilots are slow, and revenue from AI tools still isn’t moving the needle in a big way.
The anxiety feels oddly familiar. It’s reminiscent of the late-90s dot-com era, when companies poured billions into servers, networks, and fibre-optic cables long before the profits showed up.
Nevertheless, something interesting recently happened. And it didn’t come from a tech CEO, a consulting firm, or an AI startup.
It came from the St. Louis Federal Reserve.
The Fed looked at real-world data across industries, across job types and asked a simple question:
What happens to productivity when workers start using AI?
The answer was stunning.
Across the board, industries where workers save time using AI see productivity accelerate. Not marginally. Dramatically. The Fed found that for every 1% of time saved with AI, productivity in that industry grew 2.7% faster than before the pandemic.
To put that into perspective: most mature economies fight for 1–2% productivity growth a year. They plan policy around it. They set expectations around it. They celebrate when the number hits 2%.
Now imagine companies save just 5% of their working hours using AI. That alone, according to the Fed’s data, would drive a 13.5% jump in productivity. Add the usual 1.5% annual productivity growth on top, and suddenly the economy is staring at a 15% productivity surge.
Here’s why that matters:
GDP growth is basically hours worked + productivity. If hours stay flat, and productivity jumps 15%… then GDP jumps about 15% too.
Global GDP sits around $110 trillion. A 15% boost would add $16.5 trillion in economic output.
Now, let’s be realistic. Not every industry or country will hit these numbers at the same speed. A more modest scenario of a 5–7% global productivity boost still adds $5–8 trillion to the world economy. That’s equivalent to creating another Japan or Germany out of thin air.
And all of that is driven by one thing: AI helping workers get more done in less time.
So, when investors look at Big Tech and ask whether spending half a trillion dollars next year on AI infrastructure is “worth it,” they’re asking the wrong question.
The real question is: What happens if they don’t spend it?
For Big Tech, AI isn’t a nice-to-have
It’s becoming the new baseline for competitiveness, efficiency, and economic growth. Companies that deploy it will run faster, operate leaner, and scale further than those that ignore it. Countries that embrace it will grow, and those that fall behind will stay behind.
That’s why hyperscalers are spending. Not for vanity. Not for headlines. But because the ROI is showing up in the data.
And if the Fed’s numbers are even directionally right, then one could argue this era of AI investment isn’t a bubble – but rather the foundation of the next global growth cycle.
Not a subscriber to Money Morning?
You can get free daily recommendations like these with Money Morning eletter. Just sign up here.
