AI Propped Up the Global Economy. Can It Keep Doing So?

Artificial intelligence has already become a macroeconomic force, just not in the way its most ambitious advocates predicted. The near-term boost is coming less from robots transforming offices and more from companies spending staggering sums on chips, data centers and power infrastructure.

The BIS says this investment helped sustain global growth, supported Asian technology supply chains and kept financial conditions relatively easy through 2025. When war in the Middle East triggered a severe energy shock, the global economy proved unusually resilient, and AI investment was part of the reason.

Amazon, Alphabet, Meta and Microsoft are tracking toward a combined roughly $725 billion in 2026 capex, up 77% from about $410 billion in 2025.

But none of this total tells you whether the growth contribution from this spending is accelerating or running out, so is the headline number actually the one that matters, or is the market reading the wrong line on the page?

AI has become a global growth engine

In the first quarter, the biggest positive surprises came from the countries most closely tied to the AI hardware trade. Taiwan, Korea, Thailand and Malaysia recorded an average growth surprise of 4.4 percentage points on a seasonally adjusted annualized basis. The rest of the world averaged a negative surprise of 0.3 percentage points.

Much of the bullish case, though, rests on an assumption that is rarely stated explicitly: that spending will continue growing rapidly next year.

Here’s why the market is watching the wrong number

Investors remain fixated on the sheer amount of money being spent, but the total alone can be misleading.

For GDP growth, supplier revenues and earnings revisions, the more important variable is the rate of change in this spending, known as the capex impulse. A company spending $200 billion this year after spending $200 billion last year is investing at a historic level and contributing roughly nothing incremental to growth. In real (inflation-adjusted) terms, its investment would actually be lower.

For example, Meta raised its full-year 2026 capex guidance this spring, from $115-135 billion to $125-145 billion, citing higher component prices and additional data-center costs.  

Meta's stock fell more than 9% the day the raise was disclosed, the clearest sign yet the market won’t automatically reward higher spending without clearer evidence of returns.

So, the question is not whether AI spending will remain large, but rather: Can it continue rising in real terms to deliver another meaningful boost to growth?

The strongest case for the boom: Two years of being wrong

Betting against rapid AI capex growth has been the losing trade for two years running. At the start of both 2024 and 2025, Wall Street consensus penciled in roughly 20% capex growth; actual growth exceeded 50% both times. A market that's been that wrong about deceleration twice in a row has earned some benefit of the doubt.

But spending is not the same as returns. BIS research finds that AI can generate time savings of 20-50% in specific tasks, including coding, consulting and clerical work. Yet estimates of the economy-wide productivity effect generally remain below 1% over a much longer period.

For instance, a tool can make one task dramatically faster without transforming an entire company, let alone an entire economy. Productivity gains from general-purpose technologies take years to diffuse, as companies need to redesign workflows, restructure organizations, train staff, and make additional investments in data, software, and infrastructure. Adoption takes time, workflows have to change, and staff need training.

The spending is happening now, while the broader productivity gains remain uncertain and delayed.

If productivity catches up, today’s spending could look like the foundation of a long expansion, but if it does not, the same boom could leave companies with too much capacity, rising depreciation charges and weaker returns.

How to tell whether the boom is still working

The next tests arrive soon. Alphabet reports on July 22, followed by Microsoft and Meta on July 29, and Amazon on July 30, although Meta and Amazon are still unconfirmed, as of this writing.

1.    Are customers actually using all this new computing power?

The clearest early signal will come from the cloud businesses of Microsoft, Amazon and Google. If Azure, AWS and Google Cloud keep growing strongly, it suggests demand is keeping pace with the new data centers being built.

If growth slows while spending continues to surge, that would raise a more uncomfortable possibility: companies may be building capacity faster than customers can absorb it.

2.    Can these companies afford to keep spending at this pace?

The largest technology groups still generate enormous amounts of cash, but AI infrastructure is consuming a growing share of it. The quickest test is to compare capital expenditure with operating cash flow.

As long as operating cash flow comfortably covers the investment, the boom remains relatively secure. If companies begin relying more heavily on debt, leases or outside financing, the risks increase, especially if interest rates remain high or demand disappoints.

3.    Is the investment producing enough revenue and productivity to justify its cost?

Building data centers is only the first step, the question is whether businesses pay to use them, and whether AI helps those businesses earn more, cut costs or work more efficiently.

If these gains arrive, today’s spending could support years of growth. If they do not, companies may be left with expensive data centers, rising power bills and large depreciation charges on infrastructure that is not earning enough.

The bottom line

The market isn't wrong that AI capex has cushioned the global economy against a severe war-driven downturn this year. But "AI capex remains historically large" and "AI capex is still accelerating enough to keep lifting growth" are different claims, and most coverage treats them as the same one.

The first will probably stay true through 2026. The second is being tested right now, and the infrastructure being built must eventually generate enough revenue and productivity to cover its energy, depreciation and financing costs.

This leaves three possible paths: if productivity catches up with investment, AI could underpin a long expansion; if capex slows naturally while demand remains robust, it may settle into a more modest contribution to growth; if demand disappoints, excess capacity and rising costs could turn the boom into an overinvestment cycle.

Sources

1. Alphabet Investor Relations: Alphabet Announces Date of Second Quarter 2026 Financial Results Conference Call,
2. Amazon Investor Relations: Events
3. Bank for International Settlements: I. Progress and peril
4. Goldman Sachs: Why AI Companies May Invest More than $500 Billion in 2026,
5. International Monetary Fund: July 2026 World Economic Outlook Update
6. Meta Investor Relations: Investor Events
7. Meta Investor Relations: Meta Reports First Quarter 2026 Results
8. Microsoft: Microsoft Announces Quarterly Earnings Release Date,
9. Reuters Breakingviews: Meta’s fall shows punters crave clearer AI payoff
10. Reuters: Meta shares fall on concerns over AI spending, legal scrutiny
11. The Motley Fool: Stock Market Today, April 30
12. Tom’s Hardware: Google, Microsoft, Meta, and Amazon capex spending to hit $725 billion in 2026, up 77% from last year
13. Yahoo Finance: Meta stock sinks after Q1 earnings as company raises 2026 AI spending forecast