IMF warns of a “reevaluation of technology expectations” as AI spending surges—and results lag

The IMF is getting more optimistic about 2026. However, Corporate leaders are not. In its January 2026 World Economic Outlook update, the Fund raised its forecast for global growth to 3.3% in 2026 (and 3.2% in 2027) and summed up the moment in a line that reads like a verdict: “Resilient growth as technology and adaptability offset trade policy headwinds.”

On the same day, PwC’s 2026 Global CEO Survey landed with a colder message from inside companies. After surveying 4,454 CEOs across 95 countries and territories, PwC found that 56% say AI has produced no meaningful financial benefit so far, and only 12% say it has delivered both revenue gains and cost savings.

That gap—between macro optimism and micro disappointment—is the real AI story heading into 2026.

The IMF upgrade is about spending, not magic

The IMF’s revision is not framed as an AI productivity miracle arriving overnight. It’s about the near-term effects of a capital build-out: data centers, chips, networking gear, and the construction and power infrastructure that comes with them. Reporting around the IMF update describes the AI infrastructure boom as a meaningful support to growth.

This is how tech booms often show up in GDP first: not through suddenly smarter offices, but through cranes, grid upgrades, and procurement budgets.

The IMF also notes that “technology investment, fiscal and monetary support, accommodative financial conditions, and private sector adaptability” are helping offset trade-policy shocks.

In other words: the world economy looks steadier partly because the AI build-out is behaving like a stimulus program—real money, real jobs, real orders.

The IMF’s warning is also about AI

Still, the IMF is careful to label downside risks. In the same update, it flags “reevaluation of technology expectations” as a key threat.

The fear isn’t that AI is fake. It’s that the timeline is being mispriced.

If markets assume every dollar poured into compute turns quickly into revenue and productivity, then a slower reality—pilots that don’t scale, products that don’t monetise, costs that don’t come down—can trigger a sharp repricing.

Separate reporting notes the IMF warning that current resilience is increasingly reliant on US tech investment and AI-driven optimism, with a risk of correction if profitability disappoints.

This is the dot-com echo, updated for 2026: the infrastructure is real, but the cash flows are uneven—and debt and valuation narratives don’t like “uneven.”

PwC’s “AI divide”: most CEOs can’t prove the return

PwC’s numbers cut through the hype.

Despite widespread experimentation, PwC reports only 12% of CEOs claim AI is delivering both cost and revenue benefits; 33% report gains in either cost or revenue; and 56% report no significant financial benefit so far.

PwC Global Chairman Mohamed Kande described the situation as a split between companies translating AI into measurable returns and those struggling to move beyond pilots.

That line matters because it describes where the friction actually is: not in model demos, but in execution—data readiness, workflow redesign, risk controls, procurement discipline, and the unglamorous work of turning “AI features” into something customers pay for.

The Middle East is betting big on the infrastructure layer

Nowhere is the “spend first, prove later” dynamic more visible than the Gulf.

The UAE and Saudi Arabia have spent the last two years positioning themselves as AI infrastructure states: cheap energy (by global standards), access to capital, and a geopolitical strategy built around being indispensable to supply chains rather than stuck on the edge of them.

In the UAE, Microsoft and Abu Dhabi-based G42 announced a 200-megawatt data center capacity expansion, described as part of a wider multi-year investment plan with spending planned through 2029.

Saudi Arabia, meanwhile, has been building out Humain, an AI company backed by the Public Investment Fund. Reporting describes plans for early-2026 data centers and partnerships tied to advanced US chips, including Nvidia systems and a major AMD deal.

The UAE has also sought a formal role in AI and semiconductor supply chains by joining a US-led initiative aimed at strengthening those networks, underlining the country’s logistics and energy advantages.

This is the Gulf’s play: if the AI boom is real, they want to own the “pipes”—compute, power, and the sovereign cloud layer. If the boom wobbles, they still face the same exposure everyone else does: expensive assets that only pay off if demand arrives.

What changes in 2026: the story has to meet the spreadsheet

The IMF is effectively saying the AI capex cycle is strong enough to lift forecasts. PwC is saying the benefits are narrow enough to fracture the corporate world into winners and laggards.

Both can be true at once—right up until markets decide they need one clean story.

If 2026 becomes the year where AI spending keeps rising but CEO-level monetisation stays weak, the pressure won’t just be on startups. It will land on every balance sheet funding new compute: cloud providers, data center operators, utilities, and the governments selling “national AI strategy” as economic destiny.

The IMF’s headline says “resilient growth.” PwC’s data says “uneven returns.”

The space between those two statements is where the next phase of the AI cycle will be decided.