IMF Chief Economist Tenreyro: AI Investment Will Speed Up Global Growth

IMF Chief Economist Silvana Tenreyro told Bloomberg that global growth should accelerate on the back of investment in artificial intelligence. The conversation with Avril Hong took place on the sidelines of the fall IMF and World Bank meetings. The published announcement contains no specific percentages and no breakdown by country. What it does have is the claim itself, and it matters because of who is making it.

What exactly was said

The source offers few details. Tenreyro ties the expected pickup in global growth to AI investment. The announcement does not say how many percentage points this will add, over what time horizon, or in which economies. Those estimates are usually given in the World Economic Outlook, which the IMF publishes at the fall meetings. For now we have only the headline of the interview, not the fund's calculations.

For comparison, in its October 2025 report the IMF estimated global growth at 3.2% for 2025 and 3.1% for 2026. At the time, the fund also said that the AI investment boom was supporting the US economy, and warned that valuations of companies in the sector resembled the bubble of the early 2000s. The tone has now shifted: the emphasis is on possible acceleration rather than the risk of a correction. That is telling, but a single interview is not enough to conclude that the fund has changed its position.

Where the growth comes from

Today AI's macroeconomic effect flows through two channels, and they differ in reliability.

• Capital expenditure. Data centers, servers with accelerators, networks, energy, and construction. This money is being spent right now and feeds directly into GDP. Hundreds of billions of dollars a year from the largest cloud companies are visible in investment statistics.
• Productivity. The return from deploying models and agents in business processes. This channel is slower and harder to measure: the effect depends on how quickly companies restructure their work, not just on whether they buy subscriptions.
• Spillover effects. Demand for electricity, cooling equipment, construction materials, and semiconductors. Countries far from model development also benefit, for example suppliers of raw materials and components.

Today's statistics mostly capture the first channel. If growth rests on construction and hardware purchases, it holds only as long as buyers have money and confidence that the investment will pay off. The second channel determines whether the effect lasts.

Where the catch is

Investment-driven growth is vulnerable. It depends on the cost of capital, on investor optimism, and on revenue from AI services catching up with infrastructure costs. The bottlenecks are also well known: access to power grids, supplies of chips and memory, and construction timelines. If some projects are delayed or revised, the contribution to growth will shrink before any productivity gains show up.

There is also a geographic imbalance. Most of the investment is concentrated in the US and in a handful of companies, with a share going to China, the Middle East, and Europe. For the rest of the world, "AI-driven acceleration" may mean indirect demand for raw materials and components rather than a boom of its own.

Practical takeaway

For those building a business around AI, the IMF's signal is useful, but it is no substitute for your own calculations. It confirms that infrastructure demand remains a priority for large buyers and regulators. At the same time, check the numbers against the fund's report itself, not against retellings of the interview.

Infrastructure suppliers and startups should watch three indicators: interest rates, the trajectory of cloud giants' capital expenditure, and actual revenue from AI products. If the first two stay favorable and the third grows, Tenreyro's thesis holds. If revenue lags, the fund's optimism will have to be revised, as has already happened in previous technology cycles.

Next step: wait for the IMF's full forecasts and see what share of growth the fund attributes specifically to AI.