The U.S. economy continues to surprise on the upside. With second quarter data showing real GDP growth at a blistering 3.8%, it’s outpacing many of its peers. On closer inspection, however, a large portion of that strength appears to stem from one powerful engine: investment tied to artificial intelligence (AI) infrastructure.

This outsized effect flows directly into the investment component of GDP, showing up as spending on data centre construction, high-tech equipment like GPUs, and the R&D to create the models themselves.

Estimates say that, without AI investment, GDP growth in the U.S. would be flat. Despite information processing equipment and software accounting for only 4% of GDP,  AI-related spending is estimated to be responsible for nearly two-thirds of U.S. GDP growth so far this year.

This growth story is being written by just a handful of corporate giants, with firms like Meta, Alphabet, Amazon and others committing hundreds of billions to AI infrastructure.

This is where the story gets complicated, as this spending spree creates two fundamental risks.

First, the return on this massive investment is questionable. A counterpoint has emerged from the MIT Media Lab’s “State of Business in 2025” report. It finds that despite an estimated US$30-40 billion in enterprise spending on generative AI, about 95% of organizations report no measurable impact on their bottom line. Just 5% of pilots appear to be scaling into systems that generate large business returns. The principal bottlenecks are not regulation, but rather poor workflow integration and misalignment with core business processes.

Second, the source of the growth is dangerously concentrated. We are often told that the stock market is not the economy, however, the two seem to be more intertwined than ever. A handful of mega-cap technology firms dominate the AI story and U.S. growth is more vulnerable to risks like regulation of, for example, intellectual property and data privacy, chip supply disruptions, or energy constraints. To illustrate this concentration, just five tech companies (Apple, Amazon, Microsoft, Alphabet, and Nvidia) represent nearly 30% of the entire S&P 500 Index. Despite the strong headline numbers, the base of firms contributing to the boom remains worryingly narrow.

This leaves economic observers with more questions than answers. It's clear that AI spending is doing the heavy lifting for U.S. growth right now, but big questions lie ahead. How long can companies invest at this pace before investors demand returns? If AI-related capex slows before productivity gains arrive, U.S. growth could look a lot more ordinary. Or, if even a modest slice of today’s projects start to deliver, this investment could finally translate into the broad-based productivity boom we've all been waiting for.

Answer to the previous trivia question: The Exchange National Bank of Chicago opened the first drive-in teller windows on November 12, 1946.

Today’s trivia question: What city is hosting the Grey Cup on Sunday?