Capital Ideas^™ 
Investment insights from Capital Group

The United States economy is at a crossroads. Cracks are forming and growth is slowing as tariff policies weigh on hiring and curb consumer spending. On the other hand, generative artificial intelligence (AI) is driving a surge in productivity. The question is: Can AI fill the cracks in the U.S. economy?

My view is that headwinds from trade restrictions are real, but tailwinds from AI could be even more powerful. I expect GDP growth to slow to 1% or lower in the second half of 2025, but I don’t foresee a recession. In fact, the AI megatrend could become a structural tailwind and accelerate growth in 2026 and 2027.

Here are four charts to help explain why the U.S. economy may continue to show resilience.

1. Investments in AI have surpassed dot-com era

Tech companies have poured hundreds of billions of dollars into AI, despite concerns about future returns on their investment.

Most of that spending has been directed at building electricity intensive data centres. The boom has benefited certain U.S. based semiconductor, industrial and power generation companies. For example, Modine Manufacturing, which builds temperature management systems used in these centres, has seen its stock rise 22.1% this year as of August 26. Constellation Energy’s deals with tech giants Meta and Microsoft to supply data centres with nuclear power have helped propel its stock 41.5% over the same period.

Of course, all that spending begs the question: Is the U.S. headed for a bust that rivals the tech wreck in 2000? I wouldn’t rule out an “AI winter,” but the 1990s PC and internet revolution is different from today’s tech story. The former was about hardware, connectivity, networks and information, whereas AI is about extracting accumulated knowledge. And unlike the dot-com bubble, today’s companies are flush with cash and have robust earnings. Valuations vary widely, so deep, fundamental research can help separate the winners from the losers.

Ultimately, the biggest winners of the AI spending cycle may not yet exist. Early pioneers like Cisco built the internet infrastructure that gave rise to companies like Netflix, Amazon and Google. Today’s AI-infrastructure buildout could similarly pave the way for a new generation of businesses to flourish.

2. Productivity surges are rare — and this is one of them

Big surges in productivity are rare for developed economies. In the U.S., it took years for the personal computer and internet to deliver substantial productivity gains. I believe AI may have already ushered in a new era of exceptional productivity.

Producing more per hour allows companies to maintain or grow profits, even as wages and expenses rise. A high rate of productivity growth has long given the U.S. a structural advantage and contributes to its status as an economic leader. Compared to many developed economies, the U.S. leads in productivity.

I think U.S. productivity could nearly double to an annual rate of 4% per year over the next five years. This increase is constructive for GDP and may even help moderate inflation.

While I am concerned AI bottlenecks may dampen productivity potential, I’m hopeful advances could help solve these problems. Deployment costs for large language models (LLM) are 90% cheaper than they were two years ago.

I’m seeing the impact firsthand in my recent travels. At this summer’s annual conference for economists in Dublin, I noticed a significant increase in the number of people applying AI to their workflow. This story rhymes with Capital Group’s own AI usage: It’s widespread and has had cost and time-saving benefits beyond initial expectations. As a result, we’re putting more resources into employee training and agentic AI, which requires minimal human oversight.

3. Labour markets are rewiring for the future

From sewing machines to cars, new technologies have transformed labour markets. The personal computer’s impact on labour markets serves as a good reminder that tech innovations have created new, more numerous, and often higher paying jobs after an initial period of job displacement.

From 1970 to 2015, PCs destroyed 3.5 million U.S.-based jobs, largely related to typewriting, bookkeeping and auditing, according to a McKinsey & Company report. Over that same time frame, PCs also created 19.3 million jobs for a net gain of 15.8 million.

AI is likely no different, and the U.S. is witnessing some AI-related job displacement. It’s no coincidence layoffs are concentrated in tech companies heavily investing in AI. Many are focused on preserving profit margins, and several over hired during the pandemic. At the same time, they’re in bidding wars to hire AI researchers and developers.

Economists are beginning to include AI in economic models, and they’ve homed in on efficiency gains to understand AI’s impact on labour markets. Efficiency gains are higher for tasks with vast documentation such as law or coding, standardized tasks usually associated with entry-level positions, and high wage tasks ripe for cost cutting.

But it’s important to separate tasks from jobs. From my perspective, AI ultimately allows workers to focus on other, higher value activities and expand their current roles.

Of course, we can’t ignore the fact that tariffs and the uncertainty around them have led to layoffs and hiring freezes. The next few months could shed more light on the trajectory of job markets, but for now, I don’t see evidence of widespread pain.

4. Signs point to a slowdown, not a recession

There are many reasons to worry about the U.S. economy. Inflation has ticked higher, job creation is down, and GDP is weakening. I believe these are signs of a mid-cycle economic slowdown rather than a recession.

That doesn’t mean there won’t be pain, particularly since the full effects of President Trump’s tariffs have yet to be realized. Additionally, equity market valuations are high, so any shock could change the narrative. An economic slowdown typically leaves little room for error.

Nevertheless, I believe tailwinds from AI will continue to drive investments and fill cracks in the U.S. economy. Company earnings have also generally held up, with some reporting healthy consumer spending, particularly among higher income customers. Disney CFO Hugh Johnston recently declared their consumers were doing very well. Finally, the U.S. Federal Reserve may soon restart its rate-cutting cycle, which could lower borrowing costs and help steer the economy out of a rough patch.