Who gains in an AI-supercharged economy?

This commentary was originally published by Project Syndicate and is reproduced here with permission.

Vanguard projects that U.S. GDP will grow around 3% in 2027—well above most professional forecasts—suggesting ongoing support for risk assets. This would mark not just a marginal improvement but a structural step-up in the economy’s growth path. Based on extensive analysis of current AI capabilities and comparisons with earlier breakthrough technologies, Vanguard expects a major economic shift.

AI is already contributing to economic activity, though it may take another year or two to know whether its impact will rival that of the personal computer. For Vanguard’s growth outlook to materialize, AI must evolve beyond its current automation stage—where it mainly replaces human tasks—into an augmentation stage that enhances worker performance, and ultimately into a phase where it enables entirely new products, services, and industries. While today’s narrative is dominated by automation, it is these latter stages that will determine whether AI becomes a true general-purpose technology.

Before electricity became commercially viable, few could foresee electric trams, cinemas, or home appliances. Similarly, the expectation that AI will mature into a general-purpose technology underpins Vanguard’s relatively optimistic view of the labour market. Fears of widespread job loss are understandable, but pessimistic scenarios often ignore the roles that do not yet exist.

They also tend to overlook the income and spending boost that could come if workers become many times more productive. The experience of accountants after the advent of professional software is instructive: their productivity surged. Disruption is not synonymous with pure job replacement. If fully developed, AI could deliver productivity gains strong enough to offset—and likely surpass—the drag from ageing populations, lower birth rates, and weaker immigration.

Where we are in the cycle

The transition from heavy AI investment to broad-based productivity gains will take years, not quarters, much like the late-1990s internet buildout. The current capex surge likely has at least another year or two to run, despite its already massive scale. Large AI “hyperscalers” appear financially able to honour their huge investment plans, and corporates are actively rolling out AI tools.

Financial markets, however, have moved faster than the real economy. Valuations—especially for big U.S. tech names—already price in a substantial portion of AI’s potential. Over the next year or two, strong earnings growth tied to AI infrastructure spending may still justify these levels and could push markets higher. But that is a short-term dynamic. Over longer periods, investment returns tend to rebalance, particularly during episodes of rapid technological change.

The coming rotation

History shows that the firms building transformative technologies are not always the biggest long-term winners; the main beneficiaries are often the users. Electricity ultimately created more value for manufacturers running 24/7 assembly lines than for power utilities. The car industry enabled greater wealth creation for suburban developers and retailers than for the carmakers themselves.

AI is likely to follow a similar pattern. The current buildout phase—dominated by hyperscalers, semiconductor companies, and core model developers—should eventually transition to a consumption phase in which end users across sectors capture most of the gains. Many of these user businesses currently trade at value-style valuations, and a significant share are outside the U.S., in service-heavy, ageing economies where productivity improvements will be especially valuable.

Potential beneficiaries include health-care providers that can automate paperwork and improve diagnostics, financial firms that can deliver more tailored advice at lower cost, and business services companies that can combine human expertise with AI-driven analytics. These firms are already experimenting with task automation and stand to benefit further if AI evolves to truly augment human skills.

What to watch

There is no guarantee that AI will reshape the economy positively, but certain signals would indicate that it is on that path: new entrants to the workforce equipped with AI-enhanced capabilities; a pickup in startup formation outside the core tech sector; and more frequent, genuine breakthroughs—such as major medical advances—driven by AI-assisted research. As these signs appear, they would mark the early stages of AI’s economic transformation, echoing the trajectories of electricity and the personal computer.

Positioning for the transition

The emerging opportunity lies in recognising that while markets may be broadly right about AI’s economic potential, they may be mispricing which segments will benefit over a full cycle. Value-oriented U.S. equities, developed markets outside the U.S., and high-quality fixed income currently offer attractive risk–return characteristics—providing downside resilience if AI underdelivers and upside participation if it succeeds—over the next five to ten years.

The message is not to exit technology or attempt market timing. Rather, investors should acknowledge that in a mature AI environment, leadership is likely to shift from today’s AI builders to the companies that use AI most effectively. This pattern has repeated with every major technological revolution. For long-horizon investors, this prospective rotation is both a risk to manage in growth-heavy portfolios and a forward-looking opportunity to position ahead of AI’s next phase.

Copyright: Project Syndicate [project-syndicate.org].