The intelligence dividend: what AI means for portfolios

The AI investment boom is already one of the largest capital cycles in history. But its impact on economic growth, inflation and employment is less certain than the scale of spending implies.

The productivity payoff – when and how much?

Since 2024, 64 cents of every dollar of U.S. GDP growth has been attributable to tech spending.¹ Hardware investment has risen from nearly 2% of U.S. GDP to more than 3%, representing hundreds of billions of dollars.² This cycle has already seen productivity growth above 2%, ahead of last cycle’s 1.5% trend, but that largely reflects post-pandemic labor market dynamics rather than AI.³ The AI boost has yet to arrive, given that more than three in four U.S. businesses have yet to incorporate AI.⁴

That gap between investment and deployment explains why expectations of AI’s productivity impact vary so widely. The most conservative estimates put AI’s current contribution to total factor productivity at close to zero. More optimistic projections from the OECD suggest annual labour productivity gains of up to 1.3 percentage points for G7 economies with ‘high AI exposure’.⁵ Both can be true at different points in time. The internet’s productivity payoff took the better part of a decade to show up in national accounts. There is little reason to think AI will be faster as the organizational changes required are likely more complex.

A complex inflation picture

What makes the inflation outcome particularly difficult to forecast is that AI is exerting pressure in multiple directions at once.

A newly created quarterly U.S. AI-intensity index finds that AI has already made a positive and rising contribution to U.S. inflation, with the most AI-exposed sectors tending to show the highest inflation rates.⁶ The steep costs of the AI buildout are arriving faster than the productivity gains that are supposed to offset them.

Those costs are most visible in energy. U.S. power inflation ran at 6.9% year-on-year through December 2025, more than double the headline PCE gauge.⁷ Consumer electricity inflation is likely to remain near 6% through 2027, while data center demand is set to nearly double from its current approximately 4% share of total U.S. electricity by 2030.⁸ Inflation is already flowing through to input costs across manufacturing, services, and household bills.

The dynamic is further complicated by expectations. If firms and households anticipate AI’s future productivity gains, they may consume and invest today against tomorrow’s expected income, bringing inflation forward before the gains have arrived.

As AI adoption broadens and output per worker rises, the resulting fall in unit labor costs could exert meaningful downward pressure on services inflation (the same mechanism that drove the disinflationary expansion of the 1990s). But a June 2026 World Economic Forum survey finds economists now expect AI-driven productivity gains to take at least another two years to materialize across most sectors, which is longer than they thought at the start of 2026.⁹

Exposed, but not displaced, workers

The employment picture follows a similar pattern. Significant pressure is building, but only limited disruption is visible in the data so far. The IMF estimates that close to 40% of global employment is exposed to AI, and roughly 4 of 5 U.S. workers have at least 10% of their tasks exposed to large language model capabilities.¹⁰ And yet since the introduction of generative AI, economy-wide job losses and wage declines have not materialized. General-purpose technologies have historically restructured work before they reduce it, sometimes by a decade or more, and AI appears to be following that pattern¹¹

Concern that AI generates a demand collapse among displaced high-income workers, sometimes referred to as white-collar recession thesis, remains a tail risk rather than a base case. While more than 142,000 U.S. tech workers have been laid off year-to-date, which is around a 33% increase over the same period in 2025, analysts estimate that the majority is still attributable to cost discipline and an unwind of pandemic era over-hiring and only roughly 25% due to AI and automation.¹² And by our estimates, productivity booms tend to be associated with a lower unemployment rate.

Across major developed markets, unemployment in AI-exposed professional services has not risen materially, and wage growth remains resilient. But much like productivity, the labor market impact tends to arrive with a lag. Key to watch will be a sustained rise in unemployment among college-educated workers in AI-exposed sectors, accelerating headcount reductions in professional services citing AI-driven inefficiency gains, or deterioration in entry-level hiring across finance, law and tech.

Mixed signals in bond markets

How the macro dynamics of inflation and employment play out have direct consequences for investors. The fixed income market’s own read on AI is sending a counterintuitive signal. When NBER researchers studied U.S. Treasury, TIPS and corporate yields around major AI model releases in 2023 and 2024, they found that yields fell, consistently and persistently by more than 10 basis points on average and remained lower for over two weeks after each release.¹³ On the surface, bond markets priced AI as a disinflationary productivity shock, which may prove correct in the long run, but a second force is pushing in the opposite direction and the NBER cited the reaction as reflecting downward revisions in expected consumption growth.

Tech hardware investment is more than 3% of GDP, and financing the AI buildout (data centers, semiconductor supply chains, power infrastructure) can generate continued upward pressure on term premium through long-duration corporate bond issuance.¹⁴ The Dallas Federal Reserve flagged this as a structural force rather than cyclical.

Underneath both sits the question of where the neutral rate (r*) settles, the policy rate where the economy is not expanding nor contracting. A productivity-driven AI economy implies a higher r*, as stronger growth prospects mean a higher equilibrium cost of capital.

But if AI-driven gains accrue disproportionately to capital owners while displacing portions of the workforce, higher precautionary savings, political pressures for redistribution and greater social uncertainty could offset some of that upward pressure. The Cleveland Fed’s current estimate of the nominal neutral rate spans a wide confidence band from 2.9% to 4.5%.¹⁵ The central tension for fixed income investors is timing. AI’s supply-side benefits are long-dated, while its demand on capital markets is immediate.

Portfolio positioning when the AI payoff has yet to arrive

The energy data and the capex financing pipeline explain why AI is not a straightforward disinflationary story. Term premiums are being rebuilt by structural supply from AI infrastructure financing. For credit investors, the phase of the AI investment cycle matters more than broad sector exposure. For example, infrastructure credit is backed by contracted cash flows and long-duration demand, while spread pricing among AI-exposed incumbents may not reflect the disruption risk embedded in those names. The next test is whether hyperscalers can demonstrate that their capital spending converts into returns.

Two main positioning themes for investors stand out in our view. First, despite the recent move higher in longer-maturity yields, we remain cautious on duration exposure across fixed income markets. There is still potential upside to rates from the near-term, inflationary impact of AI investment. Second, we recommend a modestly risk-on stance with careful security selection within below-investment grade credit. Avoiding companies and sectors exposed to disruption is crucial, but income potential remains substantial in markets such as high yield bonds, senior loans and emerging markets where active managers can navigate the AI crosscurrents and select potential winners.