AI meets accountability

Markets are no longer underwriting AI on promise alone. Investors now want evidence that rising AI spend delivers margins, productivity, and cash flow. As enthusiasm matures, selectivity returns—and execution becomes the differentiator.

AI disruption has been a driving theme to start the year. Software & Services leads declines (-18%), with broader service industries also lagging as investors price in AI-driven disruption to labor-intensive business models. At the other extreme, Energy, Utilities, and Capital Goods lead performance—capital-intensive, asset-heavy, “old economy” sectors are benefiting from tangible assets, pricing power, and relative insulation from near-term AI displacement.

Weekly highlights

AI isn’t broken—but the market needs to see the impact

For the better part of two years, markets have treated artificial intelligence as a one‑way bet. If a company spent heavily on AI infrastructure, investors were willing to trust that the payoff would arrive in due course. That era is ending. AI is not falling apart—but the market is done underwriting it on faith alone.

The shift has been subtle but decisive. We have moved from a “build it and they will come” mentality to something closer to a “show me the money” regime. Investors increasingly want to see that large-scale AI investment is generating real financial results, or at least a credible path to them. Last quarter offered a glimpse of what this new reality looks like. Two companies, both announcing surging AI‑related capex, received sharply different market reactions: one showed clear evidence that AI was already lifting margins; the other struggled to demonstrate the link. The market now treats those differences seriously.

This is not skepticism for skepticism's sake. It is a sign that AI is maturing. In every major technological cycle, investors eventually shift from rewarding ambition to rewarding execution. That transition is healthy. It forces capital to flow to companies that can translate engineering breakthroughs into genuine productivity and genuine cash flow—not just marketing slides.

Valuations are sending a message 

Part of the reason investors have become more selective is simple arithmetic. The idea that valuations in US large-cap technology have “normalized” is wishful thinking. Even after a wobble, many AI winners still trade at stretched multiples relative to their own history—and to European and smaller‑cap peers. That does not make them uninvestable; it simply means the hurdle for incremental belief is higher. In this environment, selectivity is not an intellectual preference—it is a pricing necessity.

The market itself is enforcing that discipline. Capital is now flowing toward firms that can articulate—with evidence—how AI improves targeting, reduces costs, or boosts revenue per user. The rest are discovering that “AI exposure” is no longer enough to guarantee a premium multiple.

Capex: A surge, not a new normal

One point often lost in the noise: rising AI capex does not imply that global capital intensity is entering a new, permanently higher regime. Much of today’s spending is front‑loaded—the cost of building the computational backbone for the next decade. Once capacity and power constraints ease, growth in capex should slow, even if absolute levels remain high. For investors, this distinction matters. A world of ever‑rising capital intensity erodes free cash flow; a world of front‑loaded investment sets the stage for operating leverage later.

Labor will shift—but not collapse

The debate over AI and labor is increasingly polemical, but the most dramatic scenarios—mass, rapid displacement across knowledge industries—look implausible. Labor markets do not adjust that way. Regulation, institutional norms, and sectoral frictions slow the process. What we are far more likely to see is labor rotation: tasks shifting between roles, new job categories emerging, and younger workers absorbing the brunt of the transition. That is disruptive, but it is not dystopian.

Productivity is real—but not unbounded

AI is already lifting productivity in ways that are visible in corporate reporting. But expectations of exponential productivity growth run up against physical limits. Chips, energy, and data centers are not abstract constraints; they are binding ones. The pace of AI‑enabled productivity gains will depend as much on power grids and semiconductor supply as on advances in model architecture.

Not everything will mean‑revert

One area where investors may need to temper their optimism is the assumption that recent spikes in energy, gold, and defensives will unwind cleanly once geopolitical tensions fade. Even if spot prices settle, the risk premia behind them may not. Energy volatility now looks more structural than cyclical. Gold, too, has strengthened its case as a strategic asset. And in equities, defensives with AI‑adjacency—such as utilities and healthcare—appear far more resilient than staples, which look most exposed to reversal.

The memory of supply shocks tends to linger. Markets carry scars. One consequence is a higher-forlonger equity–bond correlation—a regime we may be living with for some time.

Inflation fear still rules policy

Even without a fresh inflation shock, the fear of inflation is enough to limit central banks’ willingness to cut aggressively. That matters. It suggests any industrial revival—tentative though it is—may struggle to gain traction without the customary support of accommodative policy. Rate relief will come, but not at the pace markets once hoped.

A more skeptical, more rational market

None of this means AI enthusiasm is misguided. The technology is transformative. However, we are now entering the phase where investors insist on evidence, not aspiration. That will make markets bumpier, but ultimately healthier. AI is not broken. It is simply being repriced—from a story to a strategy, from a theme to a testable proposition.