Thought of the day

What happened?
US equities fell again on Thursday as concerns about the AI trade combined with a more hawkish interest rate backdrop. The S&P 500 and Nasdaq Composite closed down 2% and 2.3%, respectively, and the sell-off continued in Asia on Friday. The Nikkei 225 slid over 5%, while Taiwan's benchmark index was 6.2% lower.

The market weakness was relatively broad. All major S&P 500 sectors traded lower, while declining stocks substantially outnumbered advancing stocks on both the NYSE and Nasdaq. Recent selling has followed persistent pressure on chipmakers and cloud companies as investors question the scale of investment required to build AI infrastructure.

Monetary policy uncertainty has added to fragile market sentiment. Minutes from the Federal Reserve's October meeting indicated that nearly all policymakers expected further tightening to be necessary if the US economy evolves as anticipated. Markets were pricing a near certainty of a 25-basis-point increase in December, with persistent inflation pressures also supporting the US dollar. US two-year and 10-year Treasury yields fell 4bps and 2bps, respectively, while the US dollar held recent gains. Higher rate expectations and dollar strength weighed on gold, which fell to a 14-month low of around USD 3,575 an ounce at the time of writing.

Without taking single-security views, sentiment was also dented by reports from the Financial Times and newswires that investigative research house Hunterbrook had published a short-sell report on AMD (-7.7%) over fears that its business with China may be affected by stricter US chip restrictions; NVIDIA (-3.3%) also extended its recent losses despite having issued an upbeat revenue forecast earlier in the week.

What do we think?
As outlined in our recently published Year Ahead publication, we do not think that the economic expansion, corporate earnings cycle, or structural AI investment story has broken. Our base case for 2027 remains for economic growth close to trend, rising corporate profits, further growth in AI investment, and moderating inflation later in the year. We still expect this combination to support additional equity gains and positive fixed income returns.

But the risk-reward in equities has become more balanced. We recently downgraded global equities from Attractive to Neutral, reflecting elevated expectations, high index concentration, and a wider range of potential outcomes around AI. In particular, the AI debate is shifting from whether investment will continue to whether it can generate sufficient revenues and shareholder returns. Today's price action supports this shift in focus.

Cumulative US technology investment in AI infrastructure could approach USD 5tr by the end of 2028. But we estimate that AI model and application revenues will probably have to reach at least USD 1.5tr by 2030 for the industry to earn an adequate return. While this is possible, it is unlikely that every company will reap the rewards from this: Competition, price cuts, costly monetization, and open-weight models may threaten returns. Financing, regulatory, and company-specific risks also add to uncertainty.

Tighter monetary policy expectations add another source of volatility. While we foresee further near-term rate increases (namely 25bps hikes in the US in December 2026 and March 2027), we do not anticipate a prolonged tightening cycle. The cumulative effects of higher rates look set to weigh on consumer demand, while we believe base effects will contribute to a moderation in underlying inflation.

How do we invest?
In our base case, we expect upside for equity markets in 2027. But the range of potential outcomes is broader. The challenge for investors in 2027 is constructing portfolios that can prove resilient across a variety of scenarios.

Refocus equity exposure. In our base case, we expect corporate earnings growth to support equities, with around 15% S&P 500 earnings growth in 2027. But with the top 10 companies accounting for 39% of the index, relatively small changes in expectations about a handful of AI-linked businesses can have an outsized effect on returns.

The Eurozone is our preferred equity region, with 15% earnings growth forecast for 2027, a lower valuation than the US, and earnings drivers spanning manufacturing, defense, infrastructure, and electrification. At a sector level, we favor industrials and health care, while equal-weight strategies in the US can be a convenient way to help reduce technology concentration.

Within US technology, we favor financially strong hyperscalers with visible AI monetization, alongside select leaders in compute, foundry, semiconductor equipment, and infrastructure software. We remain cautious on leveraged capacity providers, lower-quality suppliers, and companies whose valuations depend on persistent bottlenecks or distant cash flows.

Rebalance into fixed income. Elevated yields offer an attractive combination of income and potential portfolio resilience, while providing a return cushion even if rates do not decline materially. We favor intermediate duration, broadly two to 10 years. We see attractive return opportunities in quality bonds, and within fixed income, we see the best return opportunity in emerging market and high yield bonds. EM sovereign and corporate bonds currently yield roughly 7.1% and 7.0%, respectively, while US high yield offers around 8.3%.

Use volatility, rather than fear it. Investors should prepare for volatility instead of trying to predict every setback. Structured strategies can help investors build exposure in more tailored ways, potentially limiting losses, generating additional income, or allowing more measured market participation. Higher implied volatility, all else equal, may raise the appeal of such approaches. However, investors must understand the potential trade-offs, including capital loss, issuer, liquidity, and opportunity-cost risks, so we favor diversification and careful sizing of structures.

To read our views in more detail, please see Year Ahead: Sustaining resilience(PDF, 4 MB) and Monthly Letter: Sustaining resilience(PDF, 812 KB).