Thought of the day

What happened?
US stocks fell on Monday amid renewed tech sector volatility ahead of a key earnings report from NVIDIA, while the Trump administration's move to restrict Chinese companies' remote access to advanced AI chips and servers added to the selling pressure. The S&P 500, Nasdaq 100, and Philadelphia Semiconductor Index closed 1.9%, 2.7%, and 5.3% lower, respectively.

The Bureau of Industry and Security issued a new rule that marks a significant shift in how the US restricts access to advanced computing power. According to the new measure, data center operators in Singapore, Malaysia, Thailand, Indonesia, and Brazil will need to obtain a US government license before buying from US chipmakers, as it aims to restrict China and other US rivals from accessing offshore computing power that uses American-designed processors.

Markets also digested a speech from Federal Reserve Chair Kevin Warsh that added a macroeconomic dimension to existing concerns about AI investment returns. Warsh described AI as a potentially transformative technology that could lift productivity and economic growth. But he also identified the possibility that substantial capital investment and stock-market gains could be followed by disappointing returns and falling equity valuations, producing an economic slowdown and a more pronounced rise in unemployment.

Selling broadened beyond the largest technology companies, with almost all S&P 500 sectors declining. With risk-off sentiment on the rise, bonds rallied and the yield on the 10-year US Treasury fell 1 basis point to 5.13%.

What do we think?
In our view, the market reaction reflects a further positioning adjustment as investors reassess whether AI demand, utilization, and monetization can support elevated capital spending and market expectations. Recent volatility does not yet change our conviction in the structural AI opportunity or our constructive view on diversified equities.

Underlying demand for computing capacity remains robust. Recent earnings have shown accelerating cloud revenue and growing advance orders, and we forecast aggregate AI capital spending of USD 1.0tr in 2026, USD 1.4tr in 2027, and USD 1.5tr in 2028. Most spending is concentrated in large, generally well-funded technology platforms, which should limit the direct impact of financial stress among more leveraged infrastructure providers on overall investment. Project delays, repricing, and consolidation appear more likely than a broad collapse in AI spending.

However, as the cumulative investment in the AI build-out moves into trillions of US dollars, investors are placing greater weight on how this spending translates into revenue, free cash flow, utilization, and sustainable returns. NVIDIA’s results will therefore be assessed not only for chip demand, but also for evidence on data-center utilization, margins, order visibility, supply constraints, and customers’ capital-spending plans. Evidence that computing capacity is being absorbed by rising usage would support confidence in the investment cycle. Weaker utilization, slowing customer commitments, or greater difficulty converting investment into revenue could renew concerns about overcapacity and prospective returns.

The new chip restriction could affect 5-10% of next year’s AI demand, according to our estimates, but the revenue impact on US chipmakers should be limited as supply remains tight and displaced capacity can be sold elsewhere. Major US hyperscalers also appear relatively insulated, given the scale and diversity of their cloud businesses. While we see impact on smaller cloud and regional data center operators as well as standalone Chinese AI labs, the greater concern is that the rule adds to pressure on elevated 2028 capital spending and earnings expectations.

Warsh’s speech also reinforces the unusually wide range of economic and market outcomes associated with AI. The bar for performance is rising, and greater differentiation across the AI ecosystem is likely to make sharp market moves more common at this stage of the cycle.

Still, for the sell-off to indicate a lasting deterioration in the AI investment cycle, we would need to see weakening compute demand, slowing AI-related revenue growth, poor capacity utilization, or tighter financial conditions constraining spending by the largest technology platforms. The breadth of selling will also matter. Concentrated weakness in chips, data-center companies, and other highly valued growth stocks would remain consistent with a valuation and positioning adjustment. A sustained spread into financials, industrials, and other cyclical sectors, accompanied by weaker economic, credit, or earnings data, could indicate a broader reassessment. Current evidence does not yet support that conclusion.

We therefore continue to see potential for gains in broad equity markets over the next six to 12 months, supported by resilient economic growth and corporate earnings. Near-term volatility reinforces the importance of selectivity and disciplined portfolio construction rather than a broadly defensive stance.

How to invest?
We believe investors should remain engaged with the structural AI opportunity while ensuring that portfolios are not overly dependent on a single company, sector, or catalyst. Diversification across the AI value chain, other equity sectors, regions, and asset classes can help investors participate in further growth while reducing sensitivity to company-specific setbacks.

Retain diversified exposure to AI. Investors should avoid abandoning the structural opportunity in response to short-term volatility. We favor established technology platforms with tangible AI revenues, existing computing capacity, strong balance sheets, and the ability to finance investment internally. Within semiconductors and hardware, investors should emphasize competitive positioning, reasonable valuations, and lower exposure to excess capacity. Defensive technology businesses can complement more cyclical AI holdings.

Reduce excessive concentration. Investors with outsized positions in individual technology companies should use periods of market strength to reduce concentration relative to technology-sector and broad-market benchmarks. Diversifying across compute, foundries, platforms, applications, power infrastructure, and other enabling technologies can reduce dependence on a single bottleneck or business model. Investors whose overall technology allocation has moved above its intended weight should also consider rebalancing into other sectors. Our preferred US sectors include health care and industrials.

Broaden portfolios across regions and asset classes. We remain positive on Eurozone equities, where valuations are lower and the earnings outlook is supported by a manufacturing recovery and structural investment. Investors with significant bond underweights should consider rebuilding exposure to short- and medium-maturity high-quality bonds, where elevated starting yields offer attractive income and diversification potential. Broad commodities and select alternatives can provide additional return drivers for investors able to tolerate their specific risks. A strong multi-asset core can help investors remain positioned for equity gains while limiting the impact of further technology volatility.