Chip rout tests AI investment case
CIO Daily Updates
From the studio
From the studio
Thought of the day
Thought of the day
What happened?
US technology stocks led broader markets lower on Thursday, as the post-election relief rally reversed and investors reassessed the implications of the Democratic gains, especially at the state level, for AI infrastructure and data centers. The Nasdaq 100 and S&P 500 fell by around 6.4% and 4.4%, respectively.
Semiconductor and AI infrastructure companies were at the center of the decline, including Qualcomm, which fell sharply after its profit outlook disappointed investors. The Philadelphia Semiconductor Index fell 12.3%, extending its drawdown since June to around 38.3%.
Pressure extended into Friday, with North Asia's tech-heavy indices leading declines. Japan's Nikkei 225 and South Korea's Kospi closed 6% lower, and Taiwan's benchmark index shed 4.5%. Major European stock indices all stand roughly 1% lower at the time of writing.
The sharp risk-off move followed an initial post-election rally, when the removal of political uncertainty and the prospect of policy gridlock were seen as supportive for stocks. Investors now appear to be focusing more closely on how greater congressional oversight and changes in state governments might affect electricity consumption, water use, permitting, and development costs for data centers.
Oil and geopolitical developments remained a secondary source of volatility, with Brent crude rebounding to around USD 100 per barrel before coming down to around USD 98 at the time of writing. US Treasuries and perceived safe havens have gained since the market close on Tuesday. The yield on 10-year US Treasuries has fallen by 12 basis points to near 4.84%, while gold is holding near USD 4,070/oz.
What do we think?
In our view, the decline reflects uncertainty about the cost and pace of the data center buildout rather than a sudden deterioration in the outlook. It also appears to represent a de-risking event rather than a material change in fundamentals.
Further technology volatility would not be surprising. The sector has delivered strong gains in recent years, leaving valuations, expectations, and investor positioning elevated in parts of the market. Investors are hence moving from rewarding spending commitments to demanding clearer evidence of monetization, capacity utilization, potential returns, and the ability to raise capital. The sensitivity of AI-linked stocks to company guidance, regulatory developments, and changes in risk appetite has increased.
Greater congressional oversight may indeed increase scrutiny of data center construction and AI regulation. However, except in the event of a widespread AI security issue, we believe broad federal restrictions would likely face legislative and presidential constraints.
And while political opposition at a state and local level may delay, relocate, or raise the cost of data center projects, it is unlikely to change underlying demand. Proposed US capacity of around 400 gigawatts already far exceeds our buildout assumption of roughly 90 gigawatts by 2030. We therefore see regulation as potentially redistributing and extending the investment cycle rather than derailing it, with power access, permitting, financing, and execution increasingly separating leaders from laggards.
Underlying AI-related demand also remains strong. Recent earnings have shown accelerating cloud revenue and growing advance orders for computing capacity. Recent reports point to significant chip backlogs among Chinese buyers. We forecast aggregate AI capital spending of USD 1.0tr in 2026, USD 1.4tr in 2027, and USD 1.5tr in 2028.
For the sell-off to signal a lasting downturn in the AI investment cycle, we think we would need to see evidence that AI-related revenue growth is slowing, cloud utilization is weakening, or lower AI costs are failing to generate sufficient new usage to absorb expanding capacity. We would also reassess the outlook if advances in algorithmic efficiency sharply reduced compute requirements, larger investments failed to produce commensurate gains in AI capabilities, or tighter financial conditions constrained spending by the largest technology platforms and frontier labs.
How do we invest?
We see potential for gains in broad equity indices over the next six to 12 months and believe the sell-off looks overdone, given our base case for only modest changes to the outlook for the pace, timing, and cost of data center deployment. We believe the AI investment cycle remains intact, corporate earnings growth is resilient, and a positive global cyclical backdrop should support equities.
Nevertheless, we acknowledge that investors may need more certainty about the data center policy environment before volatility fully abates, and we could see a period of market indigestion following the election results. This reinforces the importance of diversification and disciplined portfolio construction.
Investors should decide in advance how they would respond to further market swings rather than making allocation decisions during periods of stress. This could include identifying target weights for equities, technology, fixed income, and alternative assets, then addressing allocation gaps progressively as markets move. Structured strategies may enable investors to respond in a disciplined way.
Manage concentration risks. Although broad indices were down sharply on Thursday, volatility was even more elevated at the individual stock level. To prepare for potentially greater differentiation within the technology sector in the weeks and months ahead, investors should review portfolios and progressively reduce excessive single-stock concentration relative to technology and broad market benchmarks.
Within AI, we prefer diversification over concentrated exposure to individual companies experiencing bottlenecks. We favor financially strong platform and application companies that can monetize growing AI adoption, together with infrastructure businesses that have durable pricing power, strong competitive positions, and access to capital. We are more selective toward highly valued companies whose investment case depends heavily on a single component, capacity constraint, customer, or stage of the data center buildout. Diversified exposure across foundry, compute, networking, power, and other enabling technologies can provide participation in structural growth while reducing dependence on any one bottleneck.
Rebalance across sectors and regions. We reduced our sector preferences for US IT and US communication services to Neutral in February. Investors whose US tech exposure has risen materially in recent years and now stands in excess of benchmark weights should consider using opportunities to rebalance toward other sectors. Within equities, our preferred sectors include industrials, health care, and financials. Regionally, we have a relative preference for the Eurozone, emerging markets, and Asia ex-Japan.
Build a strong core portfolio. To prepare for possible further equity market volatility ahead, investors with significant bond underweights should prepare for opportunities to rebuild allocations: We see attractive diversification and income potential in fixed income, particularly in high-quality bonds with short- to medium-term maturities and in emerging market bonds. Allocations to alternatives and commodities can also provide investors with exposure to a broader, differentiated range of return drivers beyond traditional asset classes.