Written by: Chao Xiang Research

The median stock in the semiconductor sector has dropped by 53% from its peak, and the Philadelphia Semiconductor Index has fallen by about 40% since mid-June. On November 8, Morgan Stanley released a 93-page global thematic strategy report with a straightforward title: "Buy the Drop." The report believes that the market has overreacted to the results of the U.S. midterm elections, and the total capital expenditure for AI has not been downgraded; only about 20% of U.S. projects (accounting for 10% of global projects) have been pushed from 2028 to 2029.

The elections impact the pace, not the total

In the 2026 midterm elections, the Democratic Party secured both the House and Senate and won gubernatorial races in Texas, Ohio, and Georgia. Texas' incoming governor, Gina Hinojosa, proposed a pause on new data center approvals, and Ohio's incoming governor, Dr. Amy Acton, proposed a conditional pause. Panic spread throughout the entire AI supply chain. The report argues that projects already under construction will not be interrupted, and the capacity forecast for 2027 remains unchanged. The impact is concentrated in 2028: about 5GW of IT capacity (accounting for 4% of the total U.S. planning) is delayed from Texas and Ohio to 2029. In other words, the elections change the pace of construction, not the AI demand itself. The report maintains the capital expenditure forecast for hyperscale cloud providers at $1.6 trillion and $1.8 trillion for 2027/28.

Delays actually extend the AI investment cycle

The report calculated that the new capacity for U.S. data centers in 2028 has been revised down from a previous forecast of 24.9GW to 19.9GW, with year-on-year growth shifting from positive to -4%. However, the delayed capacity has not disappeared but has piled up for 2029, making the new addition jump from 22.7GW to 27.6GW, with a year-on-year growth rate reaching 38%.

The report believes this "delay rather than cancellation" pattern has three ancillary benefits: a smoother rebalancing of supply and demand in the data center supply chain, improved prospects in the credit market, and rising prices for computing power leasing. Against the backdrop of limited electricity supply in the U.S., projects that can secure electricity become scarcer, and hyperscale cloud providers will be prioritized for resources.

On a global level, the total capacity for 2028 has been revised down from 224GW to 219GW, while remaining unchanged at 271GW for 2029 and beyond. The power gap has expanded from 38GW to 44GW, accounting for 38% of the required electricity.

Morgan Stanley divides "Buy the Drop" into three baskets

The report has a clear framework: three types of targets, each with its own rationale.

First basket: hyperscale cloud providers. The report is optimistic about companies with scale, capital procurement capabilities, and LLM technology iteration capabilities. Brian Nowak gives an "overweight" rating to META, GOOGL, and AMZN, while Adam Wood increases MSFT. The reason is that these companies have geographically diversified layouts, do not rely on a single state, and their capital expenditure guidance remains unchanged.

Second basket: selected data center supply chain. The report recommends buying power and infrastructure companies that have fallen below value during pullbacks, while being wary of potential inventory build-up risks in the supply chain in 2028. High conviction targets include BE (Bloom Energy), SEI (Solaris), VST (Vistra), TLN (Talen Energy), and CEG (Constellation Energy). The report believes that even if data center growth is below expectations, the risk-reward of these companies remains favorable. In terms of colocation data centers, EQIX (Equinix) has limited exposure in Texas (with only 9 operational centers out of about 280 globally) and may benefit from higher pricing.

Third basket: software enablers and AI adopters. The report anticipates that value creation is shifting from semiconductors and infrastructure to software and application layers. Cybersecurity and infrastructure software are preferred entry points, with overweight targets including PANW (Palo Alto Networks) and CRWD (CrowdStrike). In the SaaS sector, it suggests adding positions in SNOW (Snowflake), DDOG (Datadog), and MDB (MongoDB).

The semiconductor drop is deep enough, the report says it can be picked up

The report used a chart to illustrate the issue: as of November 6, the median decline of semiconductor stocks covered by Morgan Stanley has fallen by 53% from their 52-week high, with an average drop of 48%. The report believes this presents a buying opportunity because the core pillar of AI spending (demand) has not been damaged, and token usage continues to grow exponentially.

Compute chips are the report's highest conviction sub-sector. NVDA and AVGO are both around 9 times CY28 P/E, with NVDA being the preferred choice for three reasons: a broader customer base than AMD/ASIC, the ability to grow by improving revenue per watt even in a power-constrained environment (the next-gen Feynman architecture is expected to boost revenue per GW from Rubin's approximately $40 billion to over $50 billion), and a leading position in the tokens/watt metric.

In storage, the preference is for MU and SNDK. The report specifically mentions SanDisk, which, as of the closing price on November 6 at $835, could repurchase over 50% of its circulating shares by the end of 2028 if the buyback scale is around 50% of revenue.

AI adoption has moved from experimentation to quantifiable returns

The report cites an analysis by AlphaWise of around 18,000 earnings reports and conference call transcripts. 44% of companies in the S&P 500 mentioned at least one quantifiable AI benefit in the past 12 months, up from 32% a year ago. The percentage of companies citing cost savings or AI-driven revenue growth rose from 16% to 26%.

The financial industry is leading the way: 67% of financial companies mentioned tangible improvements brought by AI, compared to only 42% a year ago. The report suggests broadening investment perspectives, shifting from AI builders to AI adopters, including banks, insurance, and consumer companies.

Risk Warning

This article is a整理 and interpretation of third-party brokerage research reports by Chao Xiang Research. The ratings, target prices, earnings forecasts, and related judgments quoted in the article represent the views of the respective brokerage analysts and do not represent the views of Chao Xiang Research, nor do they constitute any investment advice.

The market has risks, and decisions should be made independently. This article should not be regarded as the basis for buying or selling any securities.