Morgan Stanley Research Report Analysis: AI Computing Power Gap Continues to Widen, Five Main Themes Present Investment Window

Written by: Rita

Recently, the AI infrastructure sector has experienced a significant correction, with NVIDIA retracing approximately 12% from its recent high, and targets such as Broadcom and Marvell declining simultaneously. Market skepticism regarding the sustainability of AI capital expenditure continues to intensify. Morgan Stanley released a global thematic strategy report on July 27, proposing a contrarian judgment: the computing power supply-demand gap will further widen, this adjustment presents allocation opportunities, and outlined Five Major Investment Directions.

AI Infrastructure Bottlenecks

Power access conditions, construction capabilities, and land resources constitute core industry barriers. The report defines companies possessing such resource endowments as "Power Infrastructure Providers," with representative targets including Talen Energy, Vistra, Bloom Energy, as well as Hut 8, Cipher Mining, TeraWulf, Riot Platforms, etc. Scarce land indicators and grid access permissions are key upstream constraints for data center construction and also the most difficult link to replicate in the AI industry chain.

Computing Power Manufacturing Ecosystem

The track covers semiconductor manufacturing, equipment, materials, and packaging segments. NVIDIA remains the preferred target for institutional semiconductor portfolios, with Broadcom and Micron also included in the recommendation list. Morgan Stanley points out that NVIDIA is currently significantly undervalued in the semiconductor sector.

China AI Solutions

Institutions judge that the development speed of domestic AI enterprises exceeds general market expectations. ByteDance may continue to increase investment, with capital expenditure expected to rise to $80 billion; Alibaba and Tencent are also continuously increasing AI business investment. The domestic AI chip self-sufficiency rate is expected to increase from 42% in 2025 to 70% in 2030, and the scale of domestic AI model deployment in overseas markets is steadily rising.

Energy Security

Energy storage, grid equipment, and natural gas power generation are key directions. Affected by the shortage of transformer supply in the United States, South Korean power equipment manufacturers are welcoming development opportunities, with HD Hyundai Electric and LS Electric as core targets. Institutions predict that the share of South Korean enterprises in the U.S. transformer import market will continue to rise.

Hyperscalers

Meta, Google, Microsoft, and Amazon are on the list. The market generally underestimates the ROI corresponding to the AI capital expenditure input of top cloud vendors; the moat of existing business plus AI commercialization potential makes fundamentals better than expected. Meta boasts 3.5 billion daily active users, and there is still room for optimization in the AI-enabled ad delivery and content recommendation systems.

Supply-Demand Contradictions and Core Market Divergences

The underlying logic of the five main lines is built on the judgment of computing power supply-demand imbalance. Google executives publicly stated that the company plans to double its computing power scale every half year, expanding 1000 times within 4 to 5 years. In contrast, NVIDIA's computing power shipment CAGR is about 140%, forming a huge divergence between demand expansion magnitude and supply growth rate, which is difficult to repair in the short term.

Regarding the two major concerns most cared about by the market, the report also provides clear arguments.

On one hand, cases of enterprises controlling AI expenditures are increasing; Uber exhausted its 2026 AI budget as early as April, and Meta has also set monthly usage limits. However, institutions define this phenomenon as short-term integration friction. Data shows that enterprise users consume about $11 worth of Tokens per month on average, and a single project AI transformation can save costs of $55, with an ROI exceeding 20 times. Budget constraints belong to phased adjustments and do not represent the arrival of a demand inflection point.

On the other hand, after the release of Kimi K3, the market once again discussed the "DeepSeek Moment," worrying that the narrowing gap between Chinese and U.S. AI technology would suppress computing power demand. Morgan Stanley holds the opposite view: improvements in model efficiency will lower unit Token costs, drive overall usage scale upward, and ultimately generate greater computing power demand, aligning with the Jevons Paradox in economics.

Supply-Side Bottlenecks and Layout Strategies

Demand-side growth certainty is strong, but supply-side constraints are becoming increasingly prominent. Institutions calculate that the U.S. data center power supply-demand gap will reach 38GW from 2026 to 2028, while the scale of projects currently under construction and capable of smooth grid connection is only 15GW each. Even adding transition power supply solutions such as gas turbines and fuel cells, there remains a persistent gap of 1GW to 11GW. Labor shortages are similarly difficult to alleviate quickly, with insufficient supply of skilled technicians. At the policy level, 14 states across the U.S. are brewing data center control bills, and New York State has implemented a one-year suspension policy.

The report also shares a set of backtesting conclusions: adding positions to 150% when the AI thematic index retraces 5%, and adding to 200% when it retraces 15%, with a holding period set at 3 months, long-term returns are superior to static holding. Volatility in the AI infrastructure sector is mostly driven by sentiment and does not represent a reversal in fundamental trends.

Comprehensive view, the core logic of this research report can be summarized as: AI computing power demand is rising exponentially, while supply is constrained by various physical conditions and can only expand linearly, causing the supply-demand gap to continue widening. Short-term stock price oscillations are more likely emotional noise, while computing power supply-demand imbalance is the medium-to-long-term main line.

Disclaimer

This article is a compilation and interpretation by TechFlow Research of a third-party broker research report (Morgan Stanley, July 27, 2026). The ratings, target prices, earnings forecasts, and related judgments cited in the text are the views of the broker's analysts, represent only the position of their affiliated institution, do not represent the views of TechFlow Research, and do not constitute any investment advice.

The market carries risks; decisions must be made independently. This article should not be used as a basis for buying or selling any securities.