Has the AI sector of US stocks ushered in a good opportunity to bottom out after being sold off? Damo is optimistic that tech giants such as NVDA.US (NVDA.US) will maintain capital expenditure forecasts that may reach 1.6 trillion US dollars in 2027

The Zhitong Finance App learned that Morgan Stanley said in the latest global research report released on November 8 that after the US midterm elections, market concerns about the prospects for data center construction triggered drastic adjustments in artificial intelligence (AI) related stocks, but the bank believes that the market has overreacted, and the long-term AI investment logic has not fundamentally changed. Morgan Stanley advises investors to adopt a selective bargain buying strategy, focusing on large technology companies, some semiconductor and data center infrastructure companies whose valuations have declined markedly, and software companies that are expected to benefit from the commercial application of AI.

Morgan Stanley believes that policy changes brought about by the US midterm elections will mainly affect the construction time of some data center projects rather than the overall AI investment requirements. Although the newly elected governor may push for more stringent data center approvals and construction requirements, projects already in the construction phase are not expected to be significantly affected.

The bank anticipates that about 20% of the US data center projects scheduled for 2028 may be postponed to 2029, involving about 5 gigawatts of IT load capacity, which is equivalent to about 10% of related projects worldwide. Affected by this, the new capacity of US data centers is expected to drop by about 4% year-on-year in 2028, then rebound by about 38% in 2029.

However, Morgan Stanley emphasized that the project extension did not equal the cancellation of the investment. On the contrary, the lengthening of the construction cycle may make AI capital expenditure increase more sustainable, help mitigate supply and demand imbalances in the supply chain, improve the financing environment, and support computing power rental prices.

Maintain AI capital expenditure forecasts of $1.6 trillion and $1.8 trillion in 2027 and 2028, respectively

Despite adjusting the construction schedule for some data center projects, Morgan Stanley currently maintains forecasts for the overall capital expenditure of large cloud computing and technology companies, which are expected to reach approximately US$1.6 trillion and US$1.8 trillion in 2027 and 2028, respectively.

The bank believes that large technology companies will still expand their computing power infrastructure as much as possible to support the promotion of big language models and AI applications to billions of users. As AI model capabilities continue to improve, the demand for computing resources for related products will remain strong.

However, Morgan Stanley also warned that if data center approval delays continue to expand, capital expenditure forecasts for 2028 may face downward pressure.

From the perspective of supply and demand, electricity supply is still one of the main bottlenecks in AI infrastructure expansion. According to Morgan Stanley's model, the global capacity gap for data centers with power supply conditions is expected to reach 44 gigawatts in 2028, an increase of about 15% over previous predictions. This means that even if construction of some projects is delayed, data centers that can receive electricity in a timely manner and put into operation still have high scarce value.

At the same time, Morgan Stanley believes that stricter construction approvals and higher project costs may further strengthen the competitive advantage of large technology companies. Compared to smaller AI infrastructure operators, companies such as Amazon, Google, and Meta have more sufficient capital, more mature technology systems, and stronger project execution capabilities.

Focus on Amazon, Google, Meta, and Microsoft; Nvidia is still the core beneficiary

In terms of specific investment directions, Morgan Stanley is first optimistic about large technology companies with advantages in scale, capital, and technology.

The bank gave Amazon (AMZN.US), Google's parent company Alphabet (GOOGL.US), Meta (META.US), and Microsoft (MSFT.US) a “plus” rating, believing that these companies can use their capital strength and AI model development capabilities to continue to expand their market share in an environment where computing power resources are limited.

According to the target price listed in the report, Morgan Stanley gave Meta a target price of 820 US dollars, corresponding to the potential increase in the report's reference share price of about 35.5%; Alphabet's target price is 375 US dollars, with a potential increase of about 32.7%; Amazon's target price is 335 US dollars, with a potential increase of about 28.7%; and Microsoft's target price is 650 US dollars, with a potential increase of about 21.3%.

Furthermore, Morgan Stanley is also optimistic about Nvidia (NVDA.US), giving it an “gain” rating and a target price of $300. The corresponding report refers to the potential share price increase of about 60.7%.

The bank pointed out that although Nvidia is not a cloud computing giant, its chips and related technology can improve the AI computing power and commercial value generated per unit of electricity. As power and data center capacity are limited, the importance of improving computing power efficiency will increase further.

Morgan Stanley also anticipates that AI model developers will pay more attention to improving computational efficiency and achieve stronger AI capabilities with fewer computing resources through model distillation, architecture optimization, and improvement of training and inference processes.

The semiconductor and data center supply chain still has opportunities, but more careful selection of targets is needed

In addition to large technology companies, Morgan Stanley believes that recent sharp stock price adjustments have also created investment opportunities for some semiconductor and AI infrastructure companies.

In the semiconductor sector, the bank favors large-scale industry leaders with strong capital strength, including Nvidia, Broadcom (AVGO.US), TSM.US (TSM.US), MediaTek, and Samsung Electronics. Among them, Morgan Stanley gave Broadcom a target price of 505 US dollars, and the corresponding report referred to a potential increase of about 72% of the stock price.

The bank is also focusing on hardware sectors where valuations have declined recently, including hard drive manufacturers Seagate Technology (STX.US), Western Digital (WDC.US), and optical communication and network equipment companies Lumentum (LITE.US), Coherent (COHR.US), and Arista Networks (ANET.US).

However, Morgan Stanley warned that delays in data center construction projects may cause some upstream supply chain companies to face the risk of slowing order growth, increasing inventory, and falling profit margins in 2028. Therefore, the bank does not recommend that investors fully buy all AI infrastructure stocks that have been sold off, but rather emphasizes the need for screening based on valuation, supply and demand conditions, and corporate profitability.

In the data center and power infrastructure sector, Morgan Stanley focuses on companies such as Equinix (EQIX.US), TeraWulf (WULF.US), Hut 8 (HUT.US), Riot Platforms (RIOT.US), and Applied Digital (APLD.US).

Meanwhile, the bank is still optimistic about some companies that can provide power solutions for data centers, including Bloom Energy (BE.US), Vistra (VST.US), Talen Energy (TLN.US), and Constellation Energy (CEG.US).

Morgan Stanley believes that as some data center construction projects in the US face approval restrictions, companies that have ready-made power resources and can quickly deliver computing power infrastructure may further enhance their competitive advantage.

AI investment opportunities are spreading from chips and infrastructure to software and applications

Morgan Stanley also emphasized that the AI investment cycle is undergoing an important change: the market focus is expected to gradually shift from chip and infrastructure construction to software services and the actual application of AI technology.

The bank pointed out that in the past computing technology development cycle, return on investment was usually focused first on semiconductor companies, then expanded to infrastructure, and finally to software and services. Similar trends are likely to reappear as companies begin putting AI technology into actual business.

In the software sector, Morgan Stanley is optimistic about companies such as Microsoft, Snowflake (SNOW.US), Datadog (DDOG.US), MongoDB (MDB.US), Cloudflare (NET.US), and ServiceNow (NOW.US).

In the field of cybersecurity, the bank focuses on Palo Alto Networks (PANW.US), CrowdStrike (CRWD.US), Okta (OKTA.US), and Rubrik (RBRK.US).

Furthermore, Morgan Stanley believes that productivity gains brought about by AI applications are gradually spreading to traditional industries such as finance, healthcare, consumer, and industry. Companies such as Bank of America (BAC.US), Citigroup (C.US), UnitedHealth (UNH.US), Procter & Gamble (PG.US), and Visa (V.US) are also included in the list of companies benefiting from AI applications that it focuses on.