AnswersPublished20 sources
Why Some AI Infrastructure Stocks Could Keep Growing as Spending Cools
Morgan Stanley’s thesis is that suppliers with bottleneck driven pricing power or room to gain share could still get earnings upgrades as AI capex growth slows—from nearly 100% in 2026 to about 60% in 2027 and 14% in... The forecasts still point to rising spending: about $980 billion in 2026, $1.58 trillion in 2027...
Published byEdited with GPT-6 LunaImages generated with GPT Image 2
Research answer
Key takeaways
- Morgan Stanley’s thesis is that suppliers with bottleneck driven pricing power or room to gain share could still get earnings upgrades as AI capex growth slows—from nearly 100% in 2026 to about 60% in 2027 and 14% in...
- The forecasts still point to rising spending: about $980 billion in 2026, $1.58 trillion in 2027 and roughly $1.8 trillion in 2028.
- A partial excerpt names Caterpillar, Cummins, Forgent Power Solutions and GE Vernova; the available sources do not verify the full chip, memory, server, storage, optics and cooling lineup or South Korean HBM suppliers.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: Which stocks named in Morgan Stanley’s 24-stock AI selection could sustain earnings growth as hyperscaler and neocloud infrastructure spendi. Article summary: Morgan Stanley’s thesis is selective resilience: suppliers with scarce products, pricing power or opportunities to gain market share could keep receiving revenue and earnings upgrades even as AI-infrastructure spending g. Topic tags: general, general web, user generated, news. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts w
Morgan Stanley’s case for 24 AI infrastructure stocks is about relative resilience, not immunity to a slowdown. The screen favors companies whose products are scarce enough to command pricing power, or that can take market share as customers build out data centers. That could support further revenue and earnings upgrades even if hyperscaler and neocloud spending grows more slowly. 5
33
The important caveat: the available reporting provides only a partial roster. It does not establish all 24 companies or confirm which South Korean high-bandwidth-memory (HBM) suppliers are included.
What the available stock list confirms
A published excerpt of the 24-stock selection names Caterpillar, Cummins, Forgent Power Solutions and GE Vernova in its power-related entries. 34 That is enough to show that the selection reaches beyond chipmakers into the equipment and power constraints involved in data-center construction, but not enough to reconstruct the full list or reliably assign every stock to a sector.
The available sources do not confirm the selection’s specific picks in memory, servers, storage, optics, cooling or electrical equipment, nor do they identify South Korean HBM suppliers as members. Naming companies in those categories would go beyond the evidence provided.
Related Morgan Stanley reporting gives AI-infrastructure exposure estimates for Nvidia, Vertiv and Broadcom: 91.7%, 90% and 81% of projected 2027 revenue, respectively. Those figures describe estimated revenue exposure—not confirmed membership in this separate 24-stock screen. 7
Why slower growth may still leave room for earnings upgrades
The screen’s logic is that capex growth and supplier growth are not the same thing. A company may keep growing if customers have few alternatives for a constrained product, are willing to pay to bring capacity online sooner, or shift more orders to that supplier. Morgan Stanley strategist Nicholas Lentini’s reported rationale points to persistent bottlenecks during the data-center buildout and the value customers place on time-to-market. 34
That makes the thesis conditional: pricing power and market-share gains could support earnings revisions, but neither is guaranteed. A slowdown in spending growth can still affect suppliers, especially if a bottleneck eases or buyers delay deployment.
Capex growth is slowing, but the forecast is not for falling spending
Morgan Stanley’s reported estimates put AI infrastructure investment by hyperscalers and neocloud providers at roughly $980 billion in 2026, $1.58 trillion in 2027 and approximately $1.8 trillion in 2028. 3
4 The reported growth rate is expected to ease from nearly 100% in 2026 to about 60% in 2027 and 14% in 2028.
3
5
The distinction matters: these forecasts describe a deceleration in the rate of growth, not a decline in the total amount spent. The 2028 figure is still higher than the 2027 estimate. They remain forecasts, however, and should not be read as guaranteed spending levels.
The $1.58 trillion 2027 estimate includes lease-related spending and is reported to be about 14% above the cash-capex measure. 2
9 A separate report puts data-center leases at roughly $195 billion and says Morgan Stanley refers to such spending as “synthetic capex”; that exact lease figure is not independently corroborated in the other excerpts available here.
11
Exposure and bottlenecks cut both ways
Nvidia’s estimated 91.7% exposure means Morgan Stanley expects a large share of its projected 2027 revenue to be tied to AI infrastructure demand. It is a measure of revenue dependence, not the company’s share of total infrastructure spending. 4
7 High exposure can benefit a supplier while investment continues, but it also makes sustained demand weakness more consequential.
Power availability is another constraint. Reuters reports that Morgan Stanley sees Nvidia and Broadcom as relatively insulated from the U.S. data-center power crunch, while deployment delays could affect memory, optical and other secondary-component suppliers. The same report describes an estimated 32-gigawatt net U.S. power shortfall through 2028; that figure is not the 57-gigawatt estimate mentioned in some versions of the question. 17
If customers have ordered components ahead of actual deployments, a slowdown could leave suppliers facing weaker orders or pressure to lower prices. That is a risk scenario, not a quantified Morgan Stanley forecast in the available excerpts. For memory, optics and other components, delayed deployment could also affect demand even when planned data-center investment remains high. 17
The practical test for investors
The central question is whether a supplier’s advantage comes from durable scarcity or market-share gains—or from temporary shortage pricing that could fade. Morgan Stanley’s screen identifies the former as a reason some companies might continue to receive earnings upgrades as capex growth cools. 5
33
But the evidence here does not support a complete stock-by-stock comparison, a verified list of Korean HBM suppliers, or a conclusion that every named company would maintain margins if spending levels off. The thesis is selective resilience, not a promise that all AI infrastructure stocks—or their earnings—will be insulated from a spending slowdown.
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "Why Some AI Infrastructure Stocks Could Keep Growing as Spending Cools"?
Morgan Stanley’s thesis is that suppliers with bottleneck driven pricing power or room to gain share could still get earnings upgrades as AI capex growth slows—from nearly 100% in 2026 to about 60% in 2027 and 14% in...
What are the key points to validate first?
Morgan Stanley’s thesis is that suppliers with bottleneck driven pricing power or room to gain share could still get earnings upgrades as AI capex growth slows—from nearly 100% in 2026 to about 60% in 2027 and 14% in... The forecasts still point to rising spending: about $980 billion in 2026, $1.58 trillion in 2027 and roughly $1.8 trillion in 2028.
What should I do next in practice?
A partial excerpt names Caterpillar, Cummins, Forgent Power Solutions and GE Vernova; the available sources do not verify the full chip, memory, server, storage, optics and cooling lineup or South Korean HBM suppliers.
Sources
finance.yahoo.comMorgan Stanley says these 24 AI infrastructure stocks are ...
uk.finance.yahoo.comMorgan Stanley says these 24 AI infrastructure stocks are ...
businessinsider.tw摩根士丹利：這24檔人工智慧基礎設施股即使資本支出放緩也能撐住 - Bu...
hellomarvisaitoday.comMorgan Stanley: 91.7% of Nvidia 2027 revenue is AI-infra exposure