Morgan Stanley: AI Investment Should Adopt a "Barbell" Strategy, Balancing Infrastructure and Emerging Adopters
nashnova research
What is the "barbell" strategy, and why now?
A barbell strategy means loading both ends of a portfolio and skipping the middle — here, AI infrastructure on one side and AI adopters on the other.
This means → Morgan Stanley sees no clear winner yet in the tug-of-war between "the companies that build AI" and "the companies that use it," so it says own both.
The report is led by analyst Stephen Byrd.
The infrastructure end — why still hold it?
AI infrastructure buildout is slowing — local opposition, energy shortages, and political friction are all dragging on data-center construction.
In plain terms = slower building extends the supply shortage, which lets hardware vendors earn for longer than a normal cycle would allow.
But the report stresses "be selective, not blanket" — only the names with the most certain demand justify a position.
Chips and software infrastructure — who made the list?
Analyst Joe Moore calls data-center demand "exceptionally strong" and recommends Nvidia and Broadcom, both benefiting from sustained spending by AI-model developers.
The report's call: "Compute demand will significantly exceed supply for years to come."
On software infrastructure, analyst Adam Wood recommends Microsoft, Snowflake, Datadog, Cloudflare, and Dynatrace — providers of cloud storage and other foundational systems that AI applications run on.
The adopter end — who has already put numbers on the board?
iRhythm Holdings (healthcare): AI is set to cut the time clinicians spend reviewing patient records by nearly half.
Airbnb: its AI assistant now resolves 40% of issues without human intervention and has significantly shortened booking times.
This means → these companies are not telling an AI story — they are showing quantified efficiency gains, which is exactly why Morgan Stanley says it is time to start adding adopter exposure.
Beyond tech — which other sectors got the nod?
Morgan Stanley holds positive ratings on AI adopters across several non-tech industries, including Home Depot, Procter & Gamble, GE Aerospace, and Coca-Cola.
The report argues that the exponential scaling of large-language-model capabilities makes AI's impact on value creation "broad and profound."
In plain terms = the AI profit opportunity is spreading from Silicon Valley into traditional industries — retail, consumer goods, and industrials are all in range.
Where is the risk in this strategy?
Whether both ends of the barbell can pay off at the same time hinges on one unresolved question: will AI market leadership migrate from the infrastructure layer to the application layer?
If the shift happens fast, infrastructure names may peak first; if it stalls, the adopter thesis cannot deliver.
This reflects Morgan Stanley's own acknowledgment: whether adopters can keep quantifying and delivering AI gains is the key validation point for this strategy — and there is no answer yet.
市场有风险，内容仅供研究参考，不构成投资建议。