Morgan Stanley Backs AI Barbell as Nvidia, Broadcom Lead Chip Legs

Morgan Stanley told clients on Thursday to run a barbell in artificial intelligence: keep the chipmakers and infrastructure names at the core of the build-out, and start adding companies that can show measurable gains from putting the technology to work. The call, led by analyst Stephen Byrd, favours Nvidia and Broadcom on the hardware leg while flagging Microsoft, Snowflake, Datadog, Cloudflare and Dynatrace in software. Home Depot traded at $292.18, down 1.53%, and Airbnb at $151.39, up 1.21%, as of 23:31 UTC today.

Context — why the AI trade needs a second leg now

Chip and infrastructure stocks have broadly outperformed the S&P 500 since the AI wave began, and the analysts still see room for further gains. Their argument for staying invested is scarcity: demand for computing power should outstrip supply for years to come.

The report attributes that shortfall to forces outside the semiconductor industry itself. Energy shortages, local backlash against data-centre construction and politics have slowed the build-out, in the team's view. Scarcity of that kind lengthens the window in which hardware suppliers can earn outsized profit.

That reasoning marks a break from earlier technology cycles, where market leadership typically rotated away from one sector as a trend matured. The analysts argue the usual handover has been delayed rather than cancelled.

The note does not tell investors to sell hardware. It tells them to stop relying on hardware alone, freeing capital for names further down the adoption curve. Software and non-technology adopters absorb that flow rather than the money leaving the AI theme.

Investors have been weighing how long the concentrated rally in AI hardware can run. A strategy that keeps enablers and adds adopters gives them a way to stay exposed while spreading bets across the wider market.

Data — what the note actually quantifies

Semiconductor analyst Joe Moore said demand from data-centre customers remains exceptionally strong, and named Nvidia and Broadcom among his preferred picks on the back of spending by AI model developers. The report does not disclose target prices, position sizes or revenue forecasts for either company.

Adam Wood, covering software, favours infrastructure names: Microsoft, Snowflake, Datadog, Cloudflare and Dynatrace. These companies supply the underlying systems on which AI applications run, a layer that sits between the chipmakers and the end users.

The most concrete adoption numbers come from two non-technology names. Healthcare firm iRhythm Holdings says AI will cut the time clinicians spend reviewing medical records by almost half. Airbnb says its AI assistant now resolves 40% of issues without a human agent and has sharply reduced booking times.

The bank also holds positive ratings on Home Depot, Procter and Gamble, GE Aerospace and Coca-Cola. In Tuesday's session, Home Depot traded between $291.09 and $299.10, while Airbnb ranged from $149.68 to $153.97, a spread of roughly 2.9% between the low and the high.

Company	AI claim in the note	Market data
Airbnb	AI assistant resolves 40% of issues without a human	$151.39, +1.21%
Home Depot	Positive rating, no metric given	$292.18, -1.53%
iRhythm	Record review time cut by almost half	Not quoted

Analysis — where the second leg of the trade points

Morgan Stanley points to new AI entrants across healthcare, real estate, autos, transportation and consumer services. The analysts argue the rapid scaling of large language model capabilities signals a broad opportunity for value creation, not one confined to the companies building the models.

That framing matters for sector selection. Model developers drive chip demand, so they carry the direct exposure to any rules governing frontier systems. Adopters and infrastructure software sit further from that regulatory line, which is why the note treats them as the less contested side of the barbell.

Regulation risk now sits alongside the bottlenecks the team already counted. Tuesday's White House meeting with technology chief executives on AI is the near-term test of how that risk prices. The report gives no read on what was discussed or agreed.

The weakest point in the argument is the scarcity thesis itself. Energy shortages, local opposition and politics can ease as quickly as they tighten, and any of the three resolving would shorten the profit window the analysts are underwriting for chip suppliers. That is the assumption to monitor.

Positioning follows the logic: money rotating out of crowded AI hardware positions moves into software and non-technology adopters rather than out of the theme. That flow, if it materialises, supports a broadening of US equity gains beyond the megacaps.

Outlook — what to watch next

Tuesday's White House meeting with technology chief executives on AI is the first catalyst. Any signal on how frontier models will be treated sets the relative risk between model builders and the adopters the note favours.

On the hardware leg, watch data-centre demand commentary from Nvidia and Broadcom and any update on the energy and permitting constraints the analysts cite. Those constraints are the load-bearing assumption behind the scarcity call.

For the adopter leg, Airbnb's disclosure that its assistant resolves 40% of issues without a human is the kind of metric that gives the thesis a number. Further disclosures of the same type from Home Depot, Procter and Gamble, GE Aerospace or Coca-Cola would validate the broadening claim.

Home Depot's session range of $291.09 to $299.10 is worth holding against its $292.18 close, since the stock finished near the bottom of that band. Airbnb closed at $151.39 against a high of $153.97. The report names no support or resistance levels, so none should be inferred from these prints.

Frequently Asked Questions

What does Morgan Stanley's AI barbell strategy mean for retail investors?

It means holding two types of exposure rather than one. The first leg is chip and infrastructure names such as Nvidia and Broadcom, where the analysts expect scarcity to extend profitability. The second is software and non-technology adopters. The intent is to stay invested in AI while reducing reliance on a single crowded group of stocks.

Why does Morgan Stanley expect hardware scarcity to last for years?

The analysts point to energy shortages, local backlash against data-centre construction and politics as the forces slowing the build-out. Those constraints keep supply of computing power behind demand, which in their view gives chip suppliers a longer window to profit than a typical technology cycle would allow.

What numbers back the claim that non-technology companies are gaining from AI?

Two disclosures carry the argument. iRhythm Holdings says AI will cut clinician time spent reviewing medical records by almost half. Airbnb says its AI assistant resolves 40% of issues without a human agent and has sharply reduced booking times. Those are the specific figures in the note.

Bottom Line

Morgan Stanley wants investors to keep the chip winners but stop depending on them alone.

Disclaimer: This article is for informational purposes only and does not constitute investment advice. CFD trading carries high risk of capital loss.