Wall Road is drawing a tough line between the shares that profit from synthetic intelligence and those who get buried by it. Goldman Sachs has carried out the maths, and the outcomes are reshaping how traders place their portfolios proper now.
The financial institution’s strategists have mapped out a rotation that’s firmly underway in February 2026. Asset-heavy corporations with factories, bodily provide chains and specialised tools are pulling forward. Software program names tied to labor-intensive workflows are absorbing the ache of a deliberate and deepening sell-off.
For on a regular basis traders, the query is not whether or not synthetic intelligence will disrupt markets. It already is. The actual query is which shares land on the fitting aspect of that divide, and Goldman has a particular framework to reply it.
Two metrics Goldman makes use of to rank AI disruption Danger
Goldman’s method depends on two particular screens. The primary is labor value as a share of income.
Goldman’s company-level metric estimates publicity to AI automation by analyzing job features and overlaying them with task-level measures of AI functionality, then combining that estimate with every agency’s labor-cost-to-revenue ratio. Software program, skilled companies, banks and media rank as essentially the most at-risk sectors by this measure.
The second display is bodily asset density. Companies anchored to factories, distribution networks or precision manufacturing tools carry a pure moat. These operations take years to duplicate and no AI mannequin can shortcut that timeline.
Collectively, the 2 measures separate genuinely sturdy companies from those who look secure however carry actual automation threat beneath.
The HALO Commerce: asset-heavy shares profitable the rotation
Goldman’s strategists have named this dynamic the “HALO effect,” standing for heavy property and low obsolescence. In response to Goldman’s shopper word, authored by strategists together with Guillaume Jaisson and Peter Oppenheimer, the financial institution’s basket of capital-intensive shares has outperformed a capital-light group by about 35% for the reason that begin of 2025.
Grids, pipelines, utilities, transport infrastructure and demanding equipment are all cited as prime examples of HALO companies.
Photograph by Michael M&interval; Santiago on Getty Photos
Jaisson wrote that markets are rewarding capability, networks, infrastructure and engineering complexity, particularly property which are expensive to duplicate and fewer uncovered to technological obsolescence. Utilities, fundamental sources and vitality are drawing the heaviest inflows as traders transfer from digital capital fashions into companies with tangible manufacturing capability.
European capital-intensive shares Goldman favors
- ASML (ASML): Holds a monopoly on excessive ultraviolet chip lithography tools with no credible substitute wherever on this planet.
- Airbus: Industrial plane meeting calls for years of precision engineering experience and world provide chain depth that synthetic intelligence can’t replicate.
- Safran: Embedded in long-cycle aerospace contracts throughout world aviation applications, giving it unusually sturdy income visibility.
- LVMH: Luxurious model dominance backed by bodily craftsmanship and heritage creates pricing energy no algorithm can manufacture.
- Air Liquide: Industrial fuel infrastructure constructed throughout a number of continents over a long time will not be a enterprise mannequin any AI startup can disrupt in a single day.
The hyperscalers are themselves changing into capital-intensive performs. Amazon (AMZN), Microsoft (MSFT), Alphabet (GOOGL), Meta (META) and Oracle (ORCL) are on observe to spend $1.5 trillion constructing AI infrastructure between 2023 and 2026, roughly twice what they invested throughout their whole historical past earlier than 2022.
In 2026 alone, their capex is on observe to exceed $650 billion.
Software program break up: Who survives and who does not
Goldman has been deliberate about distinguishing the software program corporations that can climate this shift from these that won’t.
The financial institution launched a basket that goes lengthy on corporations synthetic intelligence can’t simply displace, and shorts these whose core workflows are most susceptible to automation.
Extra AI Shares:
- Morgan Stanley units jaw-dropping Micron worth goal after occasion
- Financial institution of America updates Palantir inventory forecast after personal assembly
- Morgan Stanley drops eye-popping Broadcom worth goal
The sell-off has been extreme and regular. Goldman’s software program basket clocked its seventh consecutive each day decline in early February 2026, bringing its year-to-date loss to 19%. The rout has bled into broader tech measures, dragging the Nasdaq 100 down 1.4% up to now in 2026.
Software program shares Goldman thinks are winners
- Microsoft (MSFT): Cloud and AI infrastructure that just about each massive enterprise depends upon, with switching prices that make displacement extraordinarily troublesome.
- Oracle (ORCL): Database methods so deeply embedded in company workflows that changing them carries huge value and operational threat.
- CrowdStrike (CRWD): Cybersecurity infrastructure that turns into extra crucial, not much less, as AI instruments multiply the assault floor for unhealthy actors.
- Palo Alto Networks (PANW): Community safety with deep regulatory entrenchment throughout monetary companies, healthcare and authorities.
- Cloudflare (NET): Web infrastructure dealing with an growing share of worldwide AI-driven site visitors at scale.
Software program shares Goldman thinks are losers
- Salesforce (CRM): Core workflow automation that AI brokers are starting to duplicate internally with out a paid subscription.
- Accenture (ACN): Consulting and outsourcing companies that AI brokers are absorbing at a tempo threatening the normal billing mannequin.
- DocuSign (DOCU): Doc administration workflows that generative AI now handles from drafting by signature in a single pipeline.
- Monday.com (MNDY): Undertaking coordination instruments going through direct stress from autonomous AI assistants that deal with scheduling and process administration natively.
- Duolingo (DUOL): Language studying platform competing immediately towards AI tutors that personalize classes at a fraction of the price.
What the broader market information reveals
The size of promoting exercise is hanging. Goldman’s prime brokerage reported that notional brief promoting throughout single shares within the week of Jan. 30 to Feb. 5 was the largest on document within the financial institution’s information going again to 2016, with brief gross sales outpacing lengthy buys by a ratio of two to at least one.
Goldman CEO David Solomon acknowledged the depth of the strikes. Talking at a UBS convention in Key Biscayne, Florida, Solomon advised attendees that the sell-off narrative over the prior week had been “a little bit too broad,” and that there could be clear winners and losers amongst software program corporations slightly than a wholesale collapse.
The rotation is actual, the info backs it up, and Goldman has positioned shopper cash on either side of the commerce. For traders nonetheless checking out the place to face as synthetic intelligence rewrites the aggressive panorama, this can be a framework price understanding intently.
Associated: Goldman Sachs resets Microsoft inventory forecast
