Goldman Sachs Analysts Signal a Structural Shift in the AI Trade
TREE NEWS reports: In a newly released episode of Goldman Sachs’ The Macro Call podcast, Chief U.S. Equity Strategist Ben Schneider and TMT sector specialist Pete Callahan outlined a significant evolution in how markets are pricing the artificial intelligence theme. The central thesis: after four years of an AI trade dominated by semiconductor capital expenditure beneficiaries, investor attention is migrating up the technology stack toward the application layer — what Callahan termed the “inference economy” and the emerging race for personal AI agents.
From Selling Shovels to Selling Water
Callahan noted that the market has formed a healthy consensus around 2027 capital expenditure plans, providing roughly 12 to 15 months of visibility. But that very visibility creates a problem: what happens after 2027?
“That’s why you’re seeing multiple compression in the semiconductor space,” Callahan said. “The market is asking: is this earnings power sustainable? Are companies over-earning or under-earning? And what does the roadmap beyond 2027 look like?”
This uncertainty is pushing capital toward companies that help enterprises and consumers deploy AI — cybersecurity, data infrastructure, and tooling companies positioned to capture the “inference economy.” Callahan was careful to frame this not as a zero-sum rotation. “I think it’s more of an ‘and’ than an ‘or,'” he said. “Semiconductors are still within striking distance of all-time highs. The market has found room for both.”
The Next Narrative: Personal Agents
Perhaps the most provocative observation came during the third-quarter earnings preview. “The market is desperately hungry for the next story,” Callahan said. “Coding AI is no longer a secret — it’s already widely known as a major AI application. So how big and how fast can the personal agent story become? And the knock-on effects across the entire consumer supply chain — I think that will be fascinating to watch.”
He emphasized that during the upcoming earnings season, any signals from S&P 500 companies regarding cost savings, new revenue streams, new products, or new business formation tied to AI will be critical in sustaining the narrative.
Market Breadth: Narrowest Since the Dot-Com Bubble
Schneider addressed a topic that has increasingly worried investors: extreme market narrowness. “The median S&P stock is more than 15% below its high,” he said. “Our preferred breadth measure is the narrowest since the internet bubble.”
However, Schneider pushed back on the assumption that this must resolve through a broad sell-off. “Most investors think back to 2000 and assume it must end in a decline. But historically, narrow breadth more often resolves through catch-up rallies,” he said. His advice: own a bit of everything — maintain exposure to AI themes while also looking at beaten-down sectors like consumer experience stocks trading at low multiples.
Earnings Growth and the 2027 Slowdown
Schneider flagged a key forward-looking concern: consensus expects roughly 30% earnings growth, but that pace is expected to decelerate entering 2027. “Not only because of macro pressures, but because AI capital expenditure, while still growing, won’t maintain the same rate of growth,” he explained. The question for markets is how gentle or steep that deceleration will be.
Key Takeaways for Investors
- AI trade is broadening: Watch application-layer names — cybersecurity, data infrastructure, and AI deployment tools — alongside semiconductor leaders.
- Personal agents are the next catalyst: Any product breakthroughs in consumer-facing AI agents could drive the next leg of the trade.
- Breadth resolution likely upward: Historical precedent suggests narrow markets typically resolve through catch-up rather than collapse.
- Earnings season is critical: Q3 results from hyperscalers on ROIC and semiconductor guidance extending to 2028 will be closely scrutinized.
- Year-end base case is bullish: Strong earnings, light positioning, and reasonable valuations support a higher market into year-end.
