The next crack in the AI trade may show up in spending before it appears in revenue. Sam Rines, Macro Strategist at WisdomTree, says a slowdown in hyperscaler capital spending would likely hit semiconductors and infrastructure providers hardest, while companies using AI to improve existing businesses could prove more resilient. That puts Nvidia Corp (NVDA) and Meta Platforms, Inc. (META) on different sides of the cycle.

AI Spending Faces a New Test

Nvidia has been one of the clearest financial beneficiaries of the AI boom, supplying the chips that power the massive infrastructure buildout. But the market is increasingly asking a different question: How much longer can hyperscalers keep spending at extraordinary levels before investors demand clearer returns?

Rines says that test is already underway.

"Capex is tolerated for companies that can show the returns to it, but it is not appreciated for companies without the ability to clearly articulate the returns," he told MarketDash in an exclusive email interview.

That creates an important distinction between companies building AI infrastructure and those using AI inside established businesses.

Microsoft Corp (MSFT) is seeing its cloud business benefit from AI, while Alphabet Inc (GOOG) (GOOGL) is beginning to see Gemini contribute to revenue and improve YouTube advertising, according to Rines. Meta, meanwhile, is using AI to strengthen its core advertising business while expanding its AI product lineup.

Nvidia Has More AI-Capex Exposure

That difference matters if hyperscaler spending starts to slow.

"If AI spending starts producing weaker returns," Rines expects the impact to be felt "most by the semis and infrastructure providers." The AI implementers, he said, are likely to be "the least affected or even winners" in that scenario.

Nvidia's position makes the distinction particularly important. Its data-center business depends heavily on demand for the computing infrastructure needed to build and deploy AI systems. If Microsoft, Alphabet, Meta and other hyperscalers begin trimming or delaying infrastructure investments, that would create a different environment for Nvidia than for a company using AI to improve an existing revenue engine.

That does not mean a capex slowdown would automatically translate into a collapse in Nvidia's business. It means the semiconductor side of the AI trade could have greater direct exposure to a change in spending behavior.

Meta Could Be the Other Side

Rines sees Meta as particularly interesting as the AI story shifts from spending toward returns.

"Meta appears to be the most overlooked," he said, pointing to the company's Muse product and enterprise launch alongside AI's role in its advertising business. The argument is less about Meta becoming an AI infrastructure company and more about AI improving an already large business.

That distinction could become increasingly important if investors begin scrutinizing AI returns rather than simply rewarding bigger spending.

Rines says the first real warning sign would likely be a slowdown in hyperscaler capex.

For Nvidia investors, that makes capital-spending plans from Microsoft, Alphabet, Meta and other major AI buyers an increasingly important signal.

For Meta investors, the question may be different: whether AI can continue producing measurable gains in advertising, products and engagement even if the infrastructure spending cycle cools.