Higher Rates Split the Market — AI Stocks Get the Easy Money

The benchmark 10-year U.S. Treasury yield briefly hit 5.365% on Wednesday, its highest level since April 2002, and major U.S. equity indexes closed lower. The move landed in a week when the S&P 500 still returned to record highs, carried almost entirely by a narrow band of artificial intelligence stocks.

CNBC's Jim Cramer used his Wednesday Mad Money segment to argue that those two facts are not in conflict. They reflect the same dynamic: a credit market where AI-related companies borrow freely, while the rest of corporate America grows nervous every time a Treasury auction goes badly.

What's driving the split?

Wednesday's $39 billion 10-year Treasury note auction drew strong demand and pulled yields off their intraday highs. The relief was only partial. Investors spent the day waiting to see whether the sale would clear at a price the market could tolerate, and the answer mattered for stock prices in a way it had not for two decades.

Cramer framed that waiting game as a symptom of an unhealthy tape. "Any market where you need to wait to see the results of a Treasury auction is simply not as good as a market where you don't care about them," he said. "Every time you add a new variable into the equation, it makes owning stocks tougher."

Higher rates squeeze businesses that depend on borrowing, and Cramer named the sectors where the pressure shows up most clearly:

• Finance
• Housing
• Utilities
• Entertainment
• Retail
• Autos
• Industrials

These categories share a feature. Either their issuers, their customers, or both lean heavily on credit, and a 5.365% benchmark raises the cost of every refinancing and new issue they touch.

Which companies escape the squeeze?

AI-adjacent firms operate in a different credit environment, according to Cramer. He grouped the winners into four buckets:

• Data center builders
• Semiconductor manufacturers
• Power providers
• Cybersecurity firms

Lenders remain eager to finance their growth, Cramer said, which is why these names keep pulling indexes to new highs while the broader market worries about auction tail risk.

"They seem to be able to borrow at their leisure," Cramer said. "They're crowding out other borrowers with their demand for money."

The wider implication: the S&P 500's run to record highs this week is not a sign of broad strength. It is the visible surface of a market whose composition has tilted sharply toward AI infrastructure.

Why is SpaceX the example that matters?

The clearest test case, by Cramer's measure, is SpaceX. The Financial Times reported on Wednesday that Elon Musk's rocket and AI company is preparing to borrow $40 billion to buy Nvidia chips and sell the resulting compute — a business model it already runs for Alphabet's Google and Anthropic.

SpaceX carries a BBB credit rating. That tier is investment grade, but it sits at the bottom of the investment-grade ladder, and a $40 billion raise from a BBB issuer in a 5.365% yield environment would normally carry punishing terms. Cramer expects the opposite. Enthusiasm around the AI trade should let SpaceX secure attractive pricing despite the size of the deal.

"If it were any non-data center related company, its [borrowing] rate would skyrocket," Cramer said. "Not the data centers, though."

The SpaceX raise, if it closes near the terms anticipated, would amount to a real-time demonstration of Cramer's bifurcation thesis. Bond investors are pricing the company's future compute revenue rather than its existing balance sheet.

How does Skydance show the other side?

The contrast comes from Skydance, the company that recently closed its acquisition of Warner Bros. Discovery. Skydance issued roughly $40 billion of debt to fund the merger — a similar size to the rumored SpaceX raise — and the bonds fell quickly in price.

The buyers had to weigh a structurally challenged media business: cord cutting, advertising pressure, and a content cycle that no one is calling recession-proof. Where SpaceX's debt story is "AI will eventually monetize these chips," Skydance's is "the studio pipeline has to clear an advertising market that is shrinking."

Cramer's framing made the divide explicit: bond desks underwrite AI infrastructure at tight spreads while they underwrite traditional media at punishing ones, even when the headline deal sizes match.

What does "sainted" really mean?

Cramer called AI stocks "sainted" names on Wednesday. The label is not flattery. It describes companies the market has exempted from the usual discount-rate math.

"The AI data center stocks, aside from maybe Oracle, have nothing to do with what price the Federal government borrows at," Cramer said. "They only have to do with a future that's considered so bright that it obscures any problems, any bumps, even any pimples. The rest of corporate America should be so lucky."

The Oracle exception is significant. Oracle has built a sizable data center and cloud infrastructure business, but its existing debt load and software legacy still tether it to conventional credit metrics. Cramer's carve-out signals that the AI premium applies only to businesses that investors view as pure AI infrastructure plays, not to every name with adjacent cloud revenue.

Why it matters

The Treasury market's return as a primary driver of equity prices changes how analysts have to think about sector positioning. A 5.365% benchmark rate does not mean everything rerates together. It means the dispersion widens: the names tied to compute demand continue to attract patient capital, while the names tied to housing, autos, retail, and media absorb the full weight of higher discount rates.

Cramer's reading lines up with the fund flow data of the past two quarters. AI infrastructure ETFs and semiconductor-heavy funds have absorbed the bulk of net new equity inflows through 2024 and into this fall, while rate-sensitive sectors have seen persistent outflows. The 10-year auction on Wednesday forced the market to confront the price of that dispersion in real time.

If Treasury auctions stay volatile into year-end, the gap between the AI cohort and the rest of the S&P 500 will widen further before it narrows. The next defining test will come whenever SpaceX formally prices its chip-backed debt — a deal whose terms will either validate or undercut the credit-side version of the AI thesis in a single afternoon.