U.S. 10-yr yields surge past 5%: which stocks are most vulnerable?

The U.S. 10-year Treasury yield rose to 5.03%, its highest level since 2007, intensifying valuation and refinancing pressure across rate-sensitive assets. Semiconductors sold off sharply on Sep. 14, led by CoreWeave (-6.75%), the PHLX Semiconductor Index (-5.86%), Intel (-5.59%) and Micron (-5.25%), as higher discount rates disproportionately hurt long-duration technology and AI-infrastructure companies. REITs, utilities and highly leveraged borrowers also face margin and refinancing risks, while energy, defense and deposit-rich financials may be relative beneficiaries; the key downside risk is a further rise toward 6% amid fiscal deficits near 6% of GDP and roughly $40T in public debt.

Analysis

The critical transmission channel is not the headline yield level but the combination of a higher term premium and refinancing costs that remain elevated into 2027. CRWV is the clearest equity-duration casualty: its valuation requires both sustained AI demand and continued access to large-scale project financing, so a higher cost of capital can impair returns on incremental data-center capacity before revenue growth visibly slows. By contrast, NVDA, AVGO and MU have materially stronger cash-generation profiles; broad semiconductor selling is therefore likely to create dispersion rather than a uniform fundamental reset.

For the next 1-3 months, Treasury auction outcomes, core inflation surprises and credit spreads matter more than another marginal move in the policy-rate outlook. A disorderly rise in long yields would pressure REITs, utilities and levered infrastructure through lower asset values and weaker dividend coverage, while banks with large unrealized securities losses could see deposit competition intensify. LMT is relatively insulated because its cash flows are tied more to appropriations and backlog conversion than consumer or corporate financing conditions, though defense names are vulnerable if long-end rates force fiscal restraint rather than higher nominal defense spending.

Consensus is likely overextending the "short all tech" conclusion. If long rates stabilize without a growth recession, cash-rich AI suppliers can re-rate on earnings durability while financing-dependent AI infrastructure underperforms sharply. The cleaner expression is therefore a quality-duration pair: short the externally financed buildout layer versus long the picks-and-shovels suppliers, rather than an outright semiconductor index short.

The thesis fails if the 10-year yield retreats below roughly 4.7% alongside easing investment-grade spreads and improving data-center financing terms; that combination would restore risk appetite fastest in CRWV and other high-beta infrastructure equities. Conversely, a sustained move toward 5.5% with widening high-yield spreads would turn the current valuation issue into an earnings and liquidity problem over the next two to four quarters.

Key Decisions for Investors

• Initiate a 3-month pair trade: short CRWV / long NVDA, dollar-neutral. Target 15-25% relative downside for CRWV if long-end yields remain above 5%; cover if the 10-year closes below 4.7% for two weeks or CRWV demonstrates materially improved financing costs and contracted-capacity economics.
• Add a defensive relative-value sleeve: long LMT / short XLRE or XLU over 3-6 months. The trade monetizes the gap between contracted defense cash flows and rate-sensitive, leveraged yield proxies; reassess if real yields decline 50bp or Congress signals meaningful defense-budget compression.
• Do not short the semiconductor complex indiscriminately after the initial de-risking move. Use further broad weakness to selectively accumulate NVDA and AVGO only after the next earnings cycle confirms backlog conversion and gross-margin resilience; avoid AMD and INTC as long-duration valuation exposure until forward estimates stop being revised down.
• Place a credit-risk alert rather than a new bank-sector short: monitor BBB industrial and data-center borrower spreads. A 50bp-plus widening from current levels would justify expanding shorts in financing-dependent infrastructure and REIT exposure, because the equity impact would then reflect impaired refinancing access rather than only multiple compression.