Fed’s Barkin says bond yields remain reasonable despite recent rise

Richmond Fed President Thomas Barkin said the recent rise in U.S. Treasury yields is not being driven by inflation fears or deficit angst, calling borrowing costs 'within a reasonable range.' He also said long-term inflation expectations have not broken out and emphasized no strong forward guidance on policy, while noting AI could eventually contribute to job losses. The comments are market-relevant because they speak directly to rates, inflation, and the Fed’s current policy stance.

Analysis

The market takeaway is not that rates are “fine,” but that the burden of proof has shifted: if yields keep grinding higher without a corresponding move in inflation expectations, the equity market will increasingly treat higher real discount rates as a valuation tax rather than a macro growth signal. That is a direct headwind to long-duration AI winners, where the multiple is doing most of the work. In that setup, NVDA’s setup is less about near-term demand and more about whether the market keeps paying peak-growth multiples while the risk-free curve quietly re-prices upward.

The second-order effect is more interesting in the AI ecosystem than in the index-level macro read-through. If employers become more cautious on headcount over the next 6-18 months, software and internet names with AI-enabled labor substitution narratives can outperform hardware, because investors will start underwriting margin expansion rather than pure capex growth. That makes SMCI and APP more sensitive than NVDA to a regime where AI spend remains strong but boards demand faster ROI; those names can keep working if the market believes AI monetization is accelerating faster than wage growth.

The contrarian risk is that the bond market is not flashing a growth scare yet, which means the current move in yields can continue without a clean macro catalyst to fade it. Consensus is likely underestimating how quickly valuation compression can hit the most crowded AI names if real yields add another 25-50 bps over the next quarter. On the other hand, if the labor market softens and the Fed turns less restrictive, the AI complex can re-rate back up fast because secular earnings expectations would be preserved while the discount rate falls.

Net: this is a duration-sensitive tape where the first trade is probably not “short AI” but “own the highest quality AI cash flow, fade the weakest balance-sheet/most narrative-dependent names.” NVDA is the cleanest fundamental franchise, but it is still the most exposed to multiple compression because it is the market’s consensus proxy for AI. The better risk/reward may be in relative value and call structures rather than outright direction.

Key Decisions for Investors

• Reduce outright NVDA exposure by 15-25% over the next 1-2 weeks; re-enter only on a 5-8% pullback or if 10Y yields stabilize. Risk/reward: avoids paying peak multiple if real yields keep rising, while preserving re-accumulation optionality.
• Pair trade: long NVDA / short SMCI for 1-3 months. Thesis: NVDA is the superior franchise with stronger earnings durability, while SMCI is more exposed to multiple compression if AI hardware spend is questioned. Target 10-15% relative spread; stop if SMCI continues to print accelerating bookings and margin expansion.
• Initiate a small long APP position on 2-4 month horizon via shares or calls. If AI-driven labor substitution becomes a real narrative, software monetization beneficiaries can re-rate faster than chip suppliers; asymmetric upside if ad-tech/engagement budgets stay resilient. Keep size modest because it is more sentiment-sensitive than NVDA.
• Buy NVDA 3-6 month call spreads instead of stock into yield volatility. This preserves upside if AI capex remains intact while capping theta bleed from valuation compression; attractive if you want exposure without absorbing full duration risk.
• Set a trigger to add back to AI semis only if 10Y yields retrace by ~20 bps or if the market begins pricing a softer Fed path. If yields keep rising but inflation expectations stay anchored, treat that as a multiple compression regime rather than a buy-the-dip macro signal.