Treasury auctions and manufacturing data due Monday By Investing.com

This is less about a one-day macro calendar and more about the market using rates as the referee for the AI complex. If short-end auctions clear smoothly, it reinforces the view that financing conditions can stay benign enough for megacap capex plans to keep compounding, which is structurally supportive for NVDA’s earnings duration and valuation. The second-order effect is that any easing in front-end yields tends to keep the market willing to pay up for the AI bottleneck beneficiaries while making balance-sheet-sensitive legacy hardware names look even more like laggards.

INTC remains the more fragile setup because its re-rating thesis depends on showing credible execution while funding a multi-year catch-up cycle under a still-expensive capital backdrop. In a “rates stay higher for longer” regime, the market is likely to reward pure product momentum over turnaround optionality, which leaves INTC exposed both to competitive loss and to any disappointment in foundry-related messaging. The asymmetry is that even modestly stronger demand for compute generally flows first to the incumbent architecture leaders and only later to turnarounds.

The contrarian miss is that a rally in NVDA on the back of “AI pivots to CPUs” may actually be overstating the CPU substitution narrative. More likely, this is a workload-partitioning story: CPUs gain some share at the margin in orchestration and inference control planes, but the highest-value AI training and dense inference economics remain GPU-led. That means the trade is not NVDA versus CPUs in a zero-sum sense; it is NVDA versus lower-quality semis and obsolete server exposure, with Intel the clearest relative loser if the market keeps rewarding visible monetization and penalizing execution risk.