Nvidia Just Delivered Bad News for AMD and Intel

Nvidia delivered blowout Q2 earnings, easily beating consensus, and raised Q3 guidance above Street expectations. Most notably, CFO Colette Kress guided for 70% annual revenue growth by fiscal 2028 versus a Street model of 44%, alongside CPU momentum: Nvidia expects $20B of CPU sales in fiscal 2027 and to more than double that in fiscal 2028, including Vera CPU performance claims (1.8x faster agentic tasks and 5x bandwidth per watt). The agentic-AI shift toward CPU-orchestrated workflows is likely to pressure AMD and Intel’s share perception while reinforcing Nvidia’s dominance, keeping the upside narrative intact.

Analysis

NVDA is extending the AI trade from a single-product GPU story into a platform capture story: once the workload shifts to agent orchestration, the company can sell the CPU, GPU, and networking stack together, which raises wallet share and makes competitor displacement harder than a normal socket win. The market is likely underestimating how much this shifts procurement from "best chip" to "best system," which is a structural advantage for NVDA and a multiple-supportive argument over the next 6-18 months.

The first-order losers are AMD and INTC, but the more important second-order effect is margin pressure on rivals that rely on CPU attach rates in data centers. If NVDA can monetize agentic workflows, AMD/INTC may still grow in absolute dollars, yet lose relative share in the highest-value AI infrastructure spend; that makes relative-value shorts more attractive than outright bearish bets. Near term, the setup can overshoot because investors tend to extrapolate TAM expansion without immediately haircutting for pricing pressure and mix dilution.

The contrarian point is that TAM expansion does not automatically equal winner-take-all economics. If agentic workloads remain bandwidth- and software-integration bound, NVDA’s CPU entry may be more of a defensive wedge than a durable profit pool, and the CPU segment could carry lower margins than its core accelerator business. The thesis is falsified if NVDA’s gross margin guide slips or if the CPU revenue ramp fails to convert into visible design-win shipment acceleration over the next two quarters.

Key Decisions for Investors

• Long NVDA vs short AMD + INTC basket over 1-3 months; best risk/reward is relative outperformance, not a naked long/short because the AI spend pie is still expanding.
• Buy NVDA on pullbacks via a 6-9 month call spread rather than outright stock if you want upside to the FY27/FY28 narrative with defined downside; the catalyst is continued guidance credibility, not a one-day reaction.
• Do not chase a naked short in AMD or INTC; wait for next earnings/data-center guidance. If either company shows share loss or margin compression, then initiate the short as a follow-through trade.
• Use SMH/SOXX as a hedge proxy: long NVDA / short SOXX if you want to isolate NVDA’s platform premium versus the broader semiconductor basket.
• Set a watch item on NVDA gross margin and CPU attach rates in the next two quarters; a margin step-down or weaker-than-expected CPU ramp would materially weaken the bull case.