Here's Why Nvidia's Stock Looks ‘Relatively Cheap' According to These Experts

Deutsche Bank flags Nvidia trading at ~16x projected 2027 earnings, roughly 45% below its historical median, implying a potential buying opportunity. Broadcom trades at ~17x (~5% below its long-term median), while AMD and Marvell are about 20% and 27% below their medians respectively; Nvidia shares were near $175, down ~6% YTD, and Marvell is up ~8% in 2026. The bank notes puzzling relative-valuation dislocations in the sector, with equipment names KLAC, ASML and AMAT up ~29%, 31% and 45% in 2026 and trading at premiums.

Analysis

The visible dispersion between chip designers and equipment makers is more than a pure valuation story — it reflects differing cash-flow optionality and timing of capital intensity. Designers with software/firmware attach and recurring revenue can reallocate incremental data-center spend into margin-accretive software or services, compressing the multiple gap over a 6–12 month window as investors reprice durable revenue streams versus lumpy capex exposure.

A key risk vector is the cadence of fab orders: equipment vendors live and die by multi-quarter booking visibility, so a single negative revision from a large foundry or a delay in node transitions can cascade into outsized EPS downgrades within 3–9 months. Offsetting catalysts that would reverse a designer re-rate are material increases in hyperscaler in‑house silicon, sudden acceleration in China CAPEX, or new lithography cycle funding — all of which would re-steepen the equipment earnings curve quickly.

Consensus overlooks two second-order outcomes: first, software/recurring-revenue can convert a one-time AI build into multi-year maintenance/upgrade dollars (disproportionately favoring companies with ecosystems and M&A optionality); second, the premium in equipment stocks embodies a convexity to new-node introductions and geopolitical reshoring that is binary and concentrated. That asymmetry argues for asymmetric positioning rather than straight long-or-short bets.

Key Decisions for Investors

• Pair trade (6–12 months): Long a basket of high‑ROIC chip designers (NVDA, AVGO, MRVL) vs short capital‑goods equipment names (KLAC, AMAT) on equal notional. Size to 3–5% of portfolio, target convergence of 20–35% in relative performance, stop-loss if pair widens >20% from entry.
• Volatility-defined bullish option: Buy 6–12 month call spreads on NVDA and AVGO (buy nearer‑term strike / sell higher strike) to capture a re‑rating with capped cost. Allocate <=2% of NAV per ticker, targeting asymmetric 3:1 payoff if shares re-rate seasonally post-earnings.
• Tail protection on equipment exposure (3–9 months): Buy OTM puts on ASML/AMAT sized to cover existing long exposure or use put spreads to limit premium outlay. Expect protection cost ~1–2% of notional; activate if bookings revisions begin to print negative.
• Event-driven watch: Enter/add to designer longs on confirmed hyperscaler capex refresh announcements, TSMC/Intel fab expansion approvals, or inventory drawdowns reported by suppliers. Conversely, trim designer longs and shift to protection if quarterly guides show sequential softening in GPU/server demand.