Nvidia CEO: 40% of revenue now from non-cloud customers, not just hyperscalers

Nvidia reported Q4 FY2026 revenue of $68.13B, up 73.2% YoY, with Data Center revenue of $62.31B (+75% YoY) and Data Center Networking surging 263% YoY to $10.98B. Management says roughly 50% of Data Center revenue comes from large cloud providers and the other half — including enterprise, industrial, sovereign AI and partners — is one of the fastest-growing segments; Marvell collaboration and Marvell's $1.519B data center sales (+38% YoY) underscore the ecosystem expansion. Shares trade at $174.40 (YTD -6.48%, 1yr +60.95%) vs. a $268.22 consensus target (48 buys, 12 strong buys, 1 sell), implying the CEO's re-framing could justify a re-rating and make the YTD pullback a potential buying opportunity.

Analysis

Nvidia is transitioning from a hyperscaler-dominated revenue mix to a multi-lane infrastructure provider, which changes the nature of its moat. Enterprise and sovereign deployments create higher switching costs (rack-level integration, NVLink/firmware, and long-term service contracts) than spot hyperscaler procurement, implying more durable annuity-like revenue and stickier gross margin over a 12–36 month horizon.

The expanding ecosystem (chip suppliers, board vendors, integrators, and networking partners) creates a two-way leverage: winners see revenue and pricing power accretion as the stack deepens, but Nvidia also gains negotiating leverage vs. customers who now rely on an installed base and validated software. That amplifies optionality for software and services monetization (support, validated stacks, firmware), which can shift profit pools away from pure silicon makers toward platform owners over several years.

Key risks are structural rather than cyclical: export controls, national procurement bifurcation, and the emergence of viable alternative stacks (ASICs + open software) could meaningfully slow non-cloud adoption. Near-term catalysts to monitor are multi-quarter growth in on-prem cluster deployments, multi-vendor OEM announcements referencing NVLink, and large sovereign procurement schedules — each will re-rate expectations on different timeframes (quarters for OEMs, 12–36 months for sovereign rollouts).

From a competitive angle, incumbents without interconnect ecosystems (or full-stack software) face margin pressure; conversely, infrastructure OEMs and select silicon partners can capture disproportionate upside. The market appears to underprice this diffusion of demand into on-prem and sovereign pockets while over-indexing to hyperscaler cadence, creating asymmetrical outcomes if the enterprise wave accelerates as management suggests.

Key Decisions for Investors

• Long NVDA equity (size 3–6% portfolio): accumulate on meaningful market-quality pullbacks over the next 6–12 months; target asymmetric upside of +30–60% over 12 months with a tactical stop at -18% to limit single-name drawdown — thesis: durable enterprise annuity conversion + ecosystem monetization.
• Buy a 12–18 month NVDA call spread to express upside with defined capital (cost ~3–6% of notional): buy long-dated calls and sell higher strikes to fund premium (target 2:1+ reward/risk if enterprise/sovereign deployments accelerate); reduces Vega exposure vs outright calls.
• Pair trade: long MRVL (size 1–2%) / short AMD (size 0.5–1%) over 9–18 months — MRVL captures ecosystem supplier leverage to on-prem AI builds while shorting AMD expresses relative ecosystem weakness where NVLink+software confers premium; hedge ratio ~0.6 MRVL notional per 1 AMD notional to normalize beta.
• Tail protection: purchase 9–12 month OTM NVDA puts (small hedge, 0.5–1% portfolio cost) to protect against regime-change risks (export restrictions or an alternative stack adoption) — acceptable cost to preserve capital against a sudden valuation re-rate.