Comments (2)
王漢麟May 23, 2026
Good to know!
Sign in to comment
Good for ↑
Options market is pricing extreme uncertainty into NVDA earnings. Whether NVDA beats or misses, the elevated vol environment creates a long opportunity in AI infrastructure adjacencies — AMD and networking plays — that benefit from either NVDA's continued dominance (rising AI capex tide) or its stumble (rotation into alternatives). Long the non-NVDA AI infrastructure stack into and through earnings.
Source: MarketWatch ↗
The options market flashing elevated implied volatility into NVDA earnings is a Tier B signal — it reflects real institutional hedging activity and positioning, not just sentiment. The non-obvious read here is NOT 'NVDA is going to miss' — it's that the concentration of AI narrative risk into a single stock creates a structural rotation opportunity regardless of the outcome. Here's the transmission chain: NVDA currently commands ~70-80% of the AI accelerator market by revenue. When a single name carries this much narrative weight, institutional risk managers face binary event exposure. Pre-earnings, they trim NVDA and hedge. Post-earnings — whether up or down — capital seeks the next AI infrastructure leg. If NVDA beats massively, the entire AI capex theme re-rates and AMD, Marvell, Arista, and Broadcom all catch a bid as the rising-tide-lifts-all-boats dynamic plays out. If NVDA disappoints or guides conservatively (the tail risk the options market is pricing), hyperscalers accelerate custom silicon development, directly benefiting AMD's MI300X ramp and Broadcom's ASIC programs for Google and Meta. The market context supports this: VIX at 17.8 (down 3% on the week) suggests broad market fear is actually receding even as NVDA-specific vol spikes — this divergence means the NVDA event is idiosyncratic, not systemic. XLE leading (+5.2%) and XLB lagging (-3.7%) tells us the market is rotating toward real assets and away from materials, but tech (XLC +1.0%) is holding. The 10Y yield at 4.62% (+4.8% on the week) is a headwind for long-duration growth names, which argues for the shorter-duration, more near-term-revenue-generating AI infrastructure plays (networking, custom silicon) over pure software AI names. Target: large-cap US semiconductor and networking companies with direct AI infrastructure exposure but NOT purely dependent on NVDA's success — specifically AMD (GPU competitor), Broadcom (custom ASIC), Marvell (custom silicon + networking), and Arista Networks (AI data center networking).
The strongest bear case is that NVDA's options market signal is correctly pricing a genuine demand air pocket — hyperscalers have over-ordered H100/H200 chips and are now digesting inventory, which would mean the ENTIRE AI infrastructure stack de-rates simultaneously, not just NVDA. The specific assumption most likely wrong in this thesis is that AMD and Broadcom are 'rotation beneficiaries' — in a genuine AI capex slowdown, they are not safe havens, they are correlated losers. The data point that would flip conviction entirely: if NVDA's earnings reveal that Microsoft, Google, or Meta has materially cut AI capex guidance, the rotation thesis fails and the correct trade becomes short the entire AI infrastructure stack. Additionally, rising 10Y yields at 4.62% are a structural headwind for all growth-multiple names, and if rates continue climbing (yield up 4.8% this week alone), the multiple compression could overwhelm any earnings-driven re-rating across the sector.
The clearest historical analog is the Cisco/networking ecosystem dynamic during the 1999-2000 internet buildout. When Cisco dominated the narrative, the options market priced enormous event risk into its earnings. Savvy traders who went long the second-tier networking plays — Juniper, Foundry Networks, Extreme Networks — captured 40-200% gains in the 6 months following Cisco's peak dominance narrative, because enterprise capex was real and distributed across the ecosystem. The key insight from that period: when one company's earnings become a referendum on an entire technology cycle, the ECOSYSTEM benefits regardless of that single company's result. More recently, in Q4 2023, NVDA beat massively and AMD's MI300X narrative re-rated +35% in the following 6 weeks as investors concluded 'AI capex is real and AMD gets a slice.' The behavioral pattern is consistent: extreme single-stock event risk in a dominant player creates pre-earnings suppression and post-earnings re-rating in the ecosystem. The options market's 'ominous sign' is actually the setup, not the warning.
Elevated NVDA options implied volatility signals institutional hedging and binary event risk, causing pre-earnings selling pressure in NVDA and muted performance in adjacent AI names as capital waits on the sidelines.
Post-earnings resolution — in either direction — unlocks capital rotation. A NVDA beat re-rates the entire AI capex cycle bullishly; a miss or cautious guide accelerates hyperscaler diversification toward AMD MI300X and Broadcom/Marvell custom ASICs. Either path directs fresh institutional capital into large-cap US semiconductor and AI networking companies that have underperformed NVDA YTD.
The third-order effect is structural: as AI infrastructure spending becomes multi-vendor (driven by hyperscaler risk management and geopolitical pressure to avoid single-supplier dependence), the revenue base for AMD, Broadcom, Marvell, and Arista Networks broadens durably. Rising 10Y yields at 4.62% favor companies with near-term revenue visibility over speculative AI software plays, making the hardware infrastructure layer — with real data center contracts and shipment schedules — the preferred expression. The options market's fear is actually the market's way of telling you the AI capex cycle is real enough to hedge, which is the most bullish possible signal for the ecosystem.
Earlier connections in the same sector that share a theme with this one.
Good to know!
Sign in to comment
I like this idea! Took an entry in AVGO / long d.d. 20-05-2026 ($417.90). I read that next to NVIDIA AVGO is the stock with the most long positions. We'll see ;-)