Nvidia gave its first-ever year-ahead forecast—a 70% growth bombshell meant to silence AI bubble critics and ‘circular financing’ doomsayers

Nvidia shocked investors with a preliminary FY2028 revenue growth forecast of ~70% (implying ~$690B–$700B vs ~$570B modeled), driving an after-hours rally of more than 4% despite acknowledging supply constraints. The company also reported Q2 revenue of $96.2B (+106% YoY) vs $92.2B estimates and guided Q3 gross margin to 74% (vs 72.6% expected) and next-quarter revenue to $108B. While demand is growing ~100% and includes non-hyperscalers, CFO guidance highlighted margin pressure from memory-related cost resets (gross margins bottoming at 71%–72% in Q4).

Analysis

The immediate winner is not just NVDA; it is the entire constrained AI supply chain where scarcity still has pricing power. In the next 1-3 months, the cleaner monetization path may sit in memory, networking, and power/cooling vendors like MU, ANET, and VRT, because incremental capex has to flow through those bottlenecks before it becomes incremental chip volume. The large alternative-asset managers named here are also beneficiaries, but mostly on fee generation and capital-formation optics rather than direct balance-sheet exposure.

The main risk is that the market starts treating the story as vendor-financed growth rather than pure end-demand. That matters because credit markets will price the difference: if NVDA CDS or AI infra spreads widen again, the stock can re-rate down even if revenue keeps accelerating, since investors will question the durability of returns on capital. Over 6-18 months, the key falsifier is whether agentic AI actually drives utilization high enough to justify the buildout without a capex pause or customer pushback on pricing.

Contrarian take: consensus may be underweighting how much upside has already been pulled forward into the 2026-2027 capex cycle. The guide is extraordinary, but it also raises the bar for future upside surprise; once growth is this visible, multiple expansion becomes more sensitive to margin discipline and financing quality than to topline headlines. If margin guidance stalls near the low-70s or memory inflation starts leaking into customer economics, the trade shifts from momentum to mean reversion very quickly.

Key Decisions for Investors

• Buy NVDA only on a 2-4% post-earnings pullback; prefer 1-2 month call spreads over outright stock to limit gap risk. Falsify the trade if the next quarterly gross margin guide slips below 71% or if the post-earnings gap closes within 5 trading sessions.
• Express the scarcity trade via long MU vs short NVDA on a 1-3 month horizon, smaller size. The thesis is that memory pricing and supply tightness monetize faster than the already-famous GPU franchise; stop out if MU commentary fails to show ASP leverage or if NVDA pricing power remains intact without margin pressure.
• Overweight KKR and BX over BLK for 3-6 months as a capital-formation play on AI infrastructure. They should benefit from structuring fees and asset gathering with less direct credit risk; thesis breaks if AI infra deal flow slows or spreads on project financing widen materially.
• For diversified semiconductor exposure, buy ANET and VRT as a basket against QQQ over the next 1-3 months. These are the bottlenecks that should capture second-order spend as GPU deployments scale; reverse if hyperscaler capex guides flatten or delivery times normalize faster than expected.