Cramer Wants Nvidia to Buy Back a Tenth of Itself but One Analyst Expects the Stock To Hit $400 Without It

New Street’s Pierre Ferragu expects NVDA to reach $400 within 18 months on earnings power alone, citing Jensen Huang guidance for at least 70% growth next year and valuing NVIDIA on “single digit” forward earnings multiples versus a not-yet-fully-repriced earnings base. Jim Cramer argues the stock’s price discovery is broken and wants a buyback closer to a tenth of market value (>$500B versus NVIDIA’s already-authorized $80B additional repurchase capacity). The fundamental backdrop is strong—Q2 FY27 revenue was $96.22B (+105.8% YoY) with $2.22 non-GAAP EPS (fifth consecutive beat)—while options positioning via a 0.48 put/call ratio leans toward Ferragu’s direction.

Analysis

The buyback debate is mostly noise relative to the underlying earnings vector. For a company compounding this fast, capital returns matter more as a signaling device than as a valuation driver; a repurchase can tighten float, but it does not change the fact that the stock is trading like a long-duration cash-flow bond whose price is set by growth durability and terminal margin assumptions. That makes the real winners the holders of the highest-conviction AI exposure, while the relative losers are peers that need valuation support from scarcity rather than from their own growth inflection.

Second-order, the message to the ecosystem is that AI infrastructure spend is still being validated, which keeps pressure on the rest of the semiconductor and networking chain to stay capacity-constrained. That is supportive for TSM, AMAT, LRCX, ANET, and potentially the higher-beta power/cooling names, while AMD and INTC remain vulnerable if the market keeps treating NVDA as the default compute standard. The immediate reaction can be sentiment-driven, but the real catalyst path is 1-3 months of hyperscaler capex commentary and 6-18 months of realized earnings power; if spend normalizes, the multiple can compress very quickly even if the business remains strong.

The contrarian point is that the market may be over-fixated on headline share-count reduction when the real question is whether the forward earnings base can keep moving up faster than expectations. If growth stays near the current implied trajectory, the stock can keep re-rating without any board action; if growth slows even modestly, a buyback will not offset duration risk. The key falsifiers are a guide-down in data center growth, margin compression from mix or competition, or evidence that AI monetization is lagging capex demand by more than a quarter or two.

Key Decisions for Investors

• Stay long NVDA on pullbacks rather than chasing strength; the better entry is any 3-5% post-headline fade, with a 6-12 month horizon and upside driven by estimate revisions rather than buyback headlines.
• Pair long NVDA / short AMD or SOXX for a relative-value trade over the next 1-3 months: if AI capex remains hot, NVDA should keep winning share and multiple leadership; risk is a broad sector rerating that lifts all semis.
• Use NVDA call spreads instead of outright stock for a 3-6 month catalyst window; the market is already leaning bullish, so defined-risk upside is better than paying full delta into a crowded consensus.
• Watch hyperscaler capex prints and NVDA next-quarter data center guidance as the falsifier; if revenue growth decelerates materially or gross margin stops expanding, de-risk immediately.
• If you want broader AI exposure with less single-name event risk, favor TSM/AMAT/LRCX over NVDA for the next 6-18 months because they capture the capex cycle even if NVDA’s multiple stalls.