Jim Cramer Responds to Investor With 60% of Portfolio in NVIDIA

NVIDIA remains a strong AI leader, with Q1 FY2027 revenue of $81.6B, up 85.2% year over year, and Data Center revenue up 92% to $75.25B; management also authorized an additional $80B in buybacks. Jim Cramer reaffirmed his positive view on NVIDIA while endorsing GE Vernova as a diversification idea, citing its role in AI-related power infrastructure. GE Vernova has gained more than 105% over the past year, but Cramer said it is now at a better entry level after a nearly 9% pullback in the past month.

Analysis

The important setup is not simply that AI demand remains strong; it is that the second derivative of the trade is shifting from compute to power. That matters because the market can keep re-rating NVIDIA on earnings momentum, but the larger incremental capital pool may migrate to grid, generation, and electrification bottlenecks where supply is tighter and lead times are longer. In that sense, GE Vernova is a cleaner way to express the next leg of AI capex without paying the richest multiple in the chain.

NVDA is still a fortress asset, but the concentration risk is now the real issue: when a single name becomes ~60% of a portfolio, the investor is no longer making an alpha decision, he is making a macro beta decision on one theme. The more interesting implication is that NVIDIA’s buybacks and continued outperformance can actually increase index and retail crowding, which raises the probability of episodic air pockets if AI capex expectations pause for even one quarter. That creates a better asymmetry for monetizing gains gradually than for adding fresh capital here.

GEV’s upside is less about headline growth and more about multi-year order conversion. Data-center electrification orders already outrunning full-year prior totals suggests a backlog-upgrade cycle that should support estimates for several quarters, but the stock has already repriced a lot of that scarcity value. The key risk is that this becomes a “great story, mediocre entry” if power equipment margins or execution disappoint, so the trade works best as a rotation out of embedded consensus winner into a still-underowned beneficiary with a longer runway.

The contrarian takeaway is that the market may be underestimating how durable the AI infrastructure trade is outside semis. If power availability becomes the gating factor, then the next scarce assets are transformers, turbines, switchgear, and grid services—not just chips. That argues for owning the bottleneck providers, but with discipline: the first leg is likely multiple expansion, while the second leg requires actual backlog conversion and margin protection over the next 2-6 quarters.

Key Decisions for Investors

• Trim NVDA into strength and redeploy 25-50% of proceeds into GEV over the next 1-3 weeks; this reduces single-name concentration while keeping AI exposure, with lower theme-correlation risk if semiconductor multiples compress.
• Initiate a NVDA/GEV pair trade: short 1x NVDA against long 0.5-0.7x GEV on a value-at-risk basis; expect near-term relative underperformance in NVDA if AI spend broadens into infrastructure over the next 3-6 months.
• Buy GEV on pullbacks of 5-8% from recent highs rather than chasing; the risk/reward improves if the market gives a post-run consolidation, since backlog visibility supports a 6-12 month hold even if near-term sentiment cools.
• For traders wanting convexity, use NVDA call spreads 3-6 months out instead of outright stock; this preserves upside if AI capex reaccelerates while capping downside if the market rotates away from crowded mega-cap tech.
• Set a hard portfolio rule: no single equity above 10-15% of liquid net worth; for oversized NVDA holders, reduce in staged 10-15% tranches to avoid timing risk and preserve tax optionality.