This AI Energy Stock Is Up Nearly 2,000% in the Last Two Years. Could It Be Nvidia's Next Big Investment?

Nvidia’s 13-F shows it owns eight AI-related stocks totaling about $63.4B as of Q2, reinforcing its “kingmaker” role across foundry, neocloud, optics, telecom, EDA, and generative-biology partners. The article highlights energy as a major AI bottleneck, citing Jensen Huang’s view that AI needs ~1,000x more power than current supply, and points to Bloom Energy as a candidate to meet that demand, noting Bloom’s revenue jumped 166% in the latest quarter, it crossed $1B in quarterly revenue, and it signed a $25B AI infrastructure power partnership with Brookfield.

Analysis

The market is slowly repricing AI from a chip-supply story into a power-and-permitting story. That shifts the best risk/reward away from the obvious GPU beneficiary and toward toll-collectors on the infrastructure stack: owners of capital, grid access, thermal management, and design tools. In that frame, the most durable winners are the firms that monetize every incremental megawatt regardless of which model wins; the most fragile are the pure-play narratives whose valuation assumes uninterrupted exponential deployment.

Bloom is the right symbol for the theme, but also the most exposed to a sentiment air pocket. Its upside is tied to customer urgency and project conversion, while its downside is that any relief in grid timelines, gas turbine availability, or financing conditions can compress the scarcity premium quickly. For the next 1-3 months, this is a headline/flow trade; over 6-18 months, it becomes a question of whether distributed power is a durable moat or just a temporary bridge until utilities catch up.

The contrarian miss is that the consensus is treating ‘AI energy’ like a single winner-take-most category. More likely, the economics accrue to diversified infrastructure capital rather than to one hardware vendor with execution and margin variability. If the next two earnings cycles do not show sustained backlog conversion, margin expansion, and project bankability, the current enthusiasm around the pure-play energy name should fade.

Key Decisions for Investors

• Long BAM / short BE as a 3-6 month relative-value pair: BAM monetizes AI power buildout through fees and asset ownership with lower execution risk, while BE already embeds aggressive scarcity expectations. Thesis breaks if BE prints multiple quarters of accelerating backlog and sustained gross-margin expansion.
• Do not chase BE outright after the run; wait for a post-earnings pullback or evidence of order conversion before initiating. If entering, prefer a call spread or small starter position to cap downside from multiple compression.
• Watch CRWV and NBIS as the more direct beta to power-secured AI capacity. If either announces long-duration power access or higher utilization tied to new capacity, they are better tactical longs than BE because the revenue translation is faster. Falsifier: capacity additions that do not lift utilization/revenue per MW.
• Treat SNPS and COHR as lower-volatility beneficiaries of sustained AI capex, but not near-term momentum trades. Favor them only if data-center power constraints do not slow design-to-deployment cycles; if order timing slips, they become laggards rather than winners.