Oracle's Data Center Spending Is Exploding. Here's the Power Stock That Actually Benefits.

Bloom Energy reported record Q2 revenue of $1.065B (+166% YoY) and EPS of $0.62 versus a prior-year loss of $0.18, alongside a 668 bps jump in gross margin to 33.4%. The company raised full-year revenue guidance to $3.9B–$4.2B (+100% vs 2025) and says it is funding manufacturing expansion from positive operating cash flow. Shares are also tied to Oracle’s April contract for up to 2.8 GW of on-site fuel cell power, though valuation remains a key risk (about 290x trailing and 78x forward earnings).

Analysis

This is less a clean fuel-cell story than a monetization of the AI power bottleneck. In the near term, BE is the obvious levered winner because it converts grid delay into pricing power: every incremental hyperscaler win should carry better gross margin than legacy industrial deployments if installation speed remains the scarce asset. ORCL benefits operationally, but the market may be underestimating how much of the AI buildout is being gated by power delivery rather than chips or software demand.

The second-order effect is on capital allocation across the data-center stack. EQIX and similar operators can use distributed power to improve tenant uptime, but that also raises the bar for capex discipline and could compress returns if they start over-building behind-the-meter capacity before utilization is proven. On the loser side, utilities and grid-tied power infrastructure names face a subtle demand leak: if large-load customers can bypass interconnection queues, some future load growth never reaches the regulated grid economics.

The contrarian risk is that the market is extrapolating one marquee contract into a durable platform. The key question over the next 1-3 months is whether order flow broadens beyond a few hyperscalers; over 6-18 months, the thesis hinges on BE sustaining low-30s gross margins while scaling manufacturing without working-capital strain. A faster grid buildout, softer AI capex, or a gas-price spike would all pressure the valuation multiple and expose how much of the move is momentum versus durable earnings power.

Key Decisions for Investors

• Initiate/keep a tactical long in BE only on pullbacks or via 6-9 month call spreads; the stock can stay extended, but the position should be sized for valuation compression if backlog conversion slows.
• Use a relative-value long BE / short XLU expression for 1-3 months if the market continues to price behind-the-meter power as a structural winner over regulated grid assets.
• Add ORCL on any weakness as an indirect beneficiary of faster AI capacity deployment, but treat it as a lower-beta confirmation trade rather than the primary way to play the theme.
• Avoid chasing EQIX here; it is a second-order beneficiary, but any upside from power optionality may be offset by capex and margin drag. Reassess only if leasing and tenant expansion metrics show that power availability is translating into pricing power.
• Set a thesis-falsifier alert: if BE's next guide implies gross margins below ~30% or revenue growth decelerates sharply, fade the multiple and consider reducing exposure.