Fund Manager: Wall Street Is Using the Wrong ‘E' to Judge AI Valuations

The article argues that AI-related earnings power is being underestimated, highlighting GE Vernova’s Q1 2026 revenue of $9.34B versus $9.3B consensus, 71% organic order growth, and a record $150B backlog. It also cites Vertiv’s 252% organic order growth, Eaton’s electrical backlog up 48%, Constellation’s reaffirmed 2026 adjusted operating EPS of $11-$12, and NVIDIA’s Q1 FY2027 revenue of $81.615B, up 85.23% year over year. The message is broadly bullish for the AI power and infrastructure supply chain, with investor focus shifting from trailing P/E to accelerating fundamentals and backlog visibility.

Analysis

The key market error is not valuation level but valuation denominator drift: consensus is still treating AI-related industrial demand as a one-cycle upgrade when it is behaving like a multi-year capacity supercycle. That means the biggest upside is often in the least glamorous balance-sheet-dependent suppliers that can reprice faster than end-demand models can update, especially where capacity is structurally constrained and lead times are still extending. In that setup, earnings estimates for the picks-and-shovels layer should continue to rise even if headline AI capex eventually normalizes.

Second-order winners are the firms that sit closest to bottlenecks, not necessarily the firms with the highest direct AI revenue exposure. Power equipment, grid gear, and firm generation assets become quasi-toll roads on AI deployment, while slower-moving competitors risk margin compression if they lack pricing power or manufacturing scale. The more important spillover is that higher turbine and electrical backlog visibility should gradually pull capital toward adjacent industrials, potentially creating a broader re-rating than the market is assigning today.

The contrarian risk is that investors overpay for the obvious beneficiaries after a sharp run and miss the fact that the real earnings inflection may still be ahead. If AI capex merely stays elevated instead of accelerating, multiples could compress even as fundamentals improve, which argues for owning names with visible backlog conversion rather than pure narrative exposure. The clearest reversal trigger would be a pause in hyperscaler power commitments or a delay in data-center commissioning, because that would hit order growth first and equity multiples second, with a lag of one to three quarters.

The deeper takeaway is that the market may be underestimating how much of the AI buildout is now dictated by physical infrastructure constraints, not software adoption curves. That favors assets with contractual revenue, regulated or quasi-regulated pricing power, and long-dated purchase agreements. It also means the eventual beneficiaries are likely to be broader than the current consensus basket, because every incremental MW of compute requires a cascade of grid, cooling, and generation spend before it ever becomes an earnings story.

Key Decisions for Investors

• Add to GEV on pullbacks over the next 2-6 weeks; keep risk tight with a trailing stop because the setup is powerful but momentum-sensitive. Risk/reward favors owning backlog re-acceleration and turbine pricing power, but only if consensus keeps moving up.
• Go long VRT versus short a basket of slower-growth electrical/industrial peers for a 3-6 month relative-value trade. The thesis is that order conversion and backlog growth should outpace more diversified competitors, creating a margin and multiple gap.
• Own ETN as a lower-beta way to express the grid bottleneck theme over 6-12 months. Use it as a quality compounder, but size below GEV/VRT because the re-rating potential is smaller and the market already assigns some scarcity value.
• Buy CEG on any post-close digestion period as a 12-24 month hold, especially if long-duration PPAs continue to lengthen. The risk/reward is strongest if investors keep underpricing contracted power economics and the incremental value of firm generation.
• Maintain core NVDA exposure but hedge with short-dated calls or covered-call overwrite into strength over the next 1-2 quarters. The operating trend remains exceptional, but the stock is increasingly vulnerable to any moderation in order growth or a compressing expectation gap.