AI Power Trade: Vertiv, GE Vernova, Eaton Lead 2026 Rally

The trade that keeps working in 2026 is not the hyperscalers themselves — it is the picks-and-shovels behind them. Power equipment, grid contractors, and data-center thermal specialists are outrunning the S&P 500 by wide margins as AI capex forecasts get pushed higher and utility interconnection queues stretch out for years. On Wednesday, September 3, another leg up: Vertiv +4.73%, GE Vernova +2.16%, Eaton +1.60%, Quanta Services +1.52%, and Flex +1.83%.

The rally, in numbers

All five names below sit within their AI-data-center revenue arc — power distribution, thermal management, medium-voltage switchgear, transmission construction, and now power conversion for behind-the-meter deployments. Prices and returns are as of the September 3, 2026 close.

Ticker Company Price Day YTD Mkt Cap
VRT Vertiv Holdings $268.83 +4.73% +66.0% $103.5B
FLEX Flex Ltd. $107.89 +1.83% +78.6% $39.9B
PWR Quanta Services $620.06 +1.52% +47.0% $93.2B
GEV GE Vernova $941.84 +2.16% +44.1% $250.8B
ETN Eaton Corp. $397.12 +1.60% +24.7% $154.2B
Source: Yahoo Finance quote pages, as of the September 3, 2026 close.

Why the bid is structural, not speculative

The demand story starts on the customer side. Every incremental hyperscaler training cluster now measures in hundreds of megawatts, and the interconnection queues at U.S. utilities are pushing multi-year wait times — most acutely in ERCOT, where regulators recently paused new data-center hookups against 474 GW of speculative requests. Every gigawatt that does clear the queue drags along the same shopping list: medium-voltage switchgear, transformers, uninterruptible power supplies, liquid-cooling loops, and the transmission crews to string the new lines. That shopping list is the revenue base for the names above.

Two data points anchor how tight the market has become:

  • Quanta Services reported a record $53.4 billion backlog in Q2, with net income nearly doubling to $451.4 million; the company raised full-year guidance on the print. Backlog of that size — roughly three years of current revenue — is what “structural” looks like when it shows up on a balance sheet.
  • Eaton committed more than $242 million to a new Arkansas manufacturing facility for modular electrical enclosures, one of several capacity expansions aimed at electrical-infrastructure demand where the electrical segment already contributes roughly 70% of revenue.

Neither commitment reads like a company positioning for cyclical peak. They read like companies underwriting multi-year build-out plans and putting capex behind them.

The M&A wave is doing the same job

Strategic buyers have started paying up to close capability gaps rather than build them. Vertiv announced an acquisition of Utility Innovation Group to deepen its behind-the-meter power capabilities, and Flex disclosed an acquisition of EPC Power aimed at power conversion for AI data centers and grid applications — with Flex simultaneously guiding FY27 revenue to roughly $35.2 billion and preparing a spin-off of its Cloud and Power Infrastructure segment. The pattern rhymes with the last wave of hyperscaler cloud consolidation: when demand outruns the ability to build organically, incumbents buy specialists.

The read-through for public markets is simpler than the deal mechanics: the acquirers are signaling that this cycle is not a two-year sugar rush.

YTD performance vs the S&P 500

Every name in the basket is more than 2x the S&P 500’s roughly 13% year-to-date return, and the tightest laggard (Eaton) is still nearly 2x the index.

Year-to-date total return: AI power basket vs S&P 500 Bar chart of YTD returns as of September 3, 2026: FLEX 78.6%, VRT 66.0%, PWR 47.0%, GEV 44.1%, ETN 24.7%, S&P 500 13.2%. 0% 20% 40% 60% 80% 78.6% FLEX 66.0% VRT 47.0% PWR 44.1% GEV 24.7% ETN 13.2% S&P 500 Year-to-date total return through Sep 3, 2026 close.
Source: Yahoo Finance quote pages (FLEX, VRT, PWR, GEV, ETN); S&P 500 YTD from broker-quoted index return.

What could break the trade

The bear case is not that AI demand collapses — it is that the supply side finally catches up. Three risks are worth watching:

  1. Utility interconnection reform. If PJM, ERCOT, and MISO succeed in clearing their queues faster than expected, the scarcity premium embedded in equipment lead times compresses. That directly hits pricing power.
  2. Hyperscaler capex pause. The three biggest AI capex spenders account for a disproportionate share of the incremental order book. Any of them signaling a step-down in 2027 capex would ripple through backlogs quickly.
  3. Valuation, not fundamentals. VRT and PWR both trade near 40–58x trailing earnings; Flex is at ~42x. These are growth multiples on cyclical businesses. Historically, that combination compresses fast when growth decelerates by even a few points.

None of those risks is present in the tape today. Backlogs are still expanding, hyperscaler capex guides are still being raised, and utilities are still throttling new interconnections rather than opening the taps. The trade works until one of those changes.

Sources

Disclosure: This article is for informational purposes only and is not investment advice.

Leave a Comment