Hey there, bargain hunter. You have been sold a chip story. Nvidia up, Micron up, AMD chasing. That is the frame most investors are still working inside, and it made sense in 2023 when GPUs were the scarce resource. In August 2026, the scarce resource is electrons, heat management, and the racks that hold the whole thing together. The companies collecting that rent are not semiconductor designers. They make turbines. They make cooling units. They make switchgear. They assemble servers. And their earnings this cycle were, by any reasonable measure, more convincing than most of the chip names that get the headlines.

Scoreboard

Four companies reported in the last three weeks that together tell you more about AI economics than any GPU benchmark. Here is what the numbers actually said.

• GE Vernova (GEV): Q2 2026 orders of $24.2 billion, up 88% year-over-year. Backlog at $176 billion. Data center electrification orders topped $5 billion year-to-date, more than double the full-year 2025 total. Gas turbine backlog and slot reservation agreements hit 116 GW by June 30, up from 100 GW just three months earlier. Full-year free cash flow guidance raised to $11.5 billion to $12.5 billion.
• Vertiv (VRT): Q2 net sales of $3.27 billion, up 24% year-over-year. Adjusted operating margin expanded 410 basis points to 22.6%. Adjusted free cash flow surged 234% to $925 million. Full-year guidance raised to a midpoint of $14.0 billion in net sales.
• Eaton (ETN): Q2 revenue of $8.53 billion, up 21%. Management pointed to strong data center-driven demand within its electrical businesses and raised full-year adjusted EPS guidance to $13.40 to $13.60.
• Dell (DELL): Full fiscal year 2026 AI orders of $64.1 billion. AI server shipments of $25.2 billion for the year. AI backlog exiting fiscal 2026 at a record $43 billion. Fiscal 2027 AI revenue now guided to approximately $60 billion.

The Real Reason This Is Different

The chip cycle has peaks. Memory cycles. GPU cycles. Every investor who has been in semiconductors for more than one cycle knows the drill: demand surges, supply catches up, margins compress, the stock gives back half the run. That is not what is happening in the industrial power layer right now.

Consider what GE Vernova’s CEO Scott Strazik told investors in July: the company is already taking slot reservations for 2031 gas turbine deliveries. The gas turbine backlog and slot reservation agreements sit at 116 GW, and management expects to hit at least 125 GW under contract by year-end. That is not a demand spike. That is a manufacturing constraint on a years-long order book.

Vertiv operates on the same logic from a different angle. Every megawatt GE Vernova routes into a data center eventually generates heat that Vertiv has to manage. The shift from air cooling to liquid cooling, driven by GPU rack densities that have climbed sharply since 2023, is a direct and durable revenue tailwind that no model efficiency gain can fully offset. You can train a smarter model with fewer flops, but physics still requires you to remove the heat from whatever compute you do run.

Eaton sits one layer upstream, supplying the electrical distribution and power management equipment that gets the electricity from the substation to the rack. Management has described demand as multi-year in nature, with electrical orders and backlog supported by AI data center buildouts and broader electrification.

Deep Dive: What These Businesses Actually Do

Most retail investors can explain what Nvidia does. Fewer can explain what Vertiv does, even though Vertiv’s product is inside every data center that buys a Nvidia GPU.

GE Vernova makes the gas turbines and grid equipment that generate and transmit the electricity AI clusters consume. It is not a utility; it sells capital equipment to utilities, power producers, and increasingly to hyperscalers who are sourcing power directly. Its Electrification segment, which makes transformers and switchgear, is the piece that connects generation to data center loads. Both segments are in a multi-year order cycle driven by the same underlying demand.

Vertiv makes the power and cooling systems inside data centers: uninterruptible power supplies, thermal management units, and increasingly liquid-cooled rack systems for the densest AI clusters. It gets paid when data centers are built and when they are upgraded. The upgrade cycle alone is a recurring revenue stream that competitors cannot easily replicate because Vertiv’s equipment is integrated into the facility design.

Eaton makes electrical distribution equipment, including switchgear, circuit breakers, and now, through its Boyd Thermal acquisition, liquid cooling systems for data centers. The Boyd acquisition adds direct liquid cooling capability to a company that already dominated the electrical distribution stack.

Dell is the integrator: it assembles the GPUs, memory, networking, and storage that hyperscalers, sovereign cloud buyers, and now more than 4,000 enterprise customers need packaged into deployable AI server systems. Dell does not design the GPU. It designs the system around the GPU, and it handles deployment, services, and lifecycle support. That is a lower-margin business than fabless chip design, but it is a business with $43 billion in AI backlog and a five-quarter pipeline that kept growing even after a record-order quarter.

Data Section

The numbers across these four names, taken together, describe a layer of the AI trade that is already generating real cash flow, not forward promises.

GE Vernova:

• Q2 2026 orders: $24.2 billion, up 88% organically year-over-year
• Q2 2026 revenue: $11.1 billion, up 12% organically
• Backlog: $176 billion
• First-half free cash flow: approximately $9.9 billion
• Full-year 2026 revenue guidance: $45.5 billion to $46.5 billion
• Full-year free cash flow guidance: $11.5 billion to $12.5 billion
• Gas turbine backlog and slot reservation agreements: 116 GW, targeting at least 125 GW by year-end
• Data center electrification orders: over $5 billion year-to-date, more than double full-year 2025

Vertiv:

• Q2 2026 net sales: $3.27 billion, up 24% year-over-year
• Adjusted operating margin: 22.6%, up 410 basis points year-over-year
• Adjusted operating profit: $738 million
• Adjusted EPS: $1.52, up 60%
• Adjusted free cash flow: $925 million, up 234% year-over-year
• Full-year 2026 guidance: net sales of $13.8 billion to $14.2 billion

Eaton:

• Q2 2026 revenue: $8.53 billion, up 21% year-over-year
• Adjusted EPS: $3.15, a Q2 record
• Full-year 2026 organic growth guidance: 11% to 13%
• Full-year adjusted EPS guidance: $13.40 to $13.60

Dell:

• Full fiscal year 2026 revenue: $113.5 billion, up 19%
• Full fiscal year AI orders: $64.1 billion
• Full fiscal year AI shipments: $25.2 billion
• AI backlog exiting fiscal 2026: $43 billion (record)
• Fiscal 2027 total revenue guidance: $138 billion to $142 billion
• Fiscal 2027 AI-optimized server revenue target: approximately $60 billion
• Fiscal 2027 non-GAAP diluted EPS guidance: approximately $12.90
• Annual cash flow for fiscal 2026: more than $11 billion (record)
• AI server operating margin target: mid-single digits

Is It Cheap?

This is where the honest answer gets complicated, because none of these stocks are cheap on trailing multiples. The argument for each one is not that it is a traditional value play. The argument is that the multiples are reasonable against a growth trajectory that is already in firm orders, not in analyst projections.

GE Vernova trades at roughly 31 times trailing earnings on a backlog of $176 billion, with 2031 gas turbine production already partly spoken for via orders and slot reservation agreements. That is not a speculative multiple on a company that might grow. It is a multiple on a company that physically cannot deliver its orders fast enough. The ceiling on the multiple is the execution risk, not the demand risk.

Vertiv has expanded its adjusted operating margin from the low single digits three years ago to 22.6% today. The free cash flow machine it is building, $925 million in a single quarter, is what matters for long-term valuation. The stock has run, but the free cash flow trajectory is still accelerating.

Eaton’s valuation is more moderate than Vertiv’s, partly because of the pending Mobility segment separation and the near-term integration work from the Boyd Thermal acquisition.

Dell is the most interesting valuation case because the market has not fully credited a company targeting roughly $60 billion in AI revenue in fiscal 2027 at a mid-single-digit operating margin. That is a low-margin business, but $60 billion times even a 5% operating margin is $3.0 billion in segment operating profit from a single product line that barely existed two years ago.

Bull, Base, and Bear

Bull case: Hyperscalers keep their foot on the capex pedal. GEV converts its slot reservation agreements into firm orders at the pace management is signaling. Vertiv’s growth confirms that the AI infrastructure boom is no longer a North American story. Eaton’s electrical businesses keep converting multi-year order strength into revenue and margin. Dell’s customer base, which already exceeded 4,000 AI buyers including sovereigns, enterprises, and neo-cloud operators, keeps adding logos. In that scenario, all four names are compounding earnings faster than their current multiples price in.

Base case: Hyperscaler capex holds at current levels but does not accelerate. GEV and Vertiv continue to convert backlogs into revenue with improving but not expanding margins. Eaton benefits from Boyd but faces normal integration noise as the Mobility spin approaches. Dell sustains its fiscal 2027 AI revenue target but margin pressure from component costs limits EPS upside. All four still grow earnings meaningfully. None return the 200% gains that early investors captured. That is not a failure; it is a mature phase of a durable cycle.

Bear case: A major efficiency breakthrough in model architecture, analogous to what some teams claimed in early 2025, substantially reduces the power and cooling intensity per unit of AI output. Hyperscalers pause new data center commitments and existing orders face push-outs. GEV’s slot reservation agreements do not convert at management’s pace, and wind segment losses remain a visible drag. Vertiv’s pricing power fades as competition enters the liquid cooling market. Dell faces margin compression if AI server profitability proves structurally capped in the low-to-mid single digits. In that scenario, the backlogs are still real, but the pace of conversion slows enough to compress multiples.

Action Plan

If you already own Nvidia, you own the most-discussed part of this trade. These four names are the parts most investors have not bought yet.

The cleanest way to think about it is sequentially: electrons first, then thermal management, then distribution, then servers. GEV and ETN sit on the electron and distribution side, with orders locked years in advance and meaningful free cash flow already in the financial statements. VRT sits on thermal management, which is a function of installed base growth as much as new data center builds. DELL sits closest to the GPU cycle but with a customer base that is broadening rapidly beyond hyperscalers into sovereigns and enterprise.

• GEV: Position sizing should reflect the duration risk. This is a long-cycle business, and the backlog converts slowly. A 3% to 5% position for investors with a three-year horizon is a reasonable starting point. Scale in on any quarter where the earnings miss is driven by the wind segment, not by Power or Electrification order weakness.
• VRT: The free cash flow inflection, from a company that was generating minimal cash three years ago to $925 million in a single quarter, is the signal. A dip toward $280 to $290 per share would be a more comfortable entry than chasing the current level. The long-term margin target makes the valuation defensible at current prices for patient holders.
• ETN: The Mobility separation, targeting Q1 2027 via Reverse Morris Trust with Dana, is a re-rating event. Investors who buy ahead of that close are likely to see a more focused electrical and aerospace business with better margin optics.
• DELL: Dell’s fiscal 2027 AI revenue target, combined with $12.90 in forward EPS, is not priced like a high-growth stock. Scale in gradually; the margin sensitivity to component costs is a real risk, and the stock can move sharply on server component price changes.

Cheap Investor Checklist

• GEV gas turbine backlog and slot reservation agreements: watch for quarterly conversion rate from SRAs to firm orders
• GEV wind segment EBITDA: a negative surprise is the main risk to the thesis
• VRT adjusted operating margin trajectory: management targeting higher margins over time; any quarter coming in above recent levels validates the path
• VRT APAC organic growth: a sustained above-20% read confirms the buildout is genuinely global, not a U.S. hyperscaler story
• ETN Electrical businesses orders and backlog: any sustained weakening is your early warning signal
• ETN book-to-bill: a sustained drop below 1.0 in the electrical businesses signals demand deceleration
• ETN Mobility separation close: targeting Q1 2027, delivers approximately $1.1 billion cash distribution and sharpens the portfolio
• DELL AI server operating margin: management targeting mid-single digits; any quarter above 6% is a positive surprise
• DELL AI backlog conversion: watch quarterly shipments relative to order intake; a declining backlog is fine if shipments accelerate, not fine if orders slow
• Hyperscaler capex guidance for 2027: Microsoft, Amazon, Alphabet, and Meta guidance updates in October and January will either validate or threaten the order books at all four companies

Bottom Line

If hyperscalers continue spending at or above current levels, the constraint on AI buildout is not GPUs. It is electrons, cooling, and electrical distribution. GEV, VRT, ETN, and DELL are the companies collecting that constraint premium, with backlogs already on the books.

If capex cools, these names will see multiple compression before they see order cancellations, because the backlog provides a revenue cushion that chip designers do not have.

The framing is not buy-and-ignore. The ETN Mobility separation, VRT’s margin trajectory, and GEV’s conversion of slot reservations into firm orders are all time-stamped events that will either confirm or challenge the thesis by mid-2027. Watch those three. The rest of the story is already in the financials.

This article is for informational purposes only and does not constitute investment advice. Past performance does not guarantee future results. All investing involves risk, including the possible loss of principal.