AI infrastructure is a large market, but that does not make every company in the market a good investment.
A cooling company can grow quickly and still spend too much cash. A grid contractor can have a record backlog and still face execution risk. A data-center REIT can report strong leasing while carrying a very different debt and capital-spending profile from an equipment manufacturer.
That is why comparing these companies with one familiar number, such as P/E or revenue growth, can be misleading.
The useful approach is to ask the same four questions, then use the right numbers for each business model.
A strong AI market is only the starting point. The company still has to turn demand into profit, cash, and enough value to justify the stock price.
After reading this article, you should be able to compare an equipment maker, grid contractor, power company, regulated utility, and data-center REIT without forcing them into the same financial template. You will also have a simple eight-number worksheet you can update after each earnings report.
Before the Eight Numbers, Ask Four Questions
- Is demand real?
Are customers placing orders, signing contracts, or leasing capacity? - Can the company turn that demand into profit and cash?
Growth matters more when margins and cash flow improve with it. - Can the balance sheet fund the growth?
Infrastructure expansion often requires factories, construction, land, or new power assets before the revenue arrives. - Is the stock price already assuming a very successful future?
A good company can still be a poor investment if the price already reflects unusually optimistic expectations.
1. Revenue Growth: Is the Business Actually Delivering More?
Revenue growth tells us how quickly the company is recognizing more sales from products and services it has delivered.
For companies that make acquisitions or operate globally, it helps to separate reported growth from organic growth.
Organic growth tries to show how the existing business grew after removing much of the effect of acquisitions, divestitures, and currency changes.
Vertiv provides a useful current example. In Q2 2026, the company reported net sales of $3.274 billion, up 24% from a year earlier. Organic sales growth was 18%; acquisitions added about five percentage points and currency added about one point.[1]
Why does the distinction matter?
Reported growth tells you what reached the financial statements. Organic growth helps you see whether the company’s existing operations are growing without relying mainly on acquisitions or currency movements.
Also ask what created the growth. Was it higher selling prices, more volume, a new acquisition, or one unusually large project?
2. Orders, Backlog, Contracts, or Leasing: Is Future Demand Visible?
Revenue mostly tells you what has already been delivered.
To look forward, use the demand measure that fits the business.
| Business | Useful leading indicator | What it tells you |
|---|---|---|
| Equipment maker | orders and backlog | equipment ordered but not yet fully delivered |
| Contractor | RPO and backlog | contracted or estimated future work |
| Power generator | contracted MW and contract length | future electricity sales covered by agreements |
| Regulated utility | approved capital plan and rate base | assets approved to earn a regulated return |
| Data-center REIT | bookings, leasing, presales, signed-not-opened rent | capacity contracted before all revenue begins |
Backlog means work that has been ordered or contracted but has not yet been fully delivered as revenue. It is often translated as a “ìˆ˜ì£¼ìž”ê³ ” in Korean financial writing.
RPO, or remaining performance obligations, is contracted future work that still has to be performed under accounting disclosure rules. RPO and backlog can overlap, but they are not always identical.
Quanta Services makes that distinction clear. At June 30, 2026, it reported $33.6 billion of RPO and $53.4 billion of total backlog.[2]
The larger backlog figure also includes estimated work under some master service agreements and short-term non-fixed-price contracts.
That is why you should read the company’s definition before comparing backlog across companies.
3. Book-to-Bill—or the Closest Equivalent: Is New Demand Arriving Faster Than Revenue?
Book-to-bill means new orders divided by revenue during the same period.
If the ratio is above 1.0, new orders arrived faster than the company recognized revenue. If it stays above 1.0 for several periods, backlog may continue to grow.
Siemens Energy’s Grid Technologies business reported Q3 FY2026 orders of €5.367 billion and revenue of €3.624 billion. Its book-to-bill ratio was 1.48, and backlog reached €51 billion.[3]
The same quarter also showed why this metric needs context. Revenue was up strongly, and the company said expanded production capacity supported the increase.
So a high book-to-bill ratio is a demand signal. It is not proof that every order will be delivered quickly or profitably.
If a company does not report orders, use the closest business-specific equivalent: new leasing for a REIT, contracted megawatts for a generator, or approved rate-base investment for a regulated utility.
4. Operating Margin: Is Growth Becoming More Profitable?
Operating margin is operating profit divided by revenue.
It shows how much operating profit the company keeps from each dollar of sales after normal operating costs under the company’s accounting definition.
Vertiv’s adjusted operating margin reached 22.6% in Q2 2026, up 410 basis points from the year-earlier quarter.[1]
A basis point is one-hundredth of a percentage point. So 410 basis points equals 4.10 percentage points.
The useful question is not whether Vertiv’s margin is higher than a utility’s or a REIT’s.
It is whether Vertiv’s own margin is improving as sales and workload grow.
Compare margins within the same business model and over time.
5. Free Cash Flow: Does Accounting Profit Become Cash?
A company can report growing earnings while cash is being absorbed by inventory, customer receivables, factories, or construction.
Free cash flow usually starts with operating cash flow and subtracts capital spending, although exact company definitions can differ.
Vertiv generated $925 million of adjusted free cash flow in Q2 2026. Quanta reported about $0.9 billion of free cash flow in the same quarter.[1][2]
These numbers should not be compared mechanically because the businesses and definitions differ.
Instead, ask whether each company’s cash flow is improving as revenue and profit grow.
If management reports an adjusted cash-flow measure, read the reconciliation to understand what was added back or excluded.
6. Debt and Financial Capacity: Can the Company Afford the Growth?
AI infrastructure can require expensive factories, data centers, power plants, transmission lines, or acquisitions.
Companies may fund that spending with operating cash, debt, new shares, joint ventures, or customer advances.
For an equipment manufacturer, useful measures include net debt, net debt to EBITDA, interest expense, and liquidity.
Net debt usually means debt minus cash and cash equivalents. If cash exceeds debt, the company can be described as having a net cash position.
Vertiv ended Q2 2026 with $5.6 billion of liquidity and a net cash position.[1]
That gives the company more room to invest in capacity or absorb a downturn than a highly leveraged balance sheet would.
But the right debt level depends on the business. A regulated utility or REIT normally carries more debt because it owns long-lived assets supported by regulated or contracted cash flows.
The question is not “Does the company have debt?” It is “Can the expected cash flow support the debt through a weaker part of the cycle?”
7. Capital Spending and Return on New Capital: Is Expansion Creating Value?
Capital expenditure, or capex, is money used to build, buy, or improve long-lived assets.
Capex is not automatically a problem. It can create the factories, data centers, and power assets that produce future revenue.
But investors should separate two broad types.
Maintenance capex keeps existing assets usable.
Growth capex adds new capacity that is expected to produce additional revenue or earnings.
Equinix shows why this distinction matters for a data-center REIT. In its updated 2026 guidance, the company expected $290–$310 million of recurring capital expenditure and $4.71–$5.69 billion of non-recurring capital expenditure, excluding xScale joint ventures, future M&A, and real-estate acquisitions.[4]
Equinix also guides investors to AFFO, or adjusted funds from operations.
AFFO is a REIT cash-flow-style measure that starts with funds from operations and adjusts for recurring capital needs and other items under the company’s definition. It is commonly used to judge dividend-paying capacity and recurring economic performance.
The deeper question is whether the return earned on new assets will be higher than the cost of the debt and equity used to finance them.
If a company spends heavily but the new assets earn weak returns, growth can destroy value rather than create it.
8. Valuation: How Much Success Is Already in the Stock Price?
The first seven numbers describe the business.
Valuation connects that business to the price investors are being asked to pay.
The correct valuation measure depends on the business model.
| Business type | Common valuation measures |
|---|---|
| Equipment maker | P/E, EV/EBITDA, free-cash-flow yield |
| Contractor | P/E, EV/EBITDA, free-cash-flow yield |
| Power generator | P/E, EV/EBITDA, free-cash-flow yield |
| Regulated utility | P/E, dividend yield, price-to-book |
| Data-center REIT | price-to-AFFO, NAV premium/discount, debt ratios |
P/E is price divided by earnings per share.
EV/EBITDA compares enterprise value—roughly the value of the operating business including debt—to EBITDA.
Free-cash-flow yield compares free cash flow with the company’s market value.
NAV, or net asset value, estimates the value of a REIT’s assets after liabilities.
Do not compare a data-center REIT’s price-to-AFFO directly with an equipment manufacturer’s P/E and call one “cheaper.” They measure different economic structures.
Also remember that valuation is the one number in this article that changes every trading day. Refresh it when you update the worksheet.
A Simple Eight-Number Worksheet
| Metric | Current | One year ago | Direction | Question to answer |
|---|---|---|---|---|
| 1. Organic revenue growth | Is the existing business growing? | |||
| 2. Orders / backlog / contracts | Is future demand visible and well defined? | |||
| 3. Book-to-bill or equivalent | Is new demand arriving faster than delivery? | |||
| 4. Operating margin | Is growth becoming more profitable? | |||
| 5. Free cash flow | Does profit become cash? | |||
| 6. Net debt / financial capacity | Can the balance sheet fund growth? | |||
| 7. Capex / return on new capital | Will new assets earn more than they cost? | |||
| 8. Valuation | How much future success is already priced in? |
Update the worksheet after each earnings report.
The goal is not to react to every quarter. The goal is to see whether the direction of the business is improving, weakening, or staying roughly the same.
Three Common Mistakes
Mistake 1: Using the Same Metric for Every Company
Backlog is important for a transformer maker or contractor. It is less useful for a regulated utility.
AFFO matters for a data-center REIT. Capacity factor matters for a power generator.
Start with the business model, then choose the metric.
Mistake 2: Treating Adjusted Numbers as if They Were Standardized
Companies often report both GAAP and adjusted measures.
For example, Constellation reported Q2 2026 GAAP net income of $1.42 per share and adjusted operating earnings of $2.55 per share.[5]
The two figures answer different questions because the adjusted measure excludes or reclassifies certain items under the company’s definition.
Do not reject adjusted numbers automatically. But read the reconciliation before using them.
Mistake 3: Ignoring the Price
A company can grow revenue, margin, and free cash flow and still produce a disappointing stock return if the starting valuation already assumed even stronger growth.
The opposite can also happen. A slower-growing company can produce an acceptable return when its cash flow is durable and the purchase price is low enough.
That is why valuation belongs at the end of the eight-number process, not at the beginning and not outside it.
What Reader Questions Changed This Article?
Recent investor discussions repeatedly ask the same practical questions: Should revenue growth or backlog matter more? How much capex is too much? Is free-cash-flow conversion keeping up? How do you know whether an infrastructure company is benefiting from real demand rather than simply trading on the AI theme?
Those questions are why this article does not stop at eight definitions.
The worksheet forces each company through demand, profitability, cash, financing, capital returns, and valuation.
The Main Idea
AI infrastructure includes very different businesses: electrical equipment, cooling, construction, power generation, utilities, and data-center real estate.
You should not expect all of them to have the same margins, debt levels, capital needs, or valuation multiples.
But the basic test is similar.
Demand has to become real sales or contracts. The company has to deliver profitably. The profit has to become cash. New investment has to earn enough to justify the capital used. And the stock price still has to leave room for a reasonable return.
The eight numbers do not tell you which stock to buy. They help you test whether the business story and the stock price are supported by the same reality.
Continue Reading
- Who Gets Paid in the AI Data Center Boom? — map the value chain first.
- How to Research AI Infrastructure Companies: From Orders to Cash Flow — go deeper into demand conversion.
- The Grid Bottleneck Behind the AI Boom — see how physical constraints affect delivery.
- Who Will Power the AI Boom—and Who Will Actually Profit? — compare power business models.
- 3 Real AI Infrastructure Stock Calculations: Backlog, Power, and Rate Base — apply the framework to worked examples.
Key Terms
- organic growth: growth in the existing business after removing much of the effect of acquisitions, divestitures, and currency
- backlog: ordered or contracted work not yet fully delivered; often translated as 수주잔ê³
- RPO: remaining performance obligations, contracted work still to be performed
- book-to-bill: new orders divided by revenue during the same period
- operating margin: operating profit divided by revenue
- free cash flow: cash remaining after operating needs and capital spending under the stated definition
- net debt: debt minus cash and cash equivalents
- capex: capital expenditure used to build, buy, or improve long-lived assets
- AFFO: adjusted funds from operations, a REIT performance measure that adjusts FFO for recurring capital needs and other items
- P/E: stock price divided by earnings per share
- EV/EBITDA: enterprise value divided by EBITDA
- NAV: net asset value, an estimate of asset value after liabilities
Sources
- Vertiv — Q2 2026 results.
- Quanta Services — Q2 2026 results.
- Siemens Energy — Q3 FY2026 results.
- Equinix — Q2 2026 results and updated guidance.
- Constellation — Q2 2026 results.
Status checked October 1, 2026. Financial measures are not standardized across all companies, and valuation changes with market price. Compare each metric with the company’s own definition, scope, and prior periods before comparing it with another business.