How to Read a Tech Multiple

Why tech companies don't trade on earnings

Price-to-earnings is the default valuation tool for most of the stock market. It breaks down completely for a certain class of technology company — not because those companies are exempt from economic reality, but because their earnings today are a terrible predictor of their earnings in five years. When a company is spending 40% of its revenue on sales and marketing and another 20% on R&D, the income statement looks like a money-losing operation. But that spending is buying future customers and future product. Strip it out and the underlying economics often look very different.

The market shifts to revenue-based multiples not to ignore profitability, but to get closer to it. EV/Revenue is a shorthand for a longer question: if this company stopped growing and started optimizing for cash, what would the free cash flow margin look like, and does the current price reflect a reasonable present value of that future cash? That's why gross margin matters so much. A business with 80% gross margins has a clear path to 25–35% FCF margins at maturity. A business with 40% gross margins has an arithmetic ceiling that makes those same FCF margins nearly impossible. Same revenue multiple — completely different implied bet.

Consider what happened with Snowflake after its 2020 IPO. The company was priced at over 100x revenue, which looked like fantasy until you looked at the gross margin trajectory. Snowflake was already running 60%+ product gross margins as a relatively young business, with a clear path toward 70–75% as scale improved. The revenue multiple was always a bet on that gross margin arc converting into durable FCF at scale — not a bet that revenue multiples don't matter. By 2023, as growth decelerated and the rate environment changed, that multiple had compressed dramatically. The business did not collapse. The bet repriced.

The three numbers behind every multiple

Three inputs drive what an EV/Revenue multiple "should" be for any given company: gross margin, revenue growth rate, and FCF conversion. Gross margin sets the ceiling — it's the maximum share of each revenue dollar the business can ever keep before corporate overhead and taxes. Revenue growth determines how long you're in the reinvestment phase before cash starts compounding. FCF conversion tells you whether the business actually translates growth into cash, or whether it leaks value along the way through working capital, stock-based compensation, or perpetual capital spending.

The arithmetic of a tech multiple works backward from a terminal assumption. If you pay 10x revenue today for a company, you're implicitly making a bet about what the business looks like in seven to ten years when growth normalizes. At a 20% FCF margin at maturity — a reasonable target for a well-run software business — a company generating $500 million in revenue today would produce $100 million in annual FCF. At a 10x revenue multiple, you're paying $5 billion today for $100 million in future steady-state FCF, which is a 2% yield on today's price. That's not obviously cheap. But if revenue is growing at 30% per year, the same company generates $1.6 billion in revenue in five years and $325 million in FCF at the same margin — a 6.5% FCF yield on today's price. Whether that's a good return depends entirely on whether that 30% growth rate holds.

This is why gross margin deserves more attention than it usually gets in casual discussions of tech valuation. It is not a secondary detail — it is the single most important structural constraint on what any revenue multiple implies. A 10x multiple on an 80% gross margin business requires much less faith in the future than the same multiple on a 45% gross margin business. The higher the gross margin, the shorter the path from "revenue growth company" to "free cash flow machine." The lower the gross margin, the longer and harder that path becomes, and the more that revenue multiple is demanding from the business.

FCF conversion adds a third dimension. Some companies with strong gross margins and fast growth still struggle to translate revenue growth into cash — because they need to keep re-spending on sales capacity, or because their customer acquisition costs are rising, or because stock-based compensation is large enough to matter. The cleanest businesses show improving FCF conversion as they scale. Trailing twelve-month FCF as a percentage of revenue, compared against gross margin and growth rate, tells you whether the business model is actually tightening or whether growth is masking structural inefficiency.

The Rule of 40 and why it matters to price

The Rule of 40 is the most widely used heuristic in software investing: add a company's revenue growth rate to its FCF margin (or operating margin, if FCF isn't clean). If the sum is 40 or above, the business is considered to have a defensible multiple. Below 40, especially below 25, and the multiple becomes harder to justify without a specific narrative to explain the gap.

The rule emerged as a practical screen for distinguishing companies that are spending aggressively but efficiently from companies that are simply spending. A business growing at 40% with breakeven FCF passes — it's buying growth at scale. A business growing at 20% with 20% FCF margins also passes — it's mature enough to generate cash while still expanding. A business growing at 15% with negative 5% FCF margins scores a 10 and is in trouble, because neither its growth nor its unit economics are strong enough to justify a premium multiple.

Datadog is one of the clearest examples of the Rule of 40 translating into multiple stability. Through 2021 and 2022, the company consistently scored above 50 on the Rule of 40 — growing at 60%+ with positive and improving FCF margins. Even as the broader SaaS multiple compression hit, Datadog's multiple held better than peers because the underlying business metrics kept justifying it. Companies that dropped below 40 during the same period — particularly those where growth slowed to the mid-teens without a corresponding improvement in profitability — saw sharper multiple compression, often 60–70% in enterprise value terms.

The rule has limits. It's a screen, not a model. It treats a point of growth and a point of FCF margin as equivalent, which they are not — high-margin cash is more durable than high-growth revenue. And it says nothing about the level of the multiple, only whether it has some structural support. A company scoring 60 on the Rule of 40 at 30x revenue is still priced for an exceptional outcome. The Rule of 40 tells you whether the multiple deserves serious consideration. It doesn't tell you what to pay.

The interactive: decode any multiple

The calculator below takes three inputs — a company's current EV/Revenue multiple, its gross margin, and its revenue growth rate — and shows you what that combination actually implies about long-run returns. Enter real numbers from a company you're researching. The prefilled defaults are roughly Datadog-like: a high-gross-margin software business with strong growth and a market multiple that reflects it. Change the gross margin to 45% and watch how the math shifts. That's the whole lesson.

Not all tech is the same: SaaS vs. hardware vs. semiconductors

"Tech" is a misleading bucket. Three very different business models live inside it, each with the appropriate valuation framework — and applying the wrong framework is one of the most common mistakes retail investors make when reading analyst coverage or financial press.

• SaaS and cloud software. Recurring subscription revenue, high gross margins (typically 70–85%), low capital intensity, and a customer base that churns slowly. This is the business model that justifies EV/Revenue multiples, because gross margin is high enough that a real FCF margin at maturity is structurally achievable. Salesforce, ServiceNow, Datadog, Cloudflare — these are companies where the path from revenue to cash is well-defined. EV/NTM Revenue (enterprise value divided by next twelve months revenue) is the standard shorthand.
• Semiconductors. Cyclical, capital-intensive, lumpy. Semiconductor companies have real earnings in up-cycles and real losses (or near-losses) in down-cycles. Using an EV/Revenue multiple for Nvidia or ASML or Micron misses the whole point — the relevant question is what earnings look like through the cycle, not just at peak. Forward P/E and EV/EBITDA on mid-cycle estimates are the standard tools. Micron in a downcycle trades at 2x revenue, not because it's cheap, but because revenue multiples don't capture the inventory write-downs and margin collapse that accompany the cycle trough. The stock can still be expensive at 2x revenue if the trough lasts longer than expected.
• Hardware and devices. Lower gross margins (20–50% for most hardware companies, with some exceptions), lumpy upgrade cycles, and intense competition from contract manufacturers that can replicate product at scale. Apple is the exception — its hardware business supports a premium multiple because services revenue has pushed blended gross margins to 45%+ and created an ecosystem lock-in that's structurally different from a traditional hardware business. Pure-play hardware companies without software attached typically trade on forward earnings or EV/EBITDA, and at much lower multiples than software peers — often 1–3x revenue even when business is good.

The practical discipline is matching the framework to the business model before reading the number. An analyst who covers Nvidia on EV/Revenue is using a tool that doesn't fit the business. An analyst who covers Salesforce on forward P/E only is ignoring the structural reason the P/E looks high — that most of the company's value is in future FCF, not this year's GAAP earnings. Read the 10-K first, classify the business, then decide which multiple to apply. Once you know the right multiple, benchmark it against the sector — our P/E ratio by industry reference shows live trailing and forward P/E for all 11 S&P 500 sectors versus their five-year averages.

Why multiples compress when rates rise

A high-growth tech company is, in financial terms, a long-duration asset. Most of its value doesn't exist today — it exists in the cash flows the business will generate in years five, eight, and twelve. To price those future cash flows in today's dollars, investors apply a discount rate. When that discount rate rises, future dollars are worth less today, and the multiple the market will pay for each dollar of current revenue falls. That's the full mechanism. There's nothing more complicated to it.

The magnitude of what happened between 2021 and 2023 was extraordinary. The median EV/NTM Revenue multiple for publicly traded SaaS companies — tracked consistently by several institutional investors through the Bessemer Cloud Index and similar benchmarks — fell from approximately 15x at the start of 2021 to under 5x by late 2022. Companies whose revenue kept growing through that entire period still saw their stock prices fall 60–75%. The underlying businesses didn't deteriorate. The discount rate applied to their future cash flows changed by several hundred basis points.

This is why rate sensitivity is not a vague concern but a specific quantitative input. The longer a company's cash flow is weighted into the future, the more it behaves like a long-dated bond, and the more it reprices when rates shift. A mature company with flat growth and most of its cash flow earned in the next three years has low duration — rates barely move its multiple. A company with 40% growth where most of its value is priced into cash flows beyond year seven has high duration. The same rate increase that barely touches the multiple of a regional bank can halve the multiple of a fast-growing cloud software company.

The corollary: when rates fell sharply in 2020, the same mechanism ran in reverse. Future cash flows became worth dramatically more in present-value terms, and multiples expanded far beyond what revenue growth alone could justify. Companies with no earnings and modest growth traded at 30–50x revenue because the discount rate on their distant cash flows had dropped to near zero. Understanding that the expansion was rate-driven, not purely business-driven, would have clarified the risk in those positions before the 2022 reversal. The valuation framework doesn't predict the timing, but it does tell you what you're actually buying.

Real-world example: Broadcom (AVGO) Q3 AI guide and FY2027 forward thesis → — Broadcom's AI semiconductor revenue (+143% YoY) trades at a premium multiple because the revenue is contractual co-design lock-in, not cyclical chip demand. The case illustrates why gross margin quality and revenue visibility justify different multiple frameworks for semiconductor companies operating in AI infrastructure.

Real-world example: Microsoft Q4 FY2026: Azure AI Revenue Inflection and the Multiple Re-Rate Question → — Microsoft's Q4 FY2026 post-earnings analysis tests whether Azure AI consumption now drives the overall Azure growth rate — the threshold that separates a “cloud compounder” multiple from an “AI infrastructure winner” re-rate.

Real-world example: NVIDIA (NVDA): Blackwell's Rack-Scale Transition and the $91B Floor → — NVIDIA's shift from discrete GPU sales (~$35–40K per unit) to GB200 NVL72 rack systems ($3.0–3.8M per rack) represents a ~100x ASP expansion that defies standard semiconductor multiple frameworks. The Q2 FY2027 guide of $91B strips China H20 revenue entirely, making it a floor rather than a forecast — a case study in reading guidance conservatism.

Real-world example: NVIDIA (NVDA) Q2 FY2027 Post-Earnings — Blackwell Data Center Result, Q3 FY2027 Guidance vs. $91B+ Consensus, and China H20 Optionality → — NVIDIA's Q2 FY2027 actual result tests whether the $91B guidance floor held and whether Q3 FY2027 data center guidance cleared the $91B+ Street consensus. China H20 commentary — $15B in analyst-estimated optionality sitting outside the guided revenue range — is the live case study in how a single policy variable can reprice a semiconductor multiple without touching reported earnings.

Real-world example: Advanced Micro Devices (AMD): Helios Rack Production Ramp, MI350 Data Center GPU Roadmap vs. NVIDIA NVL72, and the ROCm Software Gap → — AMD's Helios rack enters production as the first credible rack-scale alternative to NVIDIA's NVL72 ($3.0–3.8M per system). The case tests whether a second-source competes on hardware economics alone or whether NVIDIA's CUDA ecosystem is a switching-cost moat that Helios cannot break regardless of MI350 specs. ROCm parity — or the absence of it — is the variable that determines whether hyperscaler dual-sourcing translates into sustained AMD multiple re-rate or a floor at discounted GPU spot pricing.

Real-world example: Palantir (PLTR) Q2 2026: Rule of 40 at 155, U.S. Commercial +149% YoY → — Palantir's Q2 2026 result (93% revenue growth + 62% adjusted operating margin = 155 Rule of 40 score) is an extreme case in how software multiples compress or expand based on the growth-profitability combination. At $1.94B quarterly revenue, U.S. Commercial accelerated rather than decelerated — the data point that most directly tests the “high-duration growth company” multiple thesis.

Real-world example: Datadog (DDOG) Q2 2026: Revenue Re-Accelerates to 36% at $1.12B — Fastest Growth in Four Quarters → — Datadog's Q2 2026 result demonstrates why re-acceleration at scale changes the multiple calculus. At $1.12B quarterly revenue (+36% YoY), the growth rate was faster than four prior quarters — running against the base-effect compression that should suppress it. With FCF of $279M and $100k+ ARR customers rising 23% to 4,720, the Rule of 40 score stayed well above 50. The case shows how sustained growth-profitability combinations at meaningful scale compress valuation risk even when absolute multiples look high.

Real-world example: Cloudflare (NET) Q2 2026: AI Gateway as a Transaction-Layer Revenue Model → — Cloudflare's Q2 2026 post-earnings illustrates how AI Gateway inserts a transaction layer between enterprise applications and every LLM inference call routed through Cloudflare's network — a margin profile distinct from bandwidth-priced CDN or seat-based Zero Trust. At 250+ global PoPs, Workers AI creates an edge inference advantage neither hyperscalers nor SASE-only vendors can replicate. The case shows how AI platform positioning requires a different multiple lens than traditional SaaS growth metrics.

Real-world example: Snowflake (SNOW) Q2 FY2027: Product Revenue Re-Acceleration Test and NRR Stabilization → — Snowflake's Q2 FY2027 result tests whether the cloud data platform's consumption model can sustain 30%+ product revenue growth while NRR stabilizes above 126% — the inflection point between platform expansion and installed-base contraction. Cortex AI's contribution to LLM workload consumption is the multiple-repricing variable: if AI spend compounds on top of existing data-warehouse consumption, the EV/revenue premium is justified by a new growth layer; if it merely substitutes, NRR compression continues toward the 120% floor. The case illustrates why consumption-model SaaS requires a different multiple lens than seat-based ARR companies.

EV/Revenue Calculator — look up any stock live

Type a ticker below to pull live EV/Revenue data. The calculator compares the multiple against sector benchmarks so you can immediately see whether a growth stock's valuation is in line with peers or priced for an exceptional outcome. Cross-reference the result with gross margin and the Rule of 40 score above to complete the picture.

Questions worth asking

Keep reading

• How to read an income statement
• Free cash flow explained
• How to read a 10-K
• EV/EBITDA explained
• How to value a stock