Post-SaaS Economics: The Premium Is Gone. Discipline Is the New Multiple.

For the first time in the SaaS era, public software trades at a forward earnings multiple at or below the S&P 500. That sentence would have been unthinkable to a 2021 board. It is the working assumption of an April 2026 one.

Per SaaStr’s reading of a Bloomberg chart, software’s forward P/E has walked down a staircase: roughly 84.1x in 2020–2022, 43.2x through mid-2024, 33.6x into mid-2025, 31.2x through year-end, and 22.7x for January–March 2026. We flag the trajectory as SaaStr’s. The underlying FactSet or Bloomberg series is not independently sourced. But the direction is corroborated by something harder to wave away. The iShares Expanded Tech-Software ETF (IGV) is down 27.5% year-to-date and 29% from its September 2025 peak, per BlackRock data as of April 9, 2026. That is not a rotation. That is a repricing.

If you are a CFO or a board member reading your 2026 software line items, the right question is not “is this a buying opportunity?” It is “what was the market paying a premium for, and why did it stop?”

The premium was never about the product

For twenty years, being labeled “SaaS” conferred a multiple. Recurring revenue, predictable cohorts, high gross margins, negative-churn dynamics. The premium was not paid for software. It was paid for the shape of the cash flows.

That shape had one load-bearing assumption: customer seats would keep expanding. Net revenue retention above 120% was not a sales achievement. It was a consequence of your customers hiring.

AI agents break the assumption. Not tomorrow, not uniformly, but enough that the market will no longer pay for it in advance. We covered the contract-level mechanics in Three Prices for One Agent: the per-seat chassis is already cracking inside Salesforce’s own pricing sheet. We laid out the deeper structural move in The $6-to-$1 Ratio. When AI can deliver outcomes at software margins, the category the multiple used to describe gets redrawn.

Now the market has stopped waiting. That is what 22.7x means.

Before you blame AI, engage the rates counter-thesis

A CFO at this point raises a hand. “The compression started in 2022, before any of this AI narrative mattered. Software is a duration asset. Real rates went up, multiples came down, private credit is forcing selling in leveraged names. This is the cycle, not the model.”

That is largely correct for the 2022–2024 leg. It is only partly correct for 2025–2026, and the cleanest way to see why is to look at NVIDIA.

NVIDIA is the company the market has named as the AI winner. Its forward P/E also compressed in this window, from the low-30s to roughly 20. If the software de-rating were a pure AI-disruption story, the AI winner’s multiple should have expanded. It contracted. A meaningful share of what has happened to software is duration and rates, not business-model collapse.

The broader market agrees. Goldman Sachs’ 2026 S&P 500 target is 7,600, up 13.5% from here, with the IT sector expected to deliver roughly 87% of Q1 2026 EPS growth. The index is not pricing a software apocalypse.

What remains after you subtract the rates story is dispersion. AI infrastructure names are holding earnings expectations. Application SaaS names whose revenue is indexed to seat count are not. Orlando Bravo of Thoma Bravo, the largest dedicated software buyer in private equity, has said publicly in both a January 2026 Semafor interview and a March 17 CNBC appearance that AI-driven valuation cuts in software are “very warranted.” When the biggest check-writer in the category says the haircut is deserved, that is not noise.

Hedge the macro. Manage the micro. Both are happening at once, and neither cancels the other.

The structural part the rates bulls miss

Here is where we part company with the “this is just 2022 again” view.

Even if real rates fall and multiples reflate, the underlying revenue architecture is shifting from per-seat to per-outcome. That is structural. A CFO who waits for rate relief and renews seat-based contracts unchanged is balance-sheet hedged and income-statement exposed. The multiple may recover. The contracts underneath it will not.

We are not re-arguing that AI amplifies existing organizational dynamics. The Amplifier Effect owns that claim. But John Cutler’s operator vocabulary is worth borrowing: AI does not fix the funnel, it amplifies whatever funnel you already have. Read “funnel” as “commercial model.” If your commercial model depends on customer headcount growing, AI will make it worse faster. If it is indexed to outcomes your customer cares about, AI will make it better faster. The multiple is the market trying to tell those two groups apart.

What “operating discipline” actually means for a board

When the growth premium goes away, operators do not earn multiple expansion back with a better deck. They earn it with four numbers that a board should already have in front of them every quarter.

Gross revenue retention. Before expansion, before upsell, before any accounting sleight. The rate at which last year’s dollars showed up this year without you fighting for them. Below 90% is a leak. Below 85% is a fire.

Net revenue retention, stripped of price increases. Inflation-era NRR benefited from list-price hikes customers tolerated. Strip them out. What remains is your real expansion engine. If that engine runs on seat growth, mark it down.

Free cash flow margin. Not adjusted EBITDA. Not non-GAAP operating income. Cash. The market stopped giving credit for earnings quality narratives in 2023. It is not coming back.

Rule of 40, recomputed honestly. Growth rate plus FCF margin. Forty is the Mendoza line. The interesting number is not whether you are above it, but which of the two components is doing the work. A 20/20 company and a 35/5 company are not the same company.

These are not new metrics. They are the metrics the premium let people stop looking at. The job of a board in 2026 is to look at them again and require a plan for each that assumes the growth tailwind is gone.

What this means for your 2026 software line

If you are the buyer: your vendors’ multiples have just been cut roughly in half. Your leverage in renewal conversations is structurally better than it was twelve months ago, best for contracts indexed to seats you can credibly retire. Use it deliberately, not punitively. The vendors who survive this window are the ones you will want to be working with in 2028.

If you are the seller: the pitch deck that worked in 2023 is now a liability. Growth narratives that assume customer headcount expansion will earn a polite nod and a multiple that says the board did not believe you. Rebuild the story around the four numbers above. Treat them as the product.

We made a version of this argument about capital flows in The $3 Trillion Stress Test and about value-chain economics in The Inverted Economics of the AI Value Chain. This piece is the market-signal anchor for both. The thesis that software’s multiple was built on a seat assumption is no longer a thesis. It is pricing.

The honest headline

Software is not dead. Gross margins are still 70 to 80% against an S&P 500 blend around 45%. A durable software business at 22x forward is, on paper, cheaper than it has been in the modern era, and this may yet be a generational buying opportunity for the companies on the right side of the dispersion.

But the premium for being SaaS is gone. The multiple has to be earned the old-fashioned way now. Operate well enough that a skeptical board sees it in four numbers without a narrative wrapper.

That is the quiet part of post-SaaS economics. The market stopped paying in advance for a shape of cash flow and started paying in arrears for the operating discipline that produces it.

Hedge the macro. Manage the micro. The micro is the part you control.

This analysis synthesizes The SaaS Rout of 2026 by Jason Lemkin (SaaStr, March 2026), iShares IGV ETF data (BlackRock, April 2026), Orlando Bravo on software AI valuation cuts (CNBC, March 2026), ‘Wait for it to pop’: Orlando Bravo on the AI bubble (Semafor, January 2026), and TBM 415: Demand, Mix Shaping and AI by John Cutler (March 2026).

Victorino Group runs operating-model assessments for CFOs and boards rebuilding the four numbers that now carry the multiple. Let’s talk.