SaaS Valuation Reset: AI Disruption and PE Writedowns

Introduction

Public SaaS multiples are not coming back to 2021 levels, and the 2021-2022 PE take-private cohort cannot exit at entry multiples under current market conditions. Thoma Bravo paid $6.4 billion to take customer experience software company Medallia private in 2021, modeled growth that never materialized, and is now reportedly working through a debt-for-equity restructuring that would mark a roughly $5.1 billion equity loss on the position. Orlando Bravo, the firm's founder, has publicly said his team "made a mistake" on Medallia: too aggressive on growth, too much paid at peak. Medallia is the loudest symbol of a much broader reset, not a one-off.

Public SaaS multiples have compressed roughly 60 to 70 percent from the 2021 peak. The BVP Nasdaq Emerging Cloud Index sat at roughly 15x to 20x next-twelve-months revenue during the 2021 frenzy and has spent most of 2024 to early 2026 in the 5x to 8x range, with a sharp leg lower in Q1 2026 as the market priced in AI as a structural threat to a meaningful slice of horizontal SaaS demand. The PE take-private cohort that bought through that 2021 to 2022 window, Anaplan, Coupa, SailPoint, Proofpoint, RealPage, Citrix, Medallia, and a dozen others, is mostly stuck at entry multiples the current market will not clear.

For the demand-side mechanism PE has built to manage this without crystallizing the losses, see the continuation vehicles and GP-led secondaries post. For the GP economics those vehicles reset and roll forward, see the carried interest explainer.

To understand the reset, you have to understand what was happening at the top. Through 2020 and 2021, public cloud software traded at multiples that were unprecedented in any sector outside of biotech speculation. The BVP Nasdaq Emerging Cloud Index, an equally weighted basket of leading public cloud companies maintained by Bessemer Venture Partners, sat at roughly 15x to 20x NTM revenue for the index aggregate, with the fastest-growing names trading well above 30x ARR and in some cases above 50x. Snowflake, Datadog, MongoDB, Bill.com, Asana, Confluent, HashiCorp, and dozens of others printed multiples that implied either a decade-plus of uninterrupted compounding or a willingness to overlook the discount-rate math entirely.

The Rule of 40 Era

The dominant valuation framework was the Rule of 40: a SaaS company's annual revenue growth rate plus its free cash flow margin should exceed 40 percent. Companies that cleared 40 traded at premium multiples; those that did not were penalized. The framework rewarded growth almost regardless of margin profile, because revenue compounding at 40 to 60 percent a year on a long-duration cash-flow profile dominated the discounted value of the firm at the low discount rates of the era.

Rule of 40

A heuristic for evaluating the operating quality of a software business: revenue growth rate plus free cash flow margin should be at least 40 percent. A company growing 50 percent with break-even margins (50 + 0 = 50) cleared it; a company growing 20 percent with 25 percent margins (20 + 25 = 45) also cleared it. During the 2021 peak the rule was used to justify premium multiples on names that prioritized growth over profitability, because the embedded assumption was that low discount rates and durable demand would let any Rule-of-40 compliant business grow into a defensible valuation.

The Rule of 40 was not wrong as a framework. It was the calibration that was wrong. At a 7 to 8 percent discount rate (CAPM with a 1.0 to 1.5 percent risk-free rate and a normal equity risk premium), a 40 percent grower with a long runway is worth a great deal of money. At a 9 to 10 percent discount rate (CAPM with a 4 to 4.5 percent risk-free rate), the same business is worth meaningfully less. Multiples that priced in the lower discount rate would compress mechanically when rates moved.

ARR Multiples and the Public-Private Gap

The other dominant metric was the EV / ARR multiple, the ratio of enterprise value to annual recurring revenue. The cleanest measure of how much investors would pay for a dollar of recurring software revenue, ARR multiples in 2021 routinely ran 15x to 25x on public comps for high-growth horizontal SaaS, and the private market often paid more.

The private-public gap was structurally elevated because two flywheels were spinning in the same direction. Crossover funds and growth equity vehicles (Tiger Global, Coatue, Insight Partners, General Atlantic, D1, Altimeter) were deploying record capital into late-stage software at marks set by reference to public comps. That kept private rounds priced at or above public, which gave PE sponsors looking at take-private targets a backdrop of high marks across the asset class.

Why So Many Take-Privates Happened at 18x-25x ARR

The 2021 to 2022 SaaS take-private wave was the largest concentrated PE deployment into a single sector since the buyout era began. Thoma Bravo, Vista Equity Partners, Silver Lake, Permira, Hellman & Friedman, KKR, Insight Partners, and Bain Capital collectively put hundreds of billions of dollars of equity to work on software take-privates. The economics depended on three assumptions, each of which has since broken:

First, that public SaaS multiples would normalize at roughly the 2021 levels rather than mean-revert lower. The thesis was that the cloud transition was structural and that the multiple was the new equilibrium, not a peak. Multiples have since compressed roughly 60 to 70 percent at the index level.

Second, that operating leverage and SaaS net-retention dynamics would let portfolio companies grow into their entry multiples even if public marks reset. The expectation was that 20 to 30 percent ARR growth, combined with margin expansion and net dollar retention well above 110 percent, would compound the business into the entry valuation within the hold period. Growth rates have decelerated faster than margins have expanded across most of the cohort.

Third, that leveraged-finance markets would remain accommodative, both for the entry debt package and for ongoing refinancings as the funds extended hold periods. The leveraged loan market that financed the take-privates is now repricing the same exposures at materially wider spreads, and several portfolio companies are running into covenant pressure as growth misses force lender renegotiations.

The combination is structural, not cyclical. The entry multiples assumed an environment that no longer exists. The portfolio companies have not grown fast enough to overcome the multiple compression. The lenders are unwilling to keep papering over the gap indefinitely.

The reset did not start in 2026. The compression has been running for three years, and the AI inflection in early 2026 was an acceleration of a trend already in motion. Three drivers carried most of the move.

The Rate Environment and Long-Duration Cash Flows

The most cited driver, and the one most easily quantified, is the discount-rate effect. SaaS businesses generate cash flows that are weighted toward the back end of the projection: high-growth horizontal SaaS in particular is valued primarily on the terminal value and the late-period cash flows, not on the near-term EBITDA. Long-duration cash flows are mechanically more sensitive to discount rate changes than near-term cash flows.

When the US 10-year Treasury yield moved from sub-1 percent in 2020 to roughly 4 to 4.5 percent by 2024 and stayed there, the cost of equity for high-growth SaaS rose by roughly 200 to 300 basis points (the rate move flows through directly into CAPM via the risk-free rate, and indirectly through the equity risk premium). That change alone can reduce the present value of a long-duration cash-flow stream by 25 to 40 percent depending on the growth and margin profile. The DCF math was unforgiving, and it showed up in the multiples first on the high-growth, low-margin names that had the most embedded duration.

For a refresher on how rates flow into discount rates and why long-duration assets repriced hardest, see our discount rate and WACC explainer and the WACC build guide.

Growth Slowdown as the SaaS TAM Matured

The second driver was that SaaS, after a decade and a half of category creation and budget shift away from on-premise software, started bumping into mature-market dynamics. The total addressable market did not stop growing, but the rate of expansion of enterprise SaaS spend slowed as the easy wins (CRM, HR, payments, marketing, communications, collaboration) hit higher saturation. Enterprises that had been adding new SaaS line items every quarter started consolidating their stack to cut vendor count and reduce the integration tax.

The flow-through to public-comp results was visible by mid-2023. The fast growers slowed from 40 to 50 percent into the 25 to 35 percent range; the mid-growers compressed from 25 to 35 percent into the high teens. Net dollar retention, the cleanest measure of how well a SaaS company expanded inside its installed base, drifted lower across most public names as customers either downgraded seat counts, consolidated vendors, or simply stopped expanding usage at the prior pace.

Net Dollar Retention

A SaaS metric expressing how a cohort of customers' annual recurring revenue changes from one period to the next, after accounting for upgrades, downgrades, and churn but excluding new customer additions. A net dollar retention of 120 percent means that on average the existing installed base spent 20 percent more this year than last; a retention of 95 percent means the installed base contracted by 5 percent. Net dollar retention above 110 percent has historically been a key marker of best-in-class SaaS quality; declines below 100 percent indicate the installed base is shrinking and growth has to be funded entirely by new logos.

Net Dollar Retention Pressure and Vendor Consolidation

The third driver, related but distinct from the broader growth slowdown, was the buyer-side consolidation push that started in earnest in late 2022 and accelerated through 2024. Procurement teams at large enterprises ran formal SaaS-stack rationalization exercises, often with consulting partners, and cut vendor count by 20 to 40 percent in many cases. The pressure compressed net dollar retention across the board and reset growth expectations for the next contract renewals.

The combined effect of rates, slowing growth, and consolidation took the BVP Cloud Index from roughly 15x to 20x NTM revenue at the 2021 peak to roughly 5x to 8x by late 2024, depending on the time window. That was the first leg of the reset.

The second leg arrived with the AI inflection. Through 2023, 2024, and 2025, AI was a tailwind story for SaaS: incumbents would embed LLMs into their products, charge more per seat, and ride the AI wave. The narrative held through most of 2025 as Microsoft, Salesforce, Adobe, ServiceNow, and others rolled out copilots and agent features and analysts modeled meaningful upside from AI monetization.

That narrative shifted sharply in late 2025 and early 2026. The release of more capable autonomous agent products from the major AI labs (including Anthropic's enterprise agent push in January and February 2026, OpenAI's competing agent suite, and Google's agentic offerings) reframed the relationship between AI and SaaS. The market started pricing in a more aggressive scenario: rather than incumbents capturing AI as a feature, AI agents would absorb portions of the workflow that horizontal SaaS used to own. The form of the threat was not that AI replaced any single SaaS category overnight; it was that AI created a credible substitute for the user-facing workflow in a meaningful share of horizontal SaaS, which is enough to compress multiples even before revenue declines actually appear.

The BVP Cloud Index dropped sharply in Q1 2026 as the implications were absorbed across the public comp set. Press coverage estimates suggest roughly $1 trillion in aggregate enterprise SaaS market capitalization was erased across the six weeks bracketing the AI agent releases, with the largest hits taken by horizontal SaaS names exposed to the workflow-substitution narrative. The index aggregate moved into the high single digits on NTM revenue, with the most exposed names trading at 3x to 5x revenue, and the most defensible (infrastructure, vertical, and AI-native names) holding multiples in the low double digits.

Against that backdrop, the Medallia situation is the cleanest case study in how the 2021 take-private thesis broke. Thoma Bravo announced the take-private in mid-2021 at $34 per share, an enterprise value of $6.4 billion. The thesis at the time was that Medallia was an under-monetized customer-experience platform with significant operating-leverage opportunity once removed from quarterly public-market pressure. The acquisition closed in late 2021 at the top of the multiple cycle, financed with a mix of sponsor equity and a substantial leveraged loan package.

The Medallia trajectory is not a unique mistake. Orlando Bravo has been unusually direct about that in public comments since the position turned, calling out the diagnosis at Thoma Bravo's March 2026 investor meeting in Miami.

The framing matters because Bravo went further than the headline quote: he attributed the mistake to modeling too high a growth rate and paying too much at peak SaaS multiples, and used the moment to argue more broadly that AI-exposed software names deserve a valuation cut from where they were marked in 2021. That is the exact dynamic now showing up across the take-private cohort. The debt service is a fixed obligation; the equity is the residual; the residual gets crushed when the asset reprices below the debt level. The same logic is at work, to varying degrees, across most of the 2021 to 2022 SaaS take-private wave.

The Broader 2021-2022 PE Take-Private Cohort

The Medallia situation is a leading indicator, not an outlier. The 2021 to 2022 SaaS take-private wave produced a set of comparable exposures across the major buyout sponsors. Thoma Bravo's Anaplan ($10.7B, 2022), Coupa ($8B, 2022), SailPoint ($6.9B, 2022), Proofpoint ($12.3B, 2021), RealPage ($10.2B, 2021), and Medallia ($6.4B, 2021). Vista Equity Partners' Citrix ($16.5B with Elliott, 2022), Avalara ($8.4B, 2022), and KnowBe4 ($4.6B, 2023). Silver Lake on multiple software targets. Permira on Mimecast ($5.8B, 2022) and others. Hellman & Friedman, Bain Capital, and KKR each have their own equivalents.

The cohort is large enough that the exit path matters more than any single outcome. Strategic acquirers in 2026 are not paying anything like 2021 entry multiples. Public IPO markets reopen selectively, and a return-to-IPO at a meaningful discount to the take-private price is a mark-down, not a win. That leaves three options. Sponsor-to-sponsor secondary buyouts, which are happening but at compressed multiples. Continuation vehicles, where the sponsor sells the asset to a new fund it raises alongside secondary buyers and resets the hold period (covered in depth in our continuation vehicles and GP-led secondaries post). And restructurings, which is what is unfolding at Medallia.

The flow-through to TMT IB deal flow is direct: sponsor M&A on the 2021 take-private cohort is shifting toward GP-led secondaries advisory, debt-for-equity workouts, and NAV-loan-backed dividend recaps rather than traditional sale processes. Banks with strong GP-led secondaries practices (Lazard, Evercore, Jefferies, PJT Park Hill, Campbell Lutyens) have been the principal beneficiaries.

Master interview fundamentals: Practice 1,000+ technical and behavioral questions with AI-powered feedback, including TMT and PE sector questions like the SaaS reset above. Download our iOS app for comprehensive interview prep.

The methodology stack used to value SaaS during 2021 has not been thrown out; it has been recalibrated, and the implicit assumptions inside each metric are being questioned harder than they were a few years ago.

BVP Cloud Index

The Bessemer Venture Partners (BVP) Nasdaq Emerging Cloud Index is an equally weighted index of leading public cloud software companies, maintained by Bessemer in partnership with Nasdaq and tracked as the BVP Cloud Index. Published metrics include aggregate enterprise value to next-twelve-months revenue, growth rates, and Rule of 40 scores. The index is the most widely cited benchmark for public-market SaaS valuation and is referenced by sponsors, bankers, and equity research as the SaaS multiple barometer.

Multiples That Used to Anchor the Sector

The dominant 2021 metrics were EV / ARR (or NTM revenue) for high-growth names and Rule of 40 for quality screening. Both are still used, but with materially different ranges and a much higher bar for premium multiples.

EV / NTM revenue at the index level has moved from 15x to 20x at the peak into a range of roughly 5x to 10x depending on the cut and the time window, with sub-sector dispersion widening sharply. The fastest-growing infrastructure names (Snowflake, Datadog at their best vintages) still command low double-digit revenue multiples; mature growth-stage horizontal SaaS sits in the mid single digits; the most exposed legacy and contact-center categories trade closer to 3x to 5x.

Rule of 40 is still used as a quality screen, but the threshold has effectively risen. A 40-score that was a premium signal in 2021 is now table stakes for being valued at a market-aggregate multiple. Premium multiples now require Rule of 50 or higher, often combined with infrastructure positioning, AI-native architecture, or genuine vertical-SaaS defensibility that limits the horizontal AI threat.

What's Replacing Them

The 2026 SaaS valuation conversation has incorporated several adjustments to the legacy stack:

Sub-Sector Divergence Inside SaaS

The most important shift in the 2026 methodology is the explicit recognition that SaaS is no longer a single sector for valuation purposes. The dispersion across sub-sectors has widened to the point that an aggregate "SaaS multiple" is misleading.

The interview-room implication is that anyone discussing "SaaS valuation" in 2026 needs to specify the sub-sector before quoting a multiple. The variance inside SaaS is now wider than the variance between SaaS and many traditional verticals.

The reset has flow-through effects across the three groups that built businesses around the 2021 SaaS environment.

TMT IB Deal Flow Shift

For TMT investment banking, the deal-flow mix has shifted away from traditional sponsor M&A on the 2021 cohort and toward restructuring-adjacent advisory work: GP-led secondaries on portfolio companies that need a hold-period extension, debt-for-equity workouts on portfolio companies that have hit covenant pressure, NAV loans for sponsors that need DPI relief without crystallizing the loss, and selective strategic M&A where a credible acquirer can underwrite the cash flows at current rather than peak multiples.

For analysts joining or interviewing into TMT in 2026, the practical implication is that the work coming across desks reflects a different deal mix than the headline narrative suggests. Banks with strong secondaries and restructuring capabilities have been the principal beneficiaries; pure sponsor-M&A boutiques have seen volumes contract. Our TMT investment banking guide covers the broader sector landscape; the SaaS reset is the most acute force inside that landscape in 2026.

Founding and Funding Economics for the Next Cycle

For founders, the reset has reset Series A and B valuation marks materially lower. Where a 2021 Series B for a SaaS company at $5 million ARR might have priced at a $200 million post valuation (40x ARR forward), the 2026 equivalent typically prices at $60 million to $100 million post (12x to 20x ARR forward), with a higher bar on capital efficiency, gross retention, and a credible AI-defensible product narrative.

The IPO revenue threshold has also moved. Where $100 million in ARR with 50 percent growth was a plausible IPO candidate window in 2021, the 2026 IPO market requires roughly $200 million-plus in ARR, free cash flow positive or close to it, and a Rule of 50-plus profile to clear the public-market hurdle. Several 2026 IPOs that priced in the first half of the year illustrate this: the IPO window has reopened selectively, but on much more demanding metrics.

The PE Liquidity Problem

The third implication is the most important for PE-focused candidates: the SaaS reset has been the single largest contributor to the broader PE liquidity problem of 2024 to 2026. Sponsors with substantial SaaS exposure are sitting on funds where the marks have been reset below entry, exit options at the entry valuation do not exist, and LPs are putting pressure on distributions (DPI) at exactly the moment when realizing exits crystallizes the loss.

The industry response has been a structural shift toward continuation vehicles (the GP raises a new fund to buy the asset from the prior fund at a market-clearing price, resetting the hold period), NAV loans (the GP borrows against the unrealized portfolio NAV to fund a distribution without selling), and selective writedowns where the sponsor accepts the mark and works through the restructuring. Each of these is covered in depth in our continuation vehicles post; the SaaS reset is the demand driver that has made the secondaries market explode.

Get the complete guide: Download our comprehensive 160-page PDF covering valuation, M&A, LBO mechanics, and the interview frameworks recruiters expect in 2026, access the IB Interview Guide.

In a 2026 TMT or PE interview, "tell me about the SaaS reset" or "walk me through what's happening with SaaS multiples" is a real question, and it is the question that separates candidates who have done the reading from those who are recycling 2021 framings. Three representative exchanges.

"Why are SaaS multiples down so much?"

Three drivers. First, the rate environment: long-duration cash flows repriced as the 10-year Treasury moved from below 1 percent to around 4 to 4.5 percent, compressing the present value of late-period cash flows materially. Second, a growth slowdown as the SaaS TAM matured and buyer-side vendor consolidation cut net dollar retention across the public comp set. Third, the AI inflection in late 2025 and early 2026, where autonomous agents repositioned a meaningful share of horizontal SaaS as substitutable rather than defensible, with the most exposed sub-sectors taking the worst compression. The combined effect has taken the BVP Cloud Index from roughly 15x to 20x NTM revenue at the 2021 peak to mid-single-digits to low-double-digits depending on the cut and the time window. It is a structural reset, not a cyclical dip.

"Pick a PE-owned SaaS company and walk me through what went wrong with the deal."

Medallia is the cleanest case. Thoma Bravo paid $6.4 billion in 2021 at peak SaaS multiples for a customer-experience platform that is now exposed to both horizontal-SaaS multiple compression and AI substitution risk in user-facing workflow. The sponsor is heading toward a debt-for-equity restructuring that by press estimates marks a roughly $5.1 billion equity loss. Orlando Bravo has publicly said they "made a mistake" on the deal. The deeper point is that Medallia is not an isolated mistake; it is the most visible symbol of a 2021 to 2022 take-private cohort that bought at peak multiples and cannot exit at anything close to entry valuations under current market conditions.

"How does AI actually affect SaaS valuation?"

Two channels. First, demand-side substitution: autonomous agents collapse parts of the user-facing workflow that horizontal SaaS used to own, which lowers the per-seat justification for the incumbent and reduces the long-term ARR growth runway. Second, multiple repricing: investors stop being willing to underwrite ten years of compounding ARR at the prior pace, so the multiple compresses even before revenue impact actually shows up. The substitution effect is uneven: vertical SaaS, infrastructure SaaS, and AI-native software have held multiples; horizontal user-facing SaaS (customer experience, contact center, mid-market collaboration) has compressed hardest. The right way to frame it in an interview is "AI is a sub-sector divergence, not a uniform discount."

The pattern across all three answers is the same: name the drivers, name the data, name a specific deal, and acknowledge the sub-sector divergence. Anyone reciting "AI killed SaaS" is signaling they have not done the work; anyone reciting "SaaS is fine" is signaling the same. The 2026 framing is structural reset plus sub-sector dispersion.

A short list of the recurring errors in candidate framings, ordered by how often they show up in technical and behavioral interviews.

The other recurring errors:

The SaaS valuation reset is one of the largest sector-level repricing events of the last decade. It is not finished, the AI sub-sector divergence is still being absorbed into the multiples, and the PE workout cycle on the 2021 to 2022 take-private cohort will run for several years. Medallia will not be the last writedown of size, and continuation vehicles and NAV loans will not be the last tools the industry deploys to manage the exit problem.

For candidates interviewing in 2026, the reset is the question. TMT investment banking interviewers will ask about it, PE interviewers will ask about it as part of the broader liquidity-problem conversation, growth equity and private credit interviewers will ask about it in the context of how the deal pipeline has changed. The answer is the framing in this post: name the drivers, name the data, name the deal, acknowledge the sub-sector divergence, and resist both the "SaaS is dead" and the "SaaS is fine" simplifications. The 2026 market is a more interesting one to discuss than either of those caricatures.

For deeper reading: our continuation vehicles and GP-led secondaries post covers the mechanism the industry is using to manage the SaaS exit problem. Our exit strategies post covers the broader menu of PE exit options. The common valuation multiples reference and enterprise value vs equity value guide cover the multiple frameworks underlying the conversation. And the TMT investment banking guide puts the SaaS reset in the broader sector context that recruiters will expect candidates to be fluent in.