The Paranoidist | Issue #2 By Paul Morin | February 14, 2026
Here are two things you watched happen this month. And here are some connecting points and implications you may not have considered.
Thing one: The SaaSpocalypse. In the first week of February 2026, approximately $285 billion in market value evaporated from software, data analytics, and financial data companies in 48 hours. Jefferies' equity trading desk coined the term. "Trading is very much 'get me out' style selling," said Jeffrey Favuzza. The iShares Expanded Tech Software ETF fell 28% from its recent high. The S&P North American Technology Software Index posted a 15% decline in January, its worst month since October 2008, and the selling intensified in February. Individual names were worse: Asana down 92% from its all-time high, DocuSign down 85%, with Salesforce, ServiceNow, HubSpot, Atlassian, Workday, and Adobe all hit hard. The trigger, based on Claude/Anthropic offering a new legal assistant that works with its foundational model, was the growing realization that agentic AI doesn't just augment enterprise software. It threatens to replace it. Palantir's CTO told analysts that its AI platform could now power complex SAP ERP migrations "in as little as two weeks," work that previously took years. A Palantir customer offered the line that should have kept every SaaS CEO awake that night: "We've gone all-in so much so that every other software must justify its existence, and so far they haven't been able to."
Thing two: The entire financial architecture for valuing, funding, buying, and lending against software companies is built on a single premise: that recurring revenue recurs. Annual recurring revenue, ARR, is the metric around which everything else orbits. Venture capitalists invest at a multiple of ARR. Growth equity firms price rounds at a multiple of ARR. Private equity firms finance leveraged buyouts against projected ARR. Private credit lenders underwrite loans based on the predictability and durability of ARR. Business development companies package those loans and sell them to investors, including, increasingly, retail investors in retirement accounts. Public markets value software companies on a multiple of ARR that reflects the market's confidence in future growth. Every link in this chain depends on the same foundational assumption: that the "recurring" in "annual recurring revenue" is reliable.
Now put them together.
This issue is about what happens to an entire asset class, and the financial ecosystem built on top of it, when the foundational assumption underneath the valuation model breaks.
The Chain That's Breaking To understand why the SaaSpocalypse is not a correction but a potential regime change, you need to understand how software valuations actually work, because the mechanism that built them is the same mechanism that can unwind them.
Software companies command premium valuations for specific, structural reasons. Recurring revenue is predictable: customers typically sign annual or multi-year contracts, churn rates are measurable, and net revenue retention above 100% means the customer base generates growth even before new sales. Gross margins are high, typically 70-85%, because the marginal cost of serving an additional customer is low. Per-seat pricing scales with customer headcount: as your client's company grows, your revenue grows with it, automatically. These characteristics made SaaS the preferred asset class for venture capital, growth equity, and private equity over the past decade. They are not incidental features. They are the reason the entire financial architecture exists.
Every link in the valuation chain depends on them.
At seed stage, a venture capitalist might invest at 50x to 100x ARR, which sounds absurd until you understand that they're betting on the trajectory: if ARR grows from $500,000 to $50 million over seven years, and the exit multiple is 10x, the math works. At Series A, the multiple compresses to 30x-50x. Series B, 15x-25x. By the time the company reaches a growth round or IPO, the valuation multiple must converge with public market comparables, which in the "old normal" of 2014-2019 ran 6x-10x ARR, and during the 2021 peak reached 25x at the 75th percentile.
The model works, elegantly, when two conditions hold. First, ARR keeps growing. Second, the exit multiple stays stable or expands. These conditions held, more or less, for over a decade.
Both are now under simultaneous assault.
The Double Hit Start with the multiple.
The SaaS Capital Index, which has tracked public SaaS company ARR multiples since 2008, documents four distinct valuation regimes. The "Discovery" period before 2014. The "Old Normal" of 2014-2019, when multiples ranged from 6x to 10x. The 2020-2021 peak, when the median reached 15x and the 75th percentile hit 25x. And the "New Normal" of 2023-2025, when multiples settled back to 6x-8x.
The SaaSpocalypse may be creating a fifth regime. Post-selloff, many formerly high-flying SaaS companies are clustered below 5x revenue. JP Morgan analyst Toby Ogg captured the mood: "The sector isn't just guilty until proven innocent but is now being sentenced before trial." Piper Sandler downgraded Adobe, Freshworks, and Vertex specifically because "seat-compression and vibe coding narratives could set a ceiling on multiples." The most vivid quote came from Jefferies' Favuzza: "The draconian view is that software will be the next print media or department stores, in terms of their prospects."
Whether the draconian view proves correct is, in a sense, beside the point. The damage to valuations comes not from certainty of disruption but from uncertainty about the range of outcomes. As Thomas Shipp at LPL Financial put it: "The range of outcomes for their growth has gotten wider, which means it's harder to assign fair valuations or see what looks cheap." When the confidence interval around future revenue trajectories explodes, rational investors do one of two things: demand a much larger risk premium (compressing multiples) or exit the position entirely. Both responses produce the same result for every entity that valued its holdings based on the old multiple range.
Now add the second hit: the "recurring" in recurring revenue.
The entire premium for SaaS as an asset class rested on the argument that recurring revenue is sticky, predictable, and compounds. When AI agents can replace per-seat subscriptions, the "annual" in "annual recurring revenue" becomes uncertain. This is not a future hypothetical. In the current earnings season, only 67% of software companies in the S&P 500 beat revenue expectations, compared to 83% for the broader tech sector. Median public SaaS revenue growth had already fallen to 12.2% by Q4 2025, with forecasts pointing to further deceleration. Customers are reducing seats. The metric that was supposed to be the bedrock of the valuation model is eroding.
And the pricing model transition that follows makes it worse, not better. IDC forecasts that 70% of software vendors will refactor pricing by 2028, shifting from per-seat subscriptions to usage-based, outcome-based, or platform fee models. This shift is necessary for survival, but it carries a valuation consequence that nobody is talking about: usage-based revenue is inherently less predictable than contracted seats. The entire ARR framework was built on the predictability of annual contracts. When pricing shifts to consumption, the "recurring" quality of the revenue degrades. Which logically compresses the multiple further.
This is the double hit. A lower multiple applied to a shrinking, or at minimum less predictable, revenue base. It is not a correction. It is a structural repricing of the foundational assumption.
The Cascade If the SaaSpocalypse were confined to public equity markets, it would be a painful but manageable repricing. Stocks fall. Investors take losses. Markets adjust. Life goes on.
But the SaaSpocalypse is not confined to public equities. The same valuation architecture that priced public SaaS companies extends, through a series of interconnected financial linkages, into venture capital, private equity, private credit, business development companies, and ultimately into the retirement accounts of people who have never heard the word "SaaSpocalypse." And at each link in the chain, the repricing is slower, less transparent, and potentially more destructive than in public markets.
Start with venture capital.
Harry Stebbings, one of the most prominent voices in the VC ecosystem, recently wrote: "We have a big problem. The venture model doesn't work with the current public market revenue multiples." The math is straightforward. The VC model depends on exit multiples. If exit multiples for the dominant asset class in venture portfolios (software absorbed roughly $172 billion in VC in 2025, more than half of all venture capital invested) are permanently reset 40-60% lower, the return profile of entire fund vintages changes. LPs receive smaller distributions. Fewer LPs re-up for the next fund. Less capital flows to the next generation of startups. The ecosystem contracts.
The data from SVB's 2026 State of the Markets report makes this concrete. The median revenue at IPO in 2025 was $246 million. Median operating margin was negative 16%. Companies were an average of 12.5 years old and had raised a median of $400 million in equity before going public. After 12.5 years of building, $400 million in equity raised, and $246 million in revenue at IPO, only about half of the 2025 IPO cohort was trading above their last private valuation. Less than a third were above their initial public market value.
Read those numbers again. The IPO, the event the entire venture model is built around, has become a markdown event. The last private round investors, the growth equity and late-stage VCs, are underwater in the majority of cases. And that was before the SaaSpocalypse compressed public multiples further.
The venture market's response has been bifurcation, not adjustment. In 2025, the top 1% of companies by valuation captured a full third of all VC capital, while the bottom 50% received just 7%. AI companies command valuation premiums of 10% at seed to 222% at Series D compared to non-AI peers. Capital is flooding into AI-native startups while non-AI SaaS companies, even well-run ones, are being starved. But this concentration creates its own fragility. As one investor at QED noted: "When momentary market gaps are not rapidly turned into long-lasting moats, it often leads to businesses with no enterprise value." The capital is flowing. Whether the valuations it is creating are any more durable than the ones it is destroying is an open question.
Now follow the chain to private equity.
Many of the largest software buyouts in recent years were financed at peak valuations during 2021-2022, when the combination of low interest rates and high growth expectations pushed deal prices to historic highs. When rates rose and the traditional syndicated loan market seized up, private credit stepped in to fill the financing gap. This means that a significant number of PE-owned software companies are sitting on acquisition debt that was underwritten against revenue assumptions that are now in question, at valuations that may never be recovered.
These companies are getting harder to sell. Not simply because the PE firms overpaid (though many did), but because the buyer universe has narrowed: investors are interested only in AI-related stories. The exit pathway that was supposed to generate returns for PE fund investors is constricting at both ends: lower multiples on any eventual sale, and fewer interested buyers.
Now follow the chain one more link, to private credit.
This is where the systemic risk lives, and where almost nobody is looking.
The Hidden Leverage Enterprise software companies have been a favored sector for private credit lenders since 2020. The attractions were exactly the characteristics that made SaaS a preferred asset class for equity investors: sticky revenue, high margins, predictable cash flows. Many of the largest unitranche loans in private credit history went to software and tech companies. According to 9fin's quarterly data, IT and communications account for 20-25% of private credit deal flow. Per Barclays, public BDCs issue more loans to software firms (20% of total investments) than to any other single sector.
Those loans were underwritten against the same foundational assumption that underpins every other link in the chain: recurring revenue recurs. The lending models assumed that SaaS revenue is durable, that churn is manageable, that the growth trajectory is predictable enough to service debt. The collateral, in effect, is the predictability of ARR.
When that predictability comes into question, the entire credit thesis weakens. And the data suggests it is coming into question now.
UBS credit strategist Matthew Mish published a report on January 26 estimating that 25-35% of private credit portfolios face heightened risk from AI disruption. Under UBS's "aggressive disruption" scenario, default rates for private credit could increase to 13% in 2026 if AI adoption outpaces borrowers' ability to adapt. To put that number in perspective, it is more than triple the stress projections for high-yield bonds (4%). UBS's base case is a 2-percentage-point increase in defaults to approximately 6%, which is still a significant deterioration.
The market is already moving. Apollo cut its direct lending funds' software exposure almost by half in 2025, from about 20% to roughly 10%. Private equity firms including Arcmont and Hayfin are hiring consultants to audit their portfolios for AI-vulnerable businesses. Shares of Blue Owl, TPG, Ares Management, and KKR fell by double-digit percentages in a single session during the SaaSpocalypse selloff. As of this week, Bloomberg reports that bond dealers are demanding higher compensation to trade corporate bonds issued by BDCs, reflecting growing unease about their software exposure.
There are two features of private credit that make this exposure particularly dangerous.
First, opacity. Private credit does not mark to market daily the way public equities do. Loan valuations are updated quarterly at best, and the methodologies for those valuations are at the discretion of the fund manager. This means that losses can accumulate invisibly. The adjustment, when it comes, arrives as a sudden write-down rather than a gradual price decline. A 9fin lender captured the mood: "Private credit exposure to software makes you question the risks, what your blindspots are, and perhaps even want to re-underwrite some of the most exposed names."
Second, and more concerning: payment-in-kind structures. Many SaaS companies in private credit portfolios use PIK loans, arrangements where borrowers can defer paying interest in cash, instead adding the unpaid interest to the loan balance. PIK structures are designed to give fast-growing companies time to build revenue before servicing debt. They work when growth materializes. When growth stalls or reverses, deferred interest compounds into a larger and larger liability that the borrower's cash flows can no longer support. The weakness is invisible until it isn't. In late 2025 and early 2026, BDC redemption requests exceeded $7 billion, with Blue Owl's technology-focused fund experiencing the largest outflows, a signal that investors closest to the data are acting before the quarterly marks catch up.
Now follow the chain one final link.
BDCs are not institutional-only vehicles. They are increasingly held by retail investors, in brokerage accounts and retirement portfolios. The U.S. recently gave regulatory approval for private credit managers to sell into the roughly $13 trillion defined contribution market. The democratization of private credit, which was celebrated as giving ordinary investors access to institutional-quality yields, also means that the software valuation repricing can propagate, through a chain of opaque linkages, from a compressed ARR multiple on a SaaS company to a decline in the net asset value of a retirement account held by someone who has never heard of ARR, has no idea what a BDC is, and has no way to evaluate whether the software loans in their fund are sound.
This is the cascade. Public equity repricing propagates to VC return models, which propagates to PE exit values, which propagates to private credit collateral quality, which propagates to BDC net asset values, which propagates to retail investor accounts. At each link, the adjustment is slower and less transparent than at the link before. And at the end of the chain are the people with the least information and the fewest tools to protect themselves.
The Structural Question There is a version of this story that is cyclical: multiples compressed, the market overreacted, and in 12-18 months the strong companies will recover and the weak ones will be acquired. Bank of America has argued the selloff is overdone. Some investing professionals view the current prices as an opportunity. This may prove correct for individual companies.
But the structural version of this story is different, and it is the one The Paranoidist is built to examine.
The structural question is not whether Salesforce's stock recovers. It is whether the foundational characteristics that made software a preferred asset class, the characteristics that justified the ARR multiple framework, the venture model, the private credit thesis, and the entire financial architecture built on top of them, remain intact in a world where agentic AI can replace rather than augment enterprise software.
The per-seat pricing model, which was the economic engine of SaaS, requires that companies need a growing number of human users for each software product. Agentic AI breaks this assumption directly: if one AI agent can do the work of 50 junior employees, the number of seats, and therefore the revenue, contracts. The pricing model transition to usage-based or outcome-based models may preserve revenue for some companies, but it fundamentally changes the nature of that revenue. It becomes less predictable, less contractually guaranteed, more variable, and therefore less valuable per dollar when the market applies a multiple to it.
If that transition is structural rather than cyclical, then the financial architecture built for the era of seat-based SaaS needs to be rebuilt for whatever comes next. And the rebuilding will be painful for everyone who holds assets valued under the old architecture and repriced under the new one.
The historical parallel is instructive. In the early 2010s, the transition from on-premise software (sold as perpetual licenses) to cloud-based SaaS (sold as subscriptions) destroyed the valuation framework for an entire generation of software companies. Companies that adapted thrived. Companies that clung to the old model, many of them large and profitable, watched their valuations collapse. The financial ecosystem adjusted, but the adjustment took years and produced significant losses for investors who were slow to recognize that the change was structural, not cyclical.
We may be at the beginning of an equivalent transition: from seat-based SaaS to AI-native, outcome-based software economics. The companies that adapt will command new premiums. Those that don't will join the list of businesses that were profitable right up until their business model became obsolete.
The Pricing Failure Connect this to what The Paranoidist is built to identify: risk that isn't priced.
No enterprise risk model currently includes "ARR durability under AI disruption" as a stress-tested variable. No private credit underwriting framework systematically evaluates whether a borrower's per-seat revenue model is vulnerable to agentic AI replacement. No BDC disclosure regime requires transparency about the AI disruption exposure of underlying software loans. No VC fund model stress-tests returns against a permanent 40-60% compression in software exit multiples. And no retirement fund prospectus warns investors that their yield is derived from loans to companies whose business model may be structurally threatened by the same AI technology generating the headlines they read every morning.
The equity markets, to their credit, are repricing in real time. Public SaaS stocks have been marked down 20-30%. The pain is visible and immediate. But equity markets are the most transparent link in the chain. At every preceding link (VC marks, PE valuations, private credit loan books, BDC NAVs), the adjustment is slower, less transparent, and further from the information needed to evaluate it.
UBS's Mish identified the core asymmetry: "Equity markets may be pricing in sector divergence better than credit markets right now." This is the definition of a lag that produces surprises. The public market has repriced. The private market has not. The question is not whether the private market adjustment happens, but when, and how disorderly the adjustment is when it arrives.
This is the same structural pattern that preceded the 2008 financial crisis. Equity markets began repricing mortgage risk in 2007. The credit markets, where the real leverage and the real exposure lived, didn't fully adjust until the system broke in September 2008. The lag between when the equity market prices the risk and when the credit market recognizes it is the window in which systemic damage accumulates.
What to Do About It The Paranoidist is about productive paranoia, not paralysis. Here's what I'd actually do if I were sitting in your seat.
If you're a board director: Ask your CFO and treasurer three questions at your next meeting. First: "What percentage of our enterprise value is derived from per-seat subscription revenue, and what is our plan if seat counts decline 20-30% over the next three years?" If the answer is that seat compression won't affect your company, ask for the analysis that supports that conclusion, because every SaaS company's management team believed the same thing six months ago. Second: "Are we carrying debt that was underwritten against ARR growth assumptions that no longer hold?" If your company was acquired or recapitalized during 2021-2022 at peak valuations with private credit financing, the gap between the debt structure and current market reality may be material. Third: "What does our exit or liquidity timeline look like under compressed multiples?" If the board's strategic plan assumes an exit at 8-10x ARR and the market is now pricing comparable companies at 4-5x, the plan needs revision, not next quarter but now.
If you're a CEO or founder: Run two scenarios honestly. Scenario one: your pricing model transitions successfully to usage-based or outcome-based economics, and your revenue stabilizes at a lower but sustainable level with a defensible moat. Scenario two: the transition stalls, seat counts decline faster than new pricing models can compensate, and you're caught between a shrinking revenue base and a cost structure built for growth. The difference between companies that survive structural transitions and those that don't is almost always the speed of honest assessment. The companies that waited to see if the cloud transition was "real" were the ones that didn't make it. If you have venture or PE investors expecting exits at pre-SaaSpocalypse multiples, have the conversation now about what realistic exit economics look like. Surprises compound. Early honesty preserves options.
If you're a CRO or risk leader: Three immediate actions. First, add "software valuation repricing" to your risk register as a monitored scenario, not a one-time event. Track the SaaS Capital Index, BDC software exposure data, and private credit default indicators the way you track credit spreads and VIX. The repricing is ongoing, and the private market lag means the second wave of adjustment hasn't hit yet. Second, if your organization holds BDC investments, private credit funds, or any vehicle with material software lending exposure, demand transparency on the AI disruption profile of the underlying portfolio. The Barclays data (20% BDC exposure to software) and UBS analysis (25-35% of private credit exposed to AI disruption) are sector-level estimates. You need position-level data. Third, stress-test your own revenue model. If you operate a SaaS business or depend on SaaS vendors, model the scenario where your vendors' financial stability deteriorates because their ARR is compressing. Vendor concentration risk takes on a new dimension when the vendor's business model is under structural pressure.
If you're an investor: Recognize that the opacity gradient in this cascade is itself a risk. Public equities have repriced and will continue to do so. VC portfolio marks are stale. PE exit assumptions are optimistic. Private credit loan books are quarterly-marked by the managers who originated the loans. BDC NAVs may not reflect current collateral quality. At each step down the chain, you have less information and more exposure to a delayed adjustment. If you hold assets at any point in this chain, particularly in private credit or BDCs with significant software exposure, the question is not "will this affect me?" but "when will the mark catch up to reality, and do I want to be holding the position when it does?" The prudent action is not panic selling; it is demanding transparency and stress-testing assumptions that were made in a different market environment.
If you're a citizen and a thinker: Understand that the SaaSpocalypse is not a story about technology stocks. It is a story about what happens when the financial system builds an elaborate architecture on an assumption, recurring software revenue is permanent and predictable, and then the assumption changes. The architecture doesn't adjust gracefully. It adjusts through a cascade of delayed recognitions, each one slower and less transparent than the last, until the adjustment reaches the people furthest from the information: retail investors, pension holders, and anyone whose retirement account holds products built on software lending. The same democratization of finance that gave ordinary investors access to institutional yields also gave them exposure to institutional risks, without the institutional tools to evaluate those risks. If you hold a BDC, a private credit fund, or a "yield-plus" product in your retirement account, ask your advisor what percentage of the underlying assets are software loans, and whether those loans were underwritten before or after the market began pricing AI disruption.
The Paranoidist's Assessment Probability that the SaaSpocalypse represents a structural repricing rather than a cyclical correction: Moderate to high. The cyclical case (oversold, will recover) is plausible for individual companies. The structural case (the foundational characteristics that justified SaaS premium valuations are permanently altered by agentic AI) is supported by the pricing model transition, the seat compression already visible in earnings, and the historical pattern of prior technology transitions. Both dynamics are likely operating simultaneously, which means the market will bifurcate: AI-adaptive companies recover, legacy models do not, and the average obscures both outcomes.
Probability that the private market adjustment (VC marks, PE valuations, private credit loan books) lags the public market repricing by 12-18 months: High. Private markets are structurally slower to adjust. Quarterly marks, manager discretion on valuations, and the absence of daily price discovery all contribute to delayed recognition. This is not a prediction of a crash. It is a prediction that the current gap between public and private market pricing of software risk will close, and the closing will be uncomfortable for holders of private market assets.
Probability that private credit defaults in software-exposed portfolios increase materially in 2026: Moderate to high. UBS's base case of a 2-percentage-point increase to approximately 6% is conservative but credible. The aggressive disruption scenario of 13% is a tail risk worth monitoring. The leading indicators (Apollo's exposure reduction, BDC redemption waves, rising bond dealer compensation for BDC debt) all point in the same direction.
Probability that current VC fund models adequately account for compressed software exit multiples: Near zero. The venture model was built for a world of 8-15x ARR exits. If exits settle at 4-6x for the median SaaS company, the return math for an entire generation of fund vintages does not work. This will become visible in LP reporting over the next two to four years.
Probability that retail investors in BDCs and private credit funds understand their exposure to software valuation risk: Near zero.
What I'm watching: BDC earnings season, which is starting now, will provide the first hard data on how fund managers are marking their software loan books. If marks remain stable despite the public market repricing, the gap is widening and the eventual adjustment will be larger. I'm also watching private credit default data, particularly in software and business services, and whether the PIK structures that deferred cash interest payments begin converting to realized losses. Finally, I'm watching whether any regulator, rating agency, or industry body begins requiring AI disruption stress testing for private credit portfolios. If that happens, it forces the transparency that is currently absent. If it doesn't happen, the opacity persists until a crisis forces the reckoning.
Where I might be wrong: It's possible that the SaaSpocalypse is, in fact, a classic overreaction, and that the strong SaaS companies will demonstrate resilient ARR in upcoming earnings, stabilizing multiples and reversing the cascade before it reaches the private market in force. Bank of America and several other analysts have made this argument. If the next two quarters of software earnings show that seat compression is limited to a narrow category of commodity tools rather than a broad disruption of enterprise software, the structural thesis weakens. It's also possible that the transition to usage-based pricing, rather than compressing valuations, creates a new premium for companies that successfully capture a larger share of enterprise workflow value, even at lower per-seat prices. Early signs of this would be companies reporting higher total contract values despite lower seat counts. I don't see this yet in the data, but it is the bull case and it deserves monitoring. Finally, it's possible that the private credit exposure is more manageable than the headline numbers suggest: Ares' CEO noted that software represents only 6% of total assets and less than 9% of private credit AUM. If the large, sophisticated managers have already de-risked, the losses may concentrate in smaller, less diversified funds rather than producing systemic contagion. This would limit the damage but not eliminate it, and the retail investors in those smaller funds would still bear the brunt.
The Paranoidist publishes weekly. If this changed how you think about one thing, consider subscribing. If it didn't, tell me what I'm missing. The whole point of productive paranoia is that I might be wrong, and I'd rather know now.
Paul Morin is the founder of DeepStrategy.ai and publisher of The Paranoidist, BoardroomRadar and ScenarioWatch. He has spent more than three decades in entrepreneurship, finance, risk management, and insurance, which is why he worries about the things that keep other people awake at night.
Researched, written, and edited in collaboration with Claude by Anthropic.