Apollo's "All the Marks Are Wrong" Warning: Private Equity Software Valuations Face Double Reset Catalyst

The market is delivering a stark reality check. Over the past few months, the public software sector has undergone a brutal repricing. As of February 20, the North American Tech Software Index had declined about 30% from its peak in mid-September. This isn't a minor correction; it's a sector-wide reset triggered by a single, powerful fear: that artificial intelligence could disrupt the very business model that powered two decades of growth.

The sell-off was swift and severe. When agentic AI tools capable of automating legal, financial, and data tasks launched in early February, relevant tech stocks sold off within hours. The anxiety is clear: if AI agents can perform the same functions as enterprise software for a fraction of the cost-or eliminate the need for certain categories entirely-what value remains in the traditional SaaS subscription model? This has forced a sector-wide "referendum" on durability, as one private equity executive put it, where the onus is now on firms to justify their valuations.

Against this backdrop, the disconnect between public and private markets has become glaring. While public stocks are repricing in real time, private valuations are often stale. Apollo Global Management's John Zito delivered a blunt assessment last month, telling UBS clients that "all the marks are wrong" regarding private equity's software holdings. His warning is a direct address to this expectation gap. He argued that lenders to smaller software companies could recover as little as 20 to 40 cents on the dollar, implying deep losses that private marks are failing to capture.

The driver is the same for both: fears of AI disruption. The sell-off in public software is the market's immediate reaction, a "sell the news" dynamic where the worst-case scenario is now priced in. The private market's lagging valuations, however, represent a dangerous expectation gap. They are built on a past where software was a durable, high-margin asset class. The current sell-off is a referendum on that assumption, and the private sector's outdated marks may be the next to fall.

The Mechanics of the Mispricing: Lenders, Equity, and the AI Disruption Risk

The lag in private valuations isn't just a timing issue; it's a structural flaw in how risk is layered and priced. The critical metric is the potential recovery for lenders. Apollo's John Zito warned that lenders to smaller software companies could recover as little as 20 to 40 cents on the dollar if those companies default. This stark projection reveals the core of the mispricing: private credit marks are often based on historical performance and collateral values, not on the new, existential threat posed by AI.

This creates a dangerous dual exposure. Many private equity-acquired software firms also carry significant private credit debt. As Zito noted, if the loans face difficulties, the equity positions are also affected. A markdown in a loan's value directly pressures the equity stake backing it, dragging down the overall portfolio valuation. The expectation gap here is that private equity firms may be marking their equity holdings at levels that assume a stable debt cushion, while the underlying loan portfolio is being revalued downward due to the AI disruption risk.

The AI narrative makes justifying valuations and achieving exits harder in two key ways. First, it directly challenges the durability of the business model. As one PE executive put it, "We think now there is a real referendum by limited partners on valuations of software and technology investments, and the onus is on general partners to justify them." The market's 30% sell-off in public software has set a new, lower benchmark for what's considered defensible. Second, it compresses exit windows. If a portfolio company's value is now in question, finding a buyer willing to pay a premium becomes more difficult. The once-easy path of selling a software firm at a high multiple to another buyer or a public market listing is now fraught with uncertainty.

The most exposed assets are a specific cohort. Zito identified software companies taken private between 2018 and 2022 as particularly vulnerable. These firms were often acquired during a period of high valuations and low interest rates, making them inherently less resilient. The AI disruption risk now hits them at a time when their business models may be less defensible than those of larger, public competitors. The private market's lagging marks fail to capture this confluence of timing and technology risk, leaving investors exposed to a deeper reset than the current public sell-off suggests.

The Market's Priced-In Reality: What's Left to Reset?

The market's sell-off has been a brutal reality check, but it's pricing in a specific fear: existential disruption from AI. The evidence is clear. Major software stocks like Palantir, Adobe, Salesforce, and ServiceNow are down about 22% to 30% so far this year, even as they report strong results. This disconnect is the core of the expectation gap. The market isn't reacting to current fundamentals; it's reacting to a future where AI could render entire business models obsolete.

This creates a powerful counter-narrative that the market is ignoring. The deep domain knowledge and complex integrations that companies like Palantir and ServiceNow have built over years are not easily replicable by a new AI tool. As one analysis notes, "the deep domain knowledge and complex integrations with mission-critical enterprise elements that companies like Palantir or ServiceNow provide make it nearly impossible for someone to throw together copycat software." In other words, the market is pricing in a broken logic-a fear that AI will disrupt all software equally-while ignoring the defensibility of durable platforms.

The most acute expectation gap, however, lies in the private market. The public sell-off has reset valuations for many firms, but it has not yet fully captured the risk for a specific cohort: the 2021-2022 private equity vintages. These firms deployed capital near peak valuations, often in less defensible software categories. As the evidence shows, those companies are now seeing markdowns, tighter exit windows, and growing LP scrutiny. Their private valuations, which may still reflect the frothy pre-AI era, are likely to face a more severe reset than the public market's current repricing.

The bottom line is a divergence in timing and focus. The public market is pricing in a broad, existential risk, creating a "sell the news" dynamic for stocks that are fundamentally strong. The private market, with its lagging marks, is still pricing in a past where software was a durable, high-margin asset class. The 2021-2022 PE vintages are the bridge between these two worlds, facing a double reset: their outdated valuations must now align with a market that has already repriced the sector's risk.

Catalysts and Risks: The Path to a New Equilibrium

The expectation gap between public and private software markets will narrow only when specific catalysts force a reckoning. The critical test is not a single event, but a sequence of market-driven validations that will determine whether the current repricing is a temporary scare or the start of a new, lower equilibrium.

The first and most immediate catalyst is a wave of private equity fund redemptions and mark-to-market adjustments. As limited partners demand justification for software valuations, the pressure on general partners will intensify. The evidence shows this is already happening. As one PE executive noted, "We think now there is a real referendum by limited partners on valuations of software and technology investments, and the onus is on general partners to justify them." This creates a powerful feedback loop. If Apollo's warning that lenders could face substantial losses triggers redemptions from private credit vehicles, it will force a cascade of downward valuations across the entire portfolio, from debt to equity. The pace and scale of these redemptions will be the first real-time signal of how deep the private market's mispricing runs.

The second, and more fundamental, catalyst is the pace of software M&A activity. This market is the ultimate test of durability. After growing 28% year-over-year in 2025, the market must now prove it can sustain that momentum. A slowdown or collapse in M&A would confirm the worst fears of AI disruption, making exits harder and valuations more fragile. Conversely, continued robust deal flow, especially at reasonable multiples, would signal that durable platforms still have a path to liquidity. The market will be watching for evidence that buyers still see value in complex, integrated software, not just in new AI-native entrants.

The ultimate catalyst, however, is a clear demonstration from portfolio companies themselves. The market's current sell-off is a "sell the news" dynamic, pricing in a broken logic that AI will disrupt all software equally. The path to a new equilibrium requires portfolio companies to show that AI integration is a catalyst, not a headwind, to their core business model. This means proving that AI tools are being used to enhance their existing competitive advantages-deep domain knowledge, complex integrations, and mission-critical enterprise relationships-rather than replacing them. When companies can articulate and deliver this narrative, they can begin to close the expectation gap, proving that their valuations are not just defensible, but potentially re-rated upward.