Apollo Global Management is treating AI disruption as a core software investment risk, and that matters well beyond one firm. For founders, the message is simple: revenue growth no longer answers the hardest question in a SaaS deal.
Apollo's warning on software has moved from market commentary into investment discipline. At the SuperReturn International conference in Berlin this week, deputy global head of private equity Antoine Munfakh said the firm is now screening software targets for the risk that AI could replace or weaken their products. Coming from a firm that reported $1.026 trillion in assets under management in the first quarter of 2026, that is not a small change in language. It is a signal about how one of private capital's most influential buyers is repricing a sector it once loved.
The timing is important. Software buyout activity has already slowed sharply as investors struggle to decide which companies still have durable value in an AI-driven market. According to a recent Financial Times report citing PitchBook data, software deals fell to $50 billion in the first five months of 2026, down from $88 billion in the same period last year. That is the weakest start to a year since 2020, when the pandemic froze large parts of the deal market.
For private equity, the concern is not that all software companies suddenly stop mattering. The concern is that too many were priced as if their customer relationships, workflows, and interfaces were permanent barriers to competition. AI agents have made that assumption much harder to defend. If a product mainly organizes repetitive work, fills forms, summarizes data, or moves information between systems, investors now have to ask whether a general-purpose model can do the same job more cheaply.
That changes the due diligence process. Customer concentration, net revenue retention, annual contract value growth, and gross margins still matter, but they no longer settle the argument. Apollo and other large buyers now have to look at whether a target owns proprietary data, whether customers depend on it as a system of record, and whether its AI roadmap improves the product in a way competitors cannot copy quickly. A thin AI feature set is not enough.
The most exposed companies are easy to identify. Generic workflow automation tools face pressure because their value often sits in repeatable tasks rather than hard-to-replace data. Mid-market CRM platforms without deep vertical specialization face a similar problem because large incumbents and AI-native entrants can attack from opposite ends of the market. Analytics dashboards that pull together data but do not own the underlying context also look less defensible when models can generate summaries, forecasts, and recommendations across multiple systems.
Apollo has tried to make clear that its own exposure is limited. In March, Apollo president Jim Zelter said the firm's overall software exposure was less than 2% of total assets under management, while chief executive Marc Rowan argued that concentration in one technology-exposed industry was a poor risk management posture. That is the larger point for the market. The firms that kept software as one sleeve of a diversified portfolio can be selective. The firms that made SaaS the center of the book have a harder problem.
What this means for founders still fundraising
Founders raising growth rounds in vulnerable software categories now face a different investor conversation. A strong revenue chart helps, but it does not answer who buys the company in five years and at what multiple. If the old exit plan assumed a private equity buyer paying a premium for recurring revenue, that plan now needs a more specific defense. Investors will want to know why the product survives when customers can buy AI labor directly or shift workflows into broader platforms.
This does not mean SaaS is finished. It means the bar has moved. Companies with proprietary datasets, regulated workflows, embedded compliance requirements, or deep system-of-record status have a stronger case. Salesforce, ServiceNow, and vertical software providers in healthcare, finance, and industrial markets may still have defensible positions because replacing them is not just a question of building a better interface. It is a question of trust, permissioning, migration risk, and operational history.
The weaker story is the one that says AI will simply be added later. That worked when AI was treated as a feature. It is less convincing now that buyers are asking whether AI changes the buyer's need for the software itself. In that environment, a founder has to show that AI expands the product's moat rather than compressing it. The difference will show up in valuation, deal speed, and the number of credible exit options.
Apollo's posture also explains why capital is moving toward areas where AI is a productivity tool rather than a direct substitute. Data centers, power infrastructure, industrial services, insurance, and other real-economy assets offer investors a different risk profile. They may still be affected by AI, but they are less likely to be replaced by a model update. That distinction is already reshaping where large pools of private capital are willing to lean in.
For founders, the practical takeaway is direct. The next fundraising story cannot treat AI displacement as a side question. It has to explain why the company owns something durable, why customers cannot easily switch to an AI-native alternative, and why a future buyer would still pay a meaningful multiple. The software market is not closed, but the old recurring revenue premium is no longer automatic.
Also read: SLB's new AI deal with PDVSA shows enterprise technology is reaching markets where traditional capital still fears to tread • Senator Elizabeth Warren wants Wall Street to reveal exactly how deep its AI debt exposure goes • The AI spending boom has become America's newest inflation engine
