When David Bosa, partner at Iconiq Growth, and Eric Wu, CEO of Ribbon Communications, say that private equity is about to "consume" its own software portfolio, they are not describing a trend—they're describing an arithmetic reality. According to data cited by CNBC, private equity firms control approximately $1.2 trillion in managed enterprise software assets. This figure is not just decorative; it signals a significant incentive. And by 2026, that incentive shifted dramatically.
The most visible symptom is the decline in valuations. Private equity-backed SaaS acquisitions plummeted from median multiples of 12.5x revenues in 2021 to 6.8x in Q1 2026, as reported by Bessemer Venture Partners. Such a drastic drop in multiples is not merely a market correction; it’s the market acknowledging that traditional software no longer captures the same scarcity.
The thesis presented during the "AI Eats SaaS" panel at the CNBC Disruptor 50 Summit is straightforward: AI is no longer automating isolated functions; it is automating entire layers of product and operational stacks. Wu puts it bluntly: AI is automating "the entire SaaS stack that private equity owns." Bosa translates this into power dynamics: private equity built the installed base and can dismantle it.
I view this through a specific lens: Zero Marginal Cost. Not as a tech slogan, but as an economic pressure. When the marginal cost of producing, maintaining, and customizing software decreases, value shifts. Capital that thrives on buying cash flows and expanding margins doesn’t just watch the decline; it redesigns the asset to turn the decline into an advantage.
From Multiple Era to Margin Regime
Private equity thrived during the SaaS cycle from 2016 to 2022 with a clear playbook: buy predictable growth, standardize sales, cut costs, and sell at a higher price in a market that valued ARR as though it were a perpetual bond. This script required two conditions: an abundance of cheap capital and a perception of scarcity in the "installed" software. The first condition broke with rising interest rates. The second is being undermined by AI.
The quantitative evidence of this fracture is already reflected in prices. If the median revenue multiple falls from 12.5x to 6.8x in five years, the market is rewriting the implicit contract between buyer and seller: future growth is no longer considered safe, and software ceases to look like an asset that accrues frictionless rents.
This is where the capital movement comes in. Goldman Sachs, in a note from March 12, 2026, noted $450 billion of dry powder in private equity for software, but with a more critical figure: 62% of deployments in H2 2025 focused on AI. When money shifts destinations, the power map changes. The consequence is inevitable: SaaS assets not re-platformed for AI transition from being "core" to being "restructuring material."
This is not a moral judgment about SaaS; it's an incentive reading. The private equity model seeks returns of 3-5x with operational intervention. If AI allows cost cutting, accelerates releases, reduces support, and consolidates teams, then EBITDA becomes a more achievable target through technological redesign, not just financial pruning. Bain & Co., cited in the briefing, highlights that SaaS gross margins are eroding from 70-80% down to 60% under AI competition. When gross margins compress, the only refuge is to reconstruct the cost chain.
AI as a Compression Engine for the SaaS Stack
SaaS began by selling interfaces, flows, and permissions. Its economic unit became powerful because it packaged complexity—integrations, security, analytics, training, consultancy, and support. AI is unbundling some of this. Not because it’s "smarter" in the abstract, but because it lowers the cost of producing and adapting software.
The briefing offers a striking contrast that clarifies the shock: the global SaaS market reached $247 billion in 2025 (Statista), while the AI software market reached $97 billion, projected to $450 billion by 2028 (IDC). This isn’t mere growth; it’s a reassignment of value. And the reassignment accelerates when buyers perceive that traditional software incorporates costs that AI can displace.
Wu's phrase is more than a metaphor: "Legacy SaaS is the new coal—AI is the electricity grid." Taken literally, it states that legacy SaaS becomes input: useful but replaceable, subject to a new value distribution system. If the "grid" is the AI infrastructure, then the product shifts from being a module to being the ability to perform tasks.
Commercial adoption data starts to reinforce this intuition. The briefing introduces the concept of "SaaS fatigue," with median churn rates of 15-20% (KeyBanc Q1 2026) compared to 95% renewals for AI offerings. While renewal numbers depend on category and contract, the differential sends a message: the promise of productivity anchored in AI retains spending where traditional SaaS faces budget scrutiny.
The mechanical effect for private equity is clear. If a firm controls multiple software modules performing repetitive tasks, its portfolio simultaneously becomes an asset and a target. "Consuming" the portfolio doesn’t mean destroying value for whim; it means absorbing it within an AI architecture that reduces costs and boosts pricing power. When Bosa says that private equity can "uproot" the installed base, he indicates that the asset owner has the capacity to enforce migrations, consolidations, and discontinuities without waiting for the market to do so organically.
The New Private Equity Playbook for Software
What lies ahead appears less like a buying cycle and more like a series of surgeries. Dealroom.co reports 127 exits or restructurings of private equity-backed SaaS in 2025, a 34% increase from 2024. Bloomberg, according to the briefing, noted on March 12, 2026, 15 private equity firms announcing carve-outs of AI from SaaS portfolios. It doesn’t take a hundred announcements to see the pattern: extracting "what works" in SaaS and reassembling it into AI products is becoming a method.
The ultimate incentive is multiple arbitrage. Bessemer, in the cited material, observes that ARR multiples in AI hover around 40x compared to 7x in SaaS. This gap cannot remain intact; capital moves to capture it. In operational terms, arbitrage is not solely captured by buying "pure" AI companies. It can also be captured by taking an existing SaaS asset and rebranding it as AI, provided that the re-platforming is substantive enough for customers and translates into retention, expansion, and margin.
In this sense, the phrase "AI Eats SaaS" is inaccurate if read as total replacement. The more plausible scenario is selective digestion: legacy modules become layers of data, integrations, compliance, and contracts; the interface and workflows are rewritten around agents, automation, and mass customization. Gartner, cited in the briefing, projects that by 2027 40% of private equity’s SaaS portfolios will be re-platformed for AI. If this percentage materializes, it will define an investment and restructuring cycle the size of an industry.
The risk lies on the human and contractual side of this transition. Re-platforming is not just about technology; it involves migration, pricing, support, reputational risk, and renegotiation with customers who purchased stability. As growth slows, the temptation to "force" the transition increases. And when transition is forced, the probability of churn, legal disputes, and brand erosion rises. Private equity knows how to execute cuts, but AI requires product precision.
The public market is already sensing the shift. The briefing mentions that the SaaS index (SaaS:INDX) fell 22% YTD 2026 (S&P). This price holds a message for boards and LPs: distributions may take time, and performance quality will depend on how much of the portfolio can transform without breaking contracts.
The Survival of Software Depends on Designing Abundance, Not Selling Scarcity
The pivot of private equity is not a betrayal of SaaS; it is the acknowledgment that software has entered a phase where scarcity is diminishing. If AI drives the marginal cost lower, what once justified high rents becomes debatable. In this context, the company that survives will not be the one yelling "AI" the loudest, but the one that turns AI into a lighter cost structure and a more measurable promise of results.
For software leaders, the operational mandate is stark: defending multiples no longer depends on telling a story; it depends on showing retention, expansion, and margin in an environment where prices are under pressure. For private equity leaders, the mandate is equally concrete: value no longer lies in accumulating products; it lies in reconfiguring assets around data, automation, and distribution.
The last decade rewarded software ownership as a scarce asset. The next decade will reward the ability to redesign it when it ceases to be so, and that redesign will define which funds, companies, and economies maintain their position in the new productivity regime.
