When Capital Concentrates Into the Sector That’s Resetting

Concentration met reset in 2025 — and the operator question is whether the moat the platform was designed around is the moat that survives.

PitchBook’s Q1 2026 analyst note Private Equity’s Exposure to the Software Reckoning (Garrett Hinds, February 17, 2026) opens on two figures worth holding side by side. Software accounted for approximately 18% of US PE deal value in 2025 — the highest share on record, against a long-run average closer to 14%. At the same moment, public software multiples sat more than one standard deviation below their eight-year average, a level the asset class had not seen since the 2022 reset. Concentration met reset, and the second figure is what makes the first one consequential.

The macro pressure traced in Dry Powder and the Pressure to Move lands in the software reckoning the way it lands in most sector-specific stress: capital that needed to move went to the sector with the most familiar underwriting case, and the sector then re-priced. The Q1 print is not a verdict on whether AI displaces SaaS — that argument will run for another decade. The more immediate operator question is whether the moat the platform was actually designed around is the moat that survives the reset.

The concentration print and what it actually means

A 14% long-run average climbing to 18% in a single year is not, by itself, a crisis. Sector concentration shifts every cycle — capital follows perceived durability, and software has been the asset class’s preferred underwriting case for most of the last decade. What makes the 18% print structurally different is its simultaneity with the multiple reset. The concentration happened just as public software valuations broke down through their long-run support, and the deals struck inside that 18% were almost all underwritten on multiple assumptions the public market has now invalidated. The vintage 2024–25 software cohort is, on the whole, a cohort that paid for a future the market has stopped paying for.

This is not the situation that produced the 2022 reset, when compression hit a cohort underwritten in 2020–21 at peak vintage. The current reset is hitting deals struck on the assumption that the prior one had fully cleared. That assumption was wrong, and the cohort that absorbed it is the cohort whose operating model now has to make up the difference.

The exit market priced the reset first

PitchBook’s 2025 Annual US PE Middle Market Report records the corollary: middle-market IT exits collapsed roughly 38% year-over-year in 2025, even as overall MM exit activity recovered. The exit market did the pricing work the deal market had not yet done. Sponsors holding software platforms underwritten in 2023–25 found the bid disappearing while the deal market was still pricing new software entries at concentration-era levels.

The lag between the two is not unusual; deal markets typically follow exit markets by two to four quarters in any sector reset. What is unusual is the concentration multiplier. When 18% of US PE deal value sits in a sector whose exit market has just contracted by 38%, the sponsor population carrying that exposure is unusually large — and the operating decisions that now matter are the ones that determine whether each individual platform clears, regardless of whether the sector overall does.

Where the moat actually lives

PitchBook’s analyst note frames the durable software moat with notable precision: switching costs, compliance, retraining requirements, and incumbent integration. The list does not include features, product roadmap, or AI capability. This is closer to The Industrialist’s integration-capacity vocabulary than to most sector commentary, and it is closer to the actual mechanism by which software businesses survive valuation compressions than the features-and-roadmap framing the trade press tends to default to.

The frame from The First 24 Months Now Decide the Deal applies directly. The moat the platform was designed around is one of the operating choices that becomes visible early — typically by month 24. Platforms that built operational lock-in into customer workflows, that hardened compliance and integration into their architecture, that produced retraining costs the customer cannot ignore — those platforms have a moat that survives a multiple reset because the moat is not priced in the multiple. Platforms that built features and rode growth assumptions have a moat that exists only at the multiple the market was paying.

Two software businesses, two reckonings

The reset does not punish “software” in any uniform sense. It punishes thin software — applications whose lock-in is shallow, whose compliance load is light, whose customer can switch without operational consequence — and it largely spares software whose customers cannot actually leave. The platform-selection distinction in the buy-and-build literature applies here in unusually direct form. Buy-and-build software platforms with deep organizational integration look like one kind of asset under reset conditions; thin-application SaaS roll-ups underwritten primarily on ARR growth and multiple expansion look like another.

In the cases I’ve studied, the software platforms that came through prior valuation resets intact were the ones whose customers couldn’t actually leave — not because the alternatives were worse, but because the switching cost was real. The platforms that did not survive prior resets were the ones where the lock-in was narrative rather than operational. The reset itself does the sorting; the operating model determines which side of the sort the platform lands on.

What the operating model is actually building

Most software deals struck in the concentration period were underwritten on a growth-plus-multiple case the reset has now broken. The growth assumption is independently questionable in many cases — AI capability is reshaping what enterprise customers will pay for incremental functionality — but the multiple assumption is the part already invalidated. Sponsors who paid 2024-vintage prices for businesses underwritten on 2024-vintage multiples need their platforms to clear a new market reality, and the operating choices that determine whether they can are being made right now without being framed as software-reckoning responses. Leadership configuration gets framed as operating-model design. Customer workflow integration gets framed as product roadmap. Compliance hardening gets framed as risk management. The choices add up to either a moat the new market will pay for or one it won’t.

Most of the software platforms that survived prior compressions did so because the operating work that built the lock-in had been compounding quietly for eighteen months or more before the compression arrived. The survival was a cumulative result, not a strategic response — eighteen months of operating decisions that, in retrospect, had been building the moat the new market still valued.

What 2026 settles

The software reckoning is not going to be resolved by a single quarter’s print, and the AI-displacement debate will outlast every operator currently in the asset class. What 2026 settles is something narrower: which vintage 2024–25 software platforms had the operating moat the new market still pays for, and which had the narrative moat the old market accepted in its place. The sorting is already underway in the exit market; it will reach the deal market by year-end. 2026 is the year vintage 2024–25 software platforms get marked to a market that no longer believes in multiple expansion — and the platforms that come through it intact will be the ones whose moats were operating moats, not narrative ones.