Kroll 报告：40 法则失效，增长决定估值，类别比规模更值钱

Kroll published its Summer 2026 Global Software Sector Update, covering software M&A and public comps through June 30, 2026. It’s 20 pages of deal data with a lot of detail on what the market currently pays for B2B software.

The headline numbers read like a big M&A recovery. Deal volume is on track for the second-highest year on record. Private M&A EBITDA multiples jumped 25%. Public multiples ticked up for the first time in a while. The detail underneath is less uniform. Valuations now vary enormously between categories that look nearly identical on growth and margin, which means buyers are pricing something the standard framework doesn’t capture.

But so much of the recovery is just Cursor and a few other AI deals.

Five things worth pulling out.

#1. Record Deal Count, Near-Record-Low Deal Dollars (At Least for 1H of Year)

Annualized 2026 software M&A volume is roughly 2,672 transactions, the second-highest count ever behind 2025’s 2,939. Announced deal value annualizes to about $240 billion.

One transaction accounts for half of that. SpaceX bought Cursor for $60 billion in the first half, and because Kroll is annualizing H1 figures, that single deal contributes roughly $120 billion of the $240 billion total. Excluding it, annualized 2026 deal value is closer to $120 billion, which would be one of the lowest totals on record. 2021 did $429 billion.

In Q2 alone, Cursor accounted for about 64% of all software deal value. Kroll flags that as the most concentrated single-deal impact in the ten years of data they track. The prior record was Q2 2022, when Broadcom/VMware and ICE/Black Knight together made up 50% of the quarter.

More companies are getting acquired than in almost any year on record, and the aggregate price paid for all of them is near a decade low. If you’re running a B2B company at $10M to $50M ARR and wondering why your banker has plenty of meetings and no term sheets, this is the chart that explains it.

#2. The Multiple Gap Between Subsectors No Longer Tracks Rule of 40

The subsector chart is the most useful page in the report. It ranks median EV/CY26 revenue multiples against the CY26 revenue growth and EBITDA margin behind each category.

Engineering and HCM have an identical Rule of 40 at 46%. Engineering trades at 5.2x and HCM at 3.0x, a 73% premium for the same combination of growth and profitability.

Collaboration & Productivity runs a 42% Rule of 40 and gets 3.1x. Vertical Software runs the same 42% and gets 3.7x. Marketing runs 38%, identical to Cyber Security, and gets 2.1x against Cyber’s 5.2x.

Growth plus margin explains almost none of the spread between these categories. Kroll doesn’t say what does.

Two explanations fit the data, and it matters which one you believe.

The boring one is revenue durability. Kroll’s Engineering bucket is Autodesk, Cadence, Synopsys, Dassault, PTC and Trimble. EDA and CAD, with twenty-year switching costs and almost no churn. The HCM bucket is ADP, Paychex, Paycom and Workday. Payroll is sticky too, but it’s more competitive and steadily commoditizing. That difference alone can explain a 73% multiple gap without anyone mentioning AI.

The more interesting one is that the market is pricing which categories agents make more valuable and which they make less necessary. Customer experience at 1.8x and marketing at 2.1x are the two categories where the underlying work is most directly automatable, and where a buyer can most easily picture replacing seat count with agents. ERP and supply chain at 6.5x is the system of record the agents have to read from. The Kroll data doesn’t prove any of that. It’s consistent with it, which is a weaker claim.

The founder takeaway holds under either explanation, and it has nothing to do with picking a better category.

Which comp set you get assigned to is worth more than several points of growth or margin. Two companies with the same revenue and the same margin can trade 2x apart on framing alone:

• Salesforce paid 9.5x for Fin inside a category whose public comps trade at 1.8x, because Fin got comped as an AI agent company rather than as customer-service software.
• Autodesk paid 26.7x ARR for MaintainX. Schneider Electric paid 18.2x for Cognite. Publicis paid 2.7x for LiveRamp.

Most planning cycles go into the growth and margin numbers in that chart. Almost none go into the positioning that determines which multiple gets applied to them.

Three $3B B2B Acquisitions in 30 Days: Intercom/Fin, Cognite, and MaintainX. They All Bought the Same Thing: Data for AI

#3. The Growth Cliff Got Steeper, and Margin Stopped Mattering

Public EV/CY26 revenue multiples by growth bracket:

• Below 0% growth: 2.4x median
• 0% to 10%: 3.1x
• 10% to 20%: 4.1x
• Above 20%: 7.2x

The third-quartile spread is wider still: 3.4x for shrinking companies, 11.6x for companies growing above 20%.

Median EBITDA margins for those same four buckets: 28.9%, 31.6%, 26.0%, 25.1%. They’re flat, and slightly lower at the top. The companies earning 7.2x are not more profitable than the companies earning 2.4x. The multiple difference comes entirely from growth.

Two things follow. Profitability has become table stakes rather than a differentiator, and roughly 25% to 30% EBITDA margin is now just what a public B2B company looks like. Cutting deeper buys nothing. And the step-up above 20% growth is severe enough that the difference between 19% and 21% is worth more to your valuation than anything else you’ll do this year.

The EBITDA multiples say the same thing: 7.0x for shrinking companies, 9.9x for 0 to 10%, 14.1x for 10 to 20%, 18.2x for 20%+.

#4. Getting Acquired Currently Pays Better Than Being Public

Median EV/LTM revenue for strategic acquisitions in H1 2026 was 6.0x, near the 2021 peak of 6.2x and well above the ten-year average of 4.7x.

Median EV/NTM revenue for public B2B companies is 3.3x. That’s up 6% in Q2 and still far below the 15-year median of 5.4x.

Private EBITDA multiples rose 25% to 20.2x, approaching the 2021 record of 21.4x. The public markets spent the last twelve months punishing nearly every software category. Median stock performance from June 2025 to June 2026: Financial & Accounting down 50.4%, HCM down 42.3%, Marketing down 40.8%, Vertical Software down 38.4%, Engineering down 35.6%. Infrastructure was the only category with a positive median, up 26.8%.

• Strategic buyers now account for 74% of all software transactions, up from a 69% average across 2024 and 2025.
• Private equity volume remains well below its 2021 peak, and PE multiples have been flat at roughly 4.4x since 2024.

The bid is coming almost entirely from corporate acquirers buying AI capability, proprietary data and workflow position. They’re paying close to double what the public market pays and 1.6 turns more than financial buyers. That’s strategics concluding they can’t build AI capability fast enough internally, and buying it at prices no public investor and no LBO model will support.

#5. Who You Are Matters More Than How Big You Are

Kroll splits precedent EBITDA multiples by revenue size. Companies under $100M in revenue: first quartile 11.5x, median 16.1x, third quartile 22.9x. Companies above $100M: 12.4x, 17.3x, 23.9x.

The scale premium is real but small. Crossing $100M in revenue moves the median from 16.1x to 17.3x, about 7%.

The dispersion inside each size class is much larger. Under $100M, moving from the first quartile to the third is a 100% uplift, 11.5x to 22.9x. Above $100M it’s a 93% uplift.

Being twice as big is worth 7%. Being in the top quartile of your size class is worth 100%.