AI 重建只完成 40%：增长停滞在 10%-30% 的 B2B 现实与出路

The CEOs I have the most respect for today? The ones truly rebuilding their companies for the Age of AI. Not just sort of, kind of, doing so.

Not the ones who shipped AI features and put them on the pricing page. I mean the ones tearing up how the product works, how it’s priced, how it’s sold, and what the org looks like on the other side.

Most of them are, at best, 40% of the way there.

For every rebuilding story like Fin (formerly Intercom) and its $3.6B exit to Salesforce, there are 100, probably 1,000 more B2B companies still fighting just to get there. They launched plenty of AI features. Some have an agent or two in production now. It’s helping. It hasn’t turned the ship.

Growth is still hard, and still stuck somewhere in the 10% to 30% range.

It’s friggin’ hard, man. I have some of the answers. I try to put them here every week. I don’t have all of them. And the truth is, and everyone investor sees this, there’s a lot of quitting now in the Age of AI. I’m going to withhold comment on it, but what I will say is, man, super respect to those fighting harder than ever.

10% to 30% Is Where Almost Everyone Actually Is At Scale (or Even Below)

Look at the public comps, since public companies are the ones that have to report.

Join the SaaS Capital Index per-company multiple and growth sheets as of June 30, 2026 and you get 58 constituents with both fields. The distribution:

• 18 companies growing under 10%
• 23 companies growing 10% to 20%
• 11 companies growing 20% to 30%
• Only 6 of 58 growing faster than 30%

The 60%+ public cohort has effectively disappeared.

PitchBook’s Q2 2026 comp sheet puts estimated median 2026 revenue growth at 13.2%, up from 12.2% in the Q1 report, led by DevOps, ITOps and developer/automation platforms at 21.9%, with CRM, sales, marketing and CX, plus collaboration and productivity, at the bottom.

Private isn’t materially different. SaaS Capital’s 2026 survey of private B2B companies has bootstrapped companies at a 20% median growth rate and equity-backed at 25%.

So when a CEO tells me they’re stuck at 18% and can’t find the second gear, they’re the median. That’s the part nobody says out loud at the board meeting.

The multiples tell you what the market thinks of that range. On the same June 30 data: 5.5x for the 20% to 30% growers, 3.1x for the 10% to 20% band, 1.9x for anyone under 10%. Sliding from 22% growth to 18% costs you roughly half your enterprise value.

That’s the real pressure. The question isn’t whether you can ship AI. It’s whether you can ship it fast enough to climb out of a band that’s already priced at half the company.

What a Real Rebuild Looked Like

Fin is the story everyone points to, and the shorthand version makes the whole thing sound easier than it was.

Intercom was a SaaS-era darling. It invented a lot of the patterns the rest of us copied. Around 2023 it bet the company on weeks-old LLMs and built an AI support agent. Then it spent nearly four years building around that agent. In May 2026 it renamed the entire 15-year-old company after the agent. Four weeks later, Salesforce signed a definitive agreement to acquire Fin for approximately $3.6 billion. It’s signed, not closed. Salesforce expects it to close in its fiscal Q4 2027.

The results underneath it: $400M+ ARR total, with the Fin agent itself approaching $100M ARR, 30,000+ customers, and AI agents resolving an average of 76% of support volume end to end.

Two things about that deal matter if you’re a CEO in the middle of your own rebuild.

• One: the rebuild took four years and ended with a new name on the building. A new model, a new product, a new pricing motion, and a 15-year-old brand retired to make room for the agent.
• Two: the market still paid a classic B2B multiple for it. $3.6B on $400M+ of run rate is under 9x. In the same window, Sierra raised at $15B and Decagon at $4.5B on far less revenue. The company that did the hard rebuild, with real customers and real resolution rates, exited at a fraction of the multiple assigned to companies that started clean.
• Three: it was forced on them. For all the great PR Eoghan McCabe the CEO got on this, the reality is CX and support moved to agentic models far more quickly than almost any category. So they were forced to match new and emerging competitors.

Rebuilding doesn’t get rewarded the way starting over does. You do it anyway, because the alternative is the 1.9x band.

Why “We Shipped AI Features” Doesn’t Move the Number

Almost every B2B company over $20M ARR has AI in the product now. Growth didn’t respond. The reasons are structural, and there are four of them.

The pricing unit didn’t change. If AI makes each seat more productive and you charge per seat, you built a product that shrinks your own billing base. The slowest-growing categories in the PitchBook data are the most seat-dependent ones. Every company that has moved meaningfully has changed what it charges for, not just what it ships.

The agent got bolted onto the old workflow. Adding an assistant to a screen a human still has to open is a feature. Removing the screen is a rebuild. Most companies are still on the first one because the second one breaks the org chart.

The data layer wasn’t ready. Every AI rollout I’ve watched up close, including our own at SaaStr, hits the same wall. The model is fine. The context is a mess. Stale records, duplicate entities, fields nobody maintained because no human ever read them. You don’t find any of it until an agent acts on it in production.

The GTM motion is still built for the old buyer. Selling outcomes to a budget holder who used to buy seats is a different sale, a different champion, and usually a different pricing conversation.

Any one of those is a year of work. Most CEOs are at 40% because they’ve done one or two, not all four.

The Part Where It Gets Tempting to Quit

It’s so easy to just quit now. Or to leave the keys on the table for the next guy. Take the PE offer at 3x. Let the incoming operator run the playbook.

The problem is that the next guy almost always has fewer ideas than you do. They didn’t found it. They didn’t sit with the first ten customers. They will never care as much, and caring is most of what a rebuild runs on in year two when nothing has worked yet.

I lived a small version of this. Adobe killed EchoSign’s free edition within weeks of my leaving. Not out of malice. It just wasn’t theirs. Nobody in the building had spent four years learning why that free tier existed, so it read as a line item instead of the distribution engine it actually was.

That’s what happens to a half-finished rebuild when you hand it off. The new team sees costs where you saw the mechanism. They cut the AI budget because it’s compressing gross margin, which it is, and they’re not wrong on the spreadsheet. They’re just never going to finish what you started.

Airtable Just Sold for $2.25B at 2.7x ARR. It Raised $1.4B and Was Once Worth $11.7B. This Was … Market

It’s Already Been 2+ Long Years of the Hardest Time For Many in B2B

To the CEOs still fighting through it: a quiet, silent salute.

This is likely the hardest stretch of your career as a founder, and it runs two-plus years, in public, with a board asking why growth is 17%, while the press writes about companies that got to $100M with 45 people and never had to carry any of your history.

You’re at 40% because 40% is roughly where a real rebuild sits after two years. The ones who got to 100% mostly had to rename the company to do it.

Keep going. The people who finish this will own their categories for the next decade, and most of them won’t be the ones who started fresh. They’ll be the ones who were at 40% in 2026 and didn’t hand over the keys.

I’m Still Tired. And If You Are Too, I Get It.