For most of the past decade-plus, B2B software had a beautiful, almost boring stability to it.
You’d build a product. You’d get it to $1M ARR. Then $10M. Then $100M. And the product you were selling at $100M ARR looked … pretty much the same as what you were selling at $1M. Better, sure. More features, more integrations, more enterprise bells and whistles. But fundamentally the same core product, solving the same core problem, the same core way.
Salesforce in 2025 looked a lot like Salesforce in 2015. HubSpot in 2025 looked a lot like HubSpot in 2017. Zendesk, ServiceNow, Workday — same story. You could plan product roadmaps in 3-year increments and not be wildly wrong.
That era is over. So, so, so over. And it ended so quickly.
The Stock Market “Discovered” the Threat in January. The Threat Was Already There.
What seems almost comical is that the public markets decided in January 2026 that AI was suddenly a massive existential threat to existing B2B leaders. Stocks got hammered. Multiples compressed. Analysts started writing notes with titles like “Who Gets Disrupted?”
As if this was new information.
Harry Stebbings on the 20VC x SaaStr pod put it perfectly: “If the AI companies are having the most exciting week ever, and SaaS stocks are down 30-40% in five weeks, those two things are happening at the same time. That’s not a coincidence — it’s the same story from two different angles.”
The real inflection point wasn’t January 2026. It was June 2024 — when Claude 3.5 Sonnet shipped. That was the moment it became obvious to anyone actually building with these models that AI wasn’t just going to be a feature bolted onto existing software. It was going to be a replacement layer for huge categories of work that software previously just organized.
The product of the year was Claude 3.5 (and later 3.7). Without it, we have no vibe coding. Without it, we have no Lovable, no Replit, no Cursor that really works. Many of these products existed for years. They did not work until Claude.
The old model: Software helps humans do work more efficiently. The new model: AI does the work. Maybe software orchestrates it. Maybe it doesn’t even need to.
That’s not a feature upgrade. That’s a category redefinition.
The Incumbents See It — Even If They Won’t Say It
The leaders of the biggest B2B companies know what’s happening. You just have to read between the lines of their earnings calls.
Workday CEO Carl Eschenbach on their Q1 FY25 earnings call: “We are seeing customers committing to lower headcount levels on renewals compared to what we had expected. We expect these dynamics to persist in the near term.”
One quarter later, he was telling analysts that “this whole concern around AI disruption and the potential negative impact on seat-based models are completely overblown.” By Q3, the tune changed again: “Your customers are growing headcount slowly right now… We’re focused not on just seats, but actually revenue per seat.”
That’s three quarters of narrative evolution from “customers are renewing at lower headcount” to “seat concerns are overblown” to “we’re not even focused on seats anymore.” That’s not confidence. That’s a company figuring out its story in real-time.
Benioff has been even more direct about what’s happening inside Salesforce itself: “I’ve reduced it from 9,000 heads to about 5,000, because I need less heads.” And on engineering: “In engineering this year at Salesforce, we’re seriously debating — maybe we aren’t going to hire anyone this year.”
If the biggest B2B vendor on the planet is cutting headcount because of AI, what do you think their customers are doing?
Klarna CEO Sebastian Siemiatkowski went furthest: “Thanks to AI agents + AI engineers getting prolific, you can rebuild most enterprise SaaS functionality, host for super cheap, and get basically 90%+ functionality.” Klarna is still largely an outlier. But outliers have a way of becoming the norm faster than anyone expects.
What I’m Less and Less Sure About Every Week
Here’s what I’ll admit: I am less and less sure every single week what B2B software looks like in 2027 and beyond.
Not less optimistic. Less certain.
There’s a difference.
I’ve spent 13 years at SaaStr studying how B2B companies scale. I’ve made 100+ investments. I’ve written 4,600+ posts about the playbooks. And right now, I have genuine uncertainty about which of those playbooks still apply in 36 months.
Some things I used to take for granted that I’m no longer sure about:
- Will per-seat pricing survive? If an AI agent does the work of 5 SDRs, do you still charge per seat? The math breaks for the vendor. But outcome-based pricing is brutally hard to implement and even harder to make predictable for public market investors.
- Will “workflow software” still be a category? Half the B2B market is essentially workflow management — tools that route tasks between humans. What happens when there aren’t humans in the loop for 80% of those tasks?
- Will switching costs hold? The old moat in B2B was data gravity and workflow lock-in. But if an AI layer can sit on top of any system of record and abstract away the interface, does it matter what’s underneath? Rory O’Driscoll from Scale Venture Partners nailed the framework: “There’s a race between incumbents who have distribution needing to add product, and new guys who have product needing to add distribution. How long your existing distribution advantage lasts depends entirely on how sticky your product is — which is why accounting systems get decades and task management tools get months.”
- Will vertical SaaS stay defensible? Vertical SaaS was the best category in B2B for the past 5 years. Deep domain expertise, high NRR, sticky customers. But AI is compressing the time it takes to build domain-specific software from years to months. Maybe weeks.
I don’t have confident answers to any of these. And I’m skeptical of anyone who does.
Competitive Edges Are Now Measured in Months
Here’s what really stresses a lot of us out about the instability.
When I started in B2B, you’d build something really slick and then you’d have about 12-18 months until you were copied by a startup. Then you’d have like 5 years until a big company copied you.
When we launched Adobe Sign / EchoSign, DocuSign only worked in Windows and was only partially web-based. We had like 18 months before they decided to copy what we did. Then it took 5-6 years for Adobe to decide to copy it. Google launched a clone like last year — a decade later. That’s the way it used to happen.
Not long ago I invested in a startup I love. Incredibly powerful. Within two weeks they had four clones. Then I talked with a massive company that’s building a clone themselves that’ll be out this year.
Our competitive edges are going to be measured in months when they used to be measured in years. And that compounds. If you can’t run at that pace, you’re going to fall into terminal decay.
Look at PagerDuty. They just crossed $500M in ARR. They’re worth $1 billion. 2x revenue. Their customer count has been flat at 15,000 customers for four years. DataDog came out with a clone. Startups are better. They just didn’t move fast enough. And it’s only going to accelerate. DataDog took 15 years to decide to build their competing product. Next time it might take them 8 months. And 90 days the time after that.
Don’t underestimate the rate at which this stuff is getting better.
The Instability Is the Point
What’s changed isn’t that B2B is dying. Revenue is still flowing. Budgets are still there. Companies still need software — and increasingly, AI-powered software — to operate.
What’s changed is the stability.
For a decade, if you built a great B2B product and found product-market fit, you had a 7-10 year runway to compound growth on essentially the same core product. Your competitive moat deepened every year. Your sales playbook got more refined. Your CAC payback got more efficient.
Now? I’m not sure you have 3 years of stability. Maybe 2.
That’s terrifying if you’re a $50M ARR company that just raised at 15x and your board expects you to “stay the course” on your existing product for the next 3 years. The course may not exist in 3 years.
It’s also the biggest opportunity in a generation if you’re a founder building something new. Because every incumbent’s instability is your window.
I’ve lost faith in revenue durability. I see everything decaying that isn’t growing at abnormal rates. I smell it in leads, in close rates going down, in an inability to charge more when agent competitors charge 10x less. I smell decay at the board meeting. I smell it when the investor update comes 28 days after end of month.
There are really only two types of companies right now. Private companies growing at insane rates, or unfundable. Public companies accelerating, or worthless. I don’t even care whether you’re AI, B2B, or fintech — are you growing like a beast?
“We’re a System of Record” Is Not a Growth Strategy
Here’s what I’m hearing from a lot of B2B leaders right now: “We’re safe. We’re a system of record. Nobody’s ripping us out.”
And for now? That’s probably true. If you’re the HCM system running payroll, the ERP processing invoices, the CRM that holds every customer relationship — yes, AI isn’t going to cause massive churn. Rip-and-replace of core systems of record is still brutally hard. The data gravity is real. The compliance requirements are real. The integration complexity is real.
But here’s the problem: not churning is not the same as growing.
The growth in spend, in usage, in everything that matters right now is in AI agents. Or at least AI-adjacent spend. That’s where the new budgets are going. That’s where the excitement is. That’s where the expansion revenue lives.
If you’re a system of record and you don’t own the AI agents that make your platform thrive — the agents that actually do the work your software used to just organize — you’re going to find yourself in a very uncomfortable place. Low churn. Flat growth. Slowly declining relevance.
Your customers won’t leave. But they won’t spend more, either. And over time, the AI-native layer sitting on top of your system of record will capture more and more of the value. You’ll become the database underneath someone else’s product.
Folks might not churn. But you won’t grow. And in B2B, flat is the new down. That’s not enough.
What I’d Do Right Now
If I’m running an existing B2B company, I’m doing three things immediately:
First, I’m assuming my current product has a shorter shelf life than I think. Not because it’s bad — because the definition of “good enough” is changing faster than any product team can ship. Build the AI-native version now, not next year.
Someone asked me recently if B2B will be dead in 2-3 years because you can just talk to an AI and say “write code that does that.” Six months ago, I’d have said the odds are zero. That’s ChatGPT nonsense. It’s still impossible to reproduce the amount of workflows, the amount of corner cases, the amount of complexity that complex B2B software does. But you can chip away at it. And you can chip away faster than ever. I built 11 apps in 90 days that have been used 800,000+ times. I haven’t stolen revenue from anybody — but that’s a bit of an existential threat to somebody, isn’t it?
Second, I’m watching what my customers’ employees are actually doing every day — not what my software helps them do. Because AI is going to automate the employee’s job, not just improve the software they use to do it. If your product is a tool for a role that’s about to be 80% automated, you don’t have a product problem. You have a market problem.
Third, I’m being honest with my board and my team that the next 24 months require a level of strategic flexibility that’s uncomfortable for everyone. The 3-year plan is a best guess. Update it quarterly.
B2B Isn’t Dead. But “Stable B2B” Might Be.
The total spend on B2B software and AI is going to be larger in 2027 than it is today. Probably much larger. The AI transformation of business operations is a massive tailwind for the overall category.
But the distribution of that spend — who captures it, how it’s priced, what form it takes — is genuinely uncertain in a way it hasn’t been since the shift from on-prem to cloud.
And that shift took a decade to play out. This one might take 3 years.
As Rory O’Driscoll warned: “If the AI lift starts to recede, they’ll all come down because they’re way over their long-term valuation. Pennies in front of steam wheels.” And on the other side, the opportunity is equally massive. What agents can do — we just started. Outside of parts of coding, we just started. We haven’t missed the boat as investors. We haven’t missed the boat.
I used to be able to look at a B2B company’s metrics — ARR, NRR, CAC payback, magic number — and tell you with reasonable confidence what it would look like in 3-5 years. Today, I can tell you what it looks like in 12 months. Maybe.
That’s not a reason to panic. It’s a reason to move faster, stay closer to your customers, and be willing to reinvent your own product before someone else does it for you.
