I put this chart together for my SaaStr AI Annual opener last week. Of the 5 things we kept hearing through 2025 and through Q1 2026, and what we’re actually seeing now that the agents have shipped and the budgets have hit.
#1. “Everyone will vibe-code their own CRM” → “Our CRM brings in deals itself”
Remember the panic in 2025? Every founder would spin up their own internal CRM on Replit. Salesforce was doomed, HubSpot was dead, the whole category was getting destoryed by vibe coding. It was a big part of the SaaSpocolypse.
Well, older B2B players without an AI Agent to really sell have 99 problems. But vibing your own CRM isn’t really one of them. It didn’t happen. Not at scale.
What did happen is that the CRMs got agents inside them. Agentforce running win-back campaigns at 72% open rates over here at SaaStr. Qualified closing $1M+ as our inbound BDR before a human ever touches the deal. Artisan running three concurrent outbound SDR instances. The CRM didn’t get replaced. It grew teeth.
We all now vibe-code internal tools, n=1 apps (just for us), connectors, and much more.. We did. QBee is one of them, runs our entire sponsorship operation with 70% fewer human hours. But the workflow apps that survive sit on top of the system of record. They don’t replace it. The CRM is still where the truth lives. It just stopped waiting for humans to type into it.
#2. “AI hallucinates too much to tryst” → “Our AI agents are idle too often”
The complaint a year ago: my AI agent made stuff up.
The complaint this quarter: my AI agent isn’t doing enough.
This is the actual frontier conversation with buyers right now. They turned on the agent. Hallucinations are still there but got mostly fixed with proper grounding, tool use, and the right model. And now they look at their agent dashboard and see… not much. The agent ran 14 tasks today. There are 400 things it could be doing.
#3. “SaaS is dead” → “B2B + AI ROI + AI budget = rocketship”
The narrative through Q4 last year was that B2B was finished. Margins compressed, NDR collapsed, AI was going to eat seat-based pricing alive.
The reality in May 2026: the winners are growing faster than they ever have.
Public B2B got cut in half in February. The companies that came back are the ones with measurable AI ROI and customers who have actual AI budget to spend. Look at Databricks at $5.4 billion in annual revenue as of February, up 65% year-over-year, with a $134 billion valuation after a $5 billion raise. CrowdStrike and Palo Alto both still growing 25%+ at scale. The winners are pulling away.
It’s not dead. It bifurcated. If you have AI ROI you can prove in a customer’s QBR deck, you are growing 60%+ this year. If you don’t, you’re getting churned out of the budget cycle.
#4. “Layoffs” → “Exponential productivity”
Both can be true at the same time, and both are. The layoffs are real. The productivity gains are also real, and they’re bigger than most people are admitting publicly.
SaaStr AI runs on roughly 3 humans and 20+ agents. 10K, our AI VP of Marketing, is 14,230+ lines of code and runs the Monday standup. QBee runs sponsor ops for 100+ partners. Revenue swung from -19% to +47% YoY with fewer humans, not more.
This isn’t a SaaStr-only story. Everyone truly running AI for coding and other AI agents at scale and running 3-10x output per headcount. It’s exhilirating and exhausing at the same time.
The productivity gains aren’t coming. They’re massive and already here.
#5. “IPO window shut, M&A down” → “The greatest IPOs + biggest deals of all time”
This one is the most factually wrong of the bunch, and somehow it’s still the prevailing narrative in B2B Twitter.
Start with the Big 3. SpaceX at $1.5T projected exit value. OpenAI and Anthropic both at almost $1T. Combined: $3.5 trillion. Per PitchBook, those three listings alone would create more exit value than every single US VC-backed IPO since 2000.
Combined. Meta, Uber, Snowflake, Rivian, DoorDash, Snap, Lyft, Robinhood, Coupang, every one of them. All of them added together are smaller than the Big 3. $110B of capital invested across the three. $3 trillion in projected exit value. 27x aggregate return on invested capital, if it works.
And that’s just the top of the stack. Databricks at $134B with $5.4B in revenue, up 65% YoY. Stripe finally cracking $100B+. Canva at $42B and ~$3.3B ARR. Discord, Plaid, Cerebras all queued up.
M&A is bigger that ever, too. SpaceX has an option to acquire Cursor for $60 billion by year-end, or pay a $10 billion breakup fee if they walk.
Cursor went from $1B ARR in early 2025 to $2B in February 2026, with a projected $6B run rate by year-end. That’s the most valuable application-layer AI company on the planet, and it’s getting bought by a rocket company. The BlackRock / Microsoft / Nvidia / MGX / xAI consortium bought Aligned Data Centers for roughly $40 billion, one of the largest infrastructure deals on record. Google bought Wiz for $32B. Palo Alto bought CyberArk. Total M&A volumes across sectors hit $4.3 trillion in 2025, up 39% from 2024. Technology M&A deal value in March 2026 alone reached $150.4 billion, up 31% year-over-year.
The window isn’t shut. It’s the most open it’s been since 2021, and it’s funding the biggest deals in the history of the industry.
The 2026 B2B + AI Playbook
If you’re running a B2B + AI company in 2026, the playbook is pretty clear:
- Your CRM should be running your sales motion, not the other way around. If your agents aren’t sourcing, qualifying, and closing inside your CRM, you’re paying for shelfware.
- Measure agent utilization, not agent accuracy. The accuracy needs to be managed but done right, is muchly solved. The utilization is the new bottleneck.
- Build an agentic product with true, provable AI ROI for the customer. Not “we use AI.” Customer dollars saved or customer revenue earned. In their P&L, not yours.
- Run lean & AI pilled. Three humans and twenty agents beats thirty humans and zero agents. Every single time.
- Get ready for liquidity. The window is open and the deals are the biggest they’ve ever been. Get your house ready.
The doom narrative around B2B was wrong. The takeoff narrative is, if anything, undersold.
That was the message at SaaStr AI Annual 2026. And it’s only getting more true from here.
