Morgan Stanley's 40-Point SaaS Crater Proves 'Vibe Coding' Is Real
Morgan Stanley's SaaS basket is trailing the Nasdaq by 40 points since December. HubSpot and Klaviyo are both down ~30%. Analysts are literally publishing notes titled "No Reasons to Own" on software stocks.
Welcome to the great B2B SaaS reality check of 2026.
The crash isn't happening because AI killed software—it's because customers realized they don't need to pay $50/seat/month for features they can now build themselves in an afternoon. Namanyay Goel calls this phenomenon "vibe coding"—the ability for non-technical users to quickly spin up custom tools using AI.
I've watched three previous "software is dead" cycles. The dot-com crash. Mobile eating the web. Cloud disrupting on-premise. This feels different.
<> "Traditional B2B SaaS relied on 'build once, pay forever' models when software development was complex, but AI democratizes building."/>
The math is brutal. AI-first SaaS companies are targeting 50-65% gross margins compared to the traditional 80%, thanks to inference costs. But they're compensating by slashing operations—AI chatbots are cutting sales and marketing spend from 40% to 15% of revenue.
The Retention Death Spiral Nobody Saw Coming
Here's what founders aren't admitting in board meetings: customers are canceling renewals not because they're unhappy, but because they think they can do better internally.
Goel, who runs hundreds of conversations with SaaS operators through his company Giga Catalyst, sees this pattern everywhere. Companies that spent years building "systems of record" are watching customers build "systems of preference" with ChatGPT and Claude.
The irony? The SaaS market still grew to $315B by early 2026. But that growth is incredibly selective. Vertical AI tools are unbundling legacy platforms while horizontal players scramble to become "orchestration hubs."
What Nobody Is Talking About
Every analysis focuses on AI replacing SaaS features. The real disruption is customer expectations.
When your marketing team can build a custom dashboard in 20 minutes, why wait three months for your CRM vendor's roadmap? When your ops team can automate workflows with natural language prompts, why pay for pre-built integrations?
SaaS Capital observes that AI impacts different companies "very unevenly," with most seeing only marginal effects. But marginal effects compound. A 5% churn increase here, 10% longer sales cycles there—suddenly your growth metrics look like 2022 all over again.
The winners are adapting fast:
- BetterCloud predicts AI becomes "foundational logic" by 2026, not bolt-on features
- AI marketing tools are delivering 42% CAC reduction through personalization
- Demand creation now means training AI on your support tickets and customer reviews
Meanwhile, the laggards are getting steamrolled. Goel's solution—whitelabeled AI platforms that let customers "vibe code" within existing SaaS—feels like putting a band-aid on a severed artery.
The Uncomfortable Truth
I'm skeptical of any blog post that ends with "coincidentally, here's my product that solves this exact problem." But Goel's underlying thesis rings true after watching software markets for two decades.
The build-versus-buy calculation just flipped. When building meant hiring developers, SaaS won. When building means describing what you want in plain English, the equation changes.
Smart SaaS companies are racing to become platforms where customers can extend functionality. The rest are hoping this AI thing blows over.
Spoiler: it won't.
Those Morgan Stanley analysts issuing "No Reasons to Own" ratings? They might be early, but they're not wrong. The SaaS business model that minted unicorns for the past decade is getting disrupted by customers who no longer need to wait for vendors to build what they want.
They can just vibe code it themselves.
