The data doesn’t lie. But the story is more complex than “SaaS is dead” – it’s that B2B/SaaS without AI is dead for VC investors at least. Not just at growth, but at seed as well.
We’ve never seen a sector rotation this dramatic, this fast. The latest AngelList data from the first half of 2025 tells a story that every B2B and SaaS founder, investor, and operator needs to understand immediately.
SaaS investment has fallen off a cliff. It’s almost frozen at seed stage. But it’s not that simple, as many “SaaS” deals are now AI / AI+B2B deals.
That massive 40% “AI/ML” category isn’t just infrastructure plays and foundational models. It’s packed with AI-powered B2B applications that would have been categorized as “SaaS” just two years ago.
The numbers tell a complex story:
- AI/ML captured roughly 40% of seed deals in 1H 2025
- Traditional SaaS dropped to 3-4% of deal volume
- But many AI+B2B apps are now classified as “AI/ML,” not “SaaS”
What’s Really Getting Funded in That AI/ML Category
This isn’t just OpenAI competitors and GPU infrastructure. When I dig into the actual deals, that 40% AI/ML bucket breaks down roughly like this:
~15% Pure AI Infrastructure: Foundation models, AI chips, training platforms, MLOps tools ~25% AI-Native B2B Applications: What we used to call “vertical SaaS” but now powered by LLMs and ML from day one
Think about it: that AI-powered legal research platform? That’s in the AI/ML bucket now, not SaaS. The AI customer service agent that replaces traditional helpdesk software? AI/ML category. The AI-powered financial planning tool that would have been “Fintech SaaS” in 2022? Now it’s AI/ML.
The Real Story: SaaS Isn’t Dead, It’s Been Absorbed
What we’re seeing isn’t the death of B2B software – it’s the complete transformation of what investors consider fundable B2B software.
The New Categorization Logic:
- Traditional SaaS (3%): Workflow tools, dashboards, CRMs without meaningful AI
- AI/ML (40%): Any B2B software where AI is the core value proposition
- Vertical Categories: Healthtech, Fintech, etc. – often AI-powered but industry-specific
This explains why Developer Tools (7-8%) is doing relatively well – those are increasingly AI development tools, but they’re still categorized by their end market rather than their AI capabilities.
What This Really Means for B2B Founders
The lesson isn’t “don’t build B2B software.” It’s “don’t build B2B software without AI as the primary differentiator.”
The Funded “SaaS” Companies Are All AI-First:
- Sales intelligence platforms that predict buyer intent using LLMs
- Customer success tools that automatically identify churn risk and generate interventions
- Project management software that uses AI to automatically prioritize and resource allocation
- HR platforms that use AI for candidate matching and performance prediction
These companies are getting funded and growing fast. They’re just not being categorized as “SaaS” anymore.
The Capital Deployment Reality
Here’s what’s fascinating about the capital deployment patterns (those orange bars): traditional SaaS is getting small checks, while AI-native B2B applications are commanding premium valuations and larger seed rounds.
Why? Because investors believe AI-powered B2B tools can:
- Scale faster: AI handles tasks that required human-heavy customer success
- Defend better: Network effects from AI learning create stronger moats
- Expand quicker: AI can identify upsell opportunities automatically
- Command higher prices: AI-driven insights justify premium pricing
The Categories That Are Actually Winning
Let’s reframe what’s really happening:
Healthtech (15%): AI-powered diagnostic tools, AI care coordination, AI clinical decision support Fintech (8-10%): AI fraud detection, AI underwriting, AI personal finance management
Developer Tools (7-8%): AI code generation, AI testing, AI deployment optimization
Notice the pattern? Every winning category has AI baked into the core value proposition.
What This Means for Fundraising Strategy
If you’re building B2B software in 2025, your positioning strategy needs to flip:
Don’t say: “We’re a SaaS platform with AI features”
Do say: “We’re an AI platform that delivers [specific business outcome]”
The companies raising successfully are leading with AI capabilities, not software features. They’re solving problems that genuinely couldn’t be solved before LLMs and modern ML.
The Real Threat to Traditional SaaS
The existential threat isn’t that investors don’t want B2B software. It’s that they don’t want B2B software that can be easily replicated by ChatGPT, Claude, or the next wave of AI agents.
If your SaaS product’s core value is organizing information, generating reports, or basic workflow automation, you’re competing directly with increasingly capable AI assistants. Why fund your startup when enterprises can just train an AI agent to do the same thing?
The Bottom Line (Revised)
B2B software is alive and thriving – it’s just all AI-powered now. The 40% of seed deals going to “AI/ML” includes tons of applications that solve traditional business problems, just with AI as the primary mechanism.
Traditional SaaS – software that moves data around without intelligent processing – is indeed dying at seed stage. But AI-native B2B applications are absolutely crushing it.
The question isn’t whether to build B2B software. It’s whether you’re building B2B software that leverages AI to solve problems in fundamentally new ways.
The great rotation isn’t from SaaS to AI. It’s from human-powered software to AI-powered software. And that distinction is everything.
