The Obvious AI Startup Ideas Are Already Taken. Here’s What’s Actually Open.
I spent weeks mapping claimed territories, open verticals, and a hidden infrastructure layer that Y Combinator is quietly betting everything on
Series: Part 2 of 3 — The SaaSpocalypse Series
Last week, I laid out the evidence that AI-native startups are dismantling traditional SaaS — company by company, category by category. I introduced a practical filter (probabilistic vs. deterministic tasks) to tell what’s vulnerable from what’s safe. If you missed it, start with Part 1.
This week: the opportunity map. Where the capital is flowing, which categories already have billion-dollar pseudo-incumbents, and — most importantly — where the actual white space is for founders at the pre-seed to Series A stage. Next week: five predictions for 2028, with receipts.
The $285B Didn’t Disappear — It Moved
Let me start with a number that should reframe how you think about the SaaS market crash from Part 1.
In 2025, AI startups captured 47% of all venture capital globally — $238 billion out of $506 billion total. That’s not a typo. Nearly half of all venture dollars on the planet went to AI companies in a single year, according to funding trackers. The year included twelve rounds exceeding $1 billion. OpenAI alone raised $40 billion — the largest private funding round in history. Anthropic closed $13 billion at a $183 billion valuation. xAI added $10 billion. Anysphere, the company behind Cursor, raised at a $29.3 billion valuation.
So when $285 billion vanished from SaaS market caps, it didn’t evaporate. It migrated. Capital is flowing from the companies being disrupted to the companies doing the disrupting — and that flow is accelerating.
But here’s what matters for you as a founder: most of that capital went to a handful of foundation model companies and late-stage incumbents. The application layer — the layer where you build — is still being carved up in real time. Some territories are already claimed by billion-dollar players. Others are wide open. And a third category — the infrastructure that makes AI agents actually work in production — is so early that Y Combinator’s entire Winter 2025 batch was essentially a bet on it.
That’s the map I’m going to walk you through. Three zones: claimed territories, next-wave verticals, and the meta-layer. Each one offers a different risk-reward profile and requires a different entry strategy.
Claimed Territories — The Billion-Dollar Pseudo-Incumbents
Let’s get the obvious ones out of the way first. These categories already have AI-native companies with massive valuations, significant revenue, and — critically — enough market presence that competing head-on is a losing proposition for an early-stage founder.
Customer Support is the most thoroughly claimed territory. Sierra ($150M ARR, $10B+ valuation) and Decagon ($4.5B valuation, per Bloomberg) have proven the model and locked up enterprise logos. Crescendo.ai carved out a niche with its hybrid AI-plus-human approach and hit $91M ARR. If you’re thinking “I’ll build an AI customer support agent,” you’re walking into a room where three well-funded competitors have already eaten most of the pie.
Legal belongs to Harvey. $190M ARR, $11B valuation, and institutional relationships with firms that took Thomson Reuters and LexisNexis decades to build. Harvey didn’t just build a legal AI tool — they built a legal AI platform that firms are restructuring their workflows around.
Coding and Vibe Coding is the most crowded claimed territory of all. Cursor ($29.3B), Lovable ($6.6B, per Sacra), Bolt.new, and Replit collectively represent over $48 billion in combined valuation. This is the category where the revenue velocity numbers from Part 1 came from — Lovable hitting $200M ARR with 15 employees — and it’s attracting more capital and more founders than any other.
Marketing and Content is claimed but less defensible. Writer ($1.9B) and Jasper are the leaders, but the moats here are shallower because the switching costs are lower. This is a category where a well-positioned vertical play could still win.
Now, does “claimed” mean “no opportunity”? No. But it means the obvious entry points are closed. What remains are the wedges — the specific niches within these categories that the pseudo-incumbents haven’t prioritized. In customer support, that might be industry-specific compliance requirements (healthcare support agents that understand HIPAA, financial support that handles regulatory disclosures). In coding, it might be the “code cleanup” market — tools that fix and maintain AI-generated codebases rather than creating new ones. In legal, it might be in-house corporate legal teams rather than law firms.
The lesson from claimed territories: if you can describe your startup idea in five words and a billion-dollar company already does it, you need a sharper wedge.
Next-Wave Verticals — Where the Smart Money Is Pointing
This is the section that matters most for founders at the pre-seed to Series A stage. These are verticals where the disruption thesis is proven, capital is flowing in, but no company has achieved the kind of market dominance that Sierra has in support or Harvey has in legal. The window is open. For each one, I’ll break down why now, who’s moving, and where the wedge is for early-stage founders.
Finance and Accounting
This is the vertical I’m watching most closely. Sapphire Ventures partner Rajeev Dham said it directly: “The Harvey for finance hasn’t been built yet.” And the structural conditions are screaming “now.”
The accounting industry faces what I’d call a perfect storm: 75% of CPAs are retiring within the next 15 years, complexity is increasing due to globalization and new regulations, and private equity consolidation is driving brutal productivity demands on the firms that remain. The $900 billion accounting labor market is one of the largest professional services categories on the planet, and it’s still running on workflows designed for the paper era.
Accrual just came out of stealth with $75 million in funding from General Catalyst — building what they call AI-native accounting infrastructure. Their early numbers are remarkable: 85% reduction in tax preparation time, 60% reduction in review time. John Karls, national leader of Armanino’s Private Client Advisors tax practice (a top 20 firm), said that “returns coming out of Accrual are already at manager-review quality.”
But Accrual is focused on accounting firms. Look at Y Combinator’s Winter 2025 batch and you’ll see the vertical being attacked from every angle: Finto (AI accounting for enterprise, automating invoice-to-pay, cutting 80% of daily work), FullSeam (AI employee that logs into your accounting, billing, banking, and CRM tools), Cranston (full-stack AI accounting firm targeting the entire $900B labor market), and Denki (automating internal auditing for SOX compliance).
The wedge for founders: The big players are going after large accounting firms and enterprise. The massive underserved market is SMBs — the millions of small businesses that can’t afford a top-tier accounting firm and are currently using QuickBooks and spreadsheets. An AI-native accounting agent priced at $200/month that replaces a $2,000/month bookkeeper is a wedge that none of the current players are optimizing for.
Healthcare Operations
If finance is the “perfect storm,” healthcare is the tsunami. Digital health funding hit $14.2 billion in 2025, the highest since 2022, and AI companies captured 54% of that total — up from 37% the year before. The broader healthcare AI market is projected to reach $208.2 billion by 2030, and healthcare organizations are deploying AI at 2.2x the rate of the broader economy.
But here’s the pattern I find fascinating:
