The SaaSpocalypse Is Real. Here's What It Actually Means for Founders.
I tracked the revenue of 12 AI-native startups disrupting SaaS. The numbers tell a story Wall Street is just waking up to.
I've spent the last several weeks tracking how AI-native startups are dismantling traditional SaaS companies, company by company and category by category. The picture that emerged was too big for a single issue, so I'm breaking it into three parts. This week: the evidence that the disruption is already here, and a practical filter to tell what's vulnerable from what's safe. Next week: the opportunity map. The week after: five predictions for 2028, with receipts.
The Receipts Are In
On February 3rd, 2026, $285 billion in market cap evaporated from SaaS stocks in a single trading session. Workday, ServiceNow, Salesforce, HubSpot — the entire sector bled red. The trigger was Anthropic’s launch of Claude Cowork, a desktop agent that automates tasks across multiple applications simultaneously. Pundits immediately dubbed it the “SaaSpocalypse.”
But here’s what most of the coverage missed: the market crash was a lagging indicator. The disruption didn’t start on February 3rd. It started years earlier — and by now, it comes with receipts.
I’ve been tracking a dozen AI-native startups that are actively replacing traditional SaaS products. Not “threatening to disrupt” or “positioning to challenge.” Actually replacing them with real revenue, real customers, and growth rates that make traditional SaaS benchmarks look quaint.
Sierra, the AI customer support company founded by former Salesforce co-CEO Bret Taylor and former Google executive Clay Bavor, went from roughly $20 million in annual recurring revenue to $150 million in a single year, according to Sacra’s estimates. They don’t charge per seat like Zendesk. They charge approximately $1.50 per resolution. Their client list — WeightWatchers, SiriusXM, ADT, Sonos — reads like a who’s who of companies that used to pay Zendesk and Intercom millions annually.
Harvey, the legal AI platform, hit $190 million ARR in roughly three years from founding, according to TechCrunch reporting. That’s in a market — legal research and document review — where Thomson Reuters and LexisNexis have held a stranglehold for decades. Harvey is now valued at $11 billion.
And then there’s Lovable, the Swedish “vibe coding” platform. This one still makes me do a double-take every time I look at the numbers. Lovable hit $100 million ARR in eight months — the fastest in recorded startup history — and then doubled to $200 million ARR by late 2025. With 15 employees. Fifteen.
These aren’t projections in a pitch deck. They’re not “if adoption continues at this rate” forecasts. They’re actual revenue numbers from companies that are actively taking market share from SaaS incumbents.
And the pipeline behind them is staggering. Eighty-eight percent of Y Combinator’s Summer 2025 batch — 141 out of 160 startups — was classified as AI-native. That’s the highest concentration in YC’s history. The world’s most influential startup accelerator has essentially become an AI agent factory. The smartest founders have already placed their bet.
The Disruption Mechanism: Four Layers
So what’s actually happening here? The SaaSpocalypse isn’t one disruption — it’s four, happening simultaneously across different layers of how software gets built, sold, and used. Understanding which layer is being attacked tells you which SaaS categories are vulnerable and which are safe.
The Interface Layer is the most visible disruption, and it’s happening right now. AI agents are becoming the primary interface to software — not as a chatbot bolted onto an existing product, but as a complete replacement for the human who used to operate the tool. Sierra and Decagon (which just raised at a $4.5 billion valuation, per Bloomberg) don’t add AI to Zendesk. They eliminate the need for a human to use Zendesk. The customer talks to the agent. The agent resolves the issue. Zendesk becomes irrelevant.
This is what a16z general partner Sarah Wang meant when she wrote in the firm’s Big Ideas 2026 report that “the traditional system of record slips into the background as a commodity persistence tier.” The system of record — CRM, helpdesk, ITSM — doesn’t disappear. It gets demoted. It becomes the database that the AI agent reads from and writes to, while the agent itself captures all the strategic value.
