Palantir Drops 7% as Agentic AI Panic Hits Software Stocks

Palantir Technologies fell 7% on Friday, April 10, capping a brutal stretch for enterprise software that has seen the IGV software ETF drop more than 35% from its recent high and 28% year-to-date in 2026. The proximate cause: investor Michael Burry's claim that Anthropic is "eating Palantir's lunch" with its Claude Cowork agentic AI platform, combined with inflation data that killed hopes of rate cuts for the rest of the year.

The sell-off isn't really about Palantir, though. It's about a bigger question the market is now trying to price: does the SaaS (software-as-a-service - subscription software) business model survive a world where AI agents can do the same work for less?

The Specific Trigger

Burry's Palantir criticism landed in a market already primed to panic. Anthropic's annual recurring revenue reportedly grew from $9 billion to $30 billion in the four months before April, and Claude Cowork — Anthropic's agentic platform for automating legal, enterprise, and operational workflows — is being positioned as a direct replacement for entire categories of SaaS.

The specific data point that spooked investors: the legal tech sell-off that started in early February, when Anthropic released its Legal plugin for Claude Cowork. That plugin automates contract review, NDA triage, compliance workflows, and legal briefings — capabilities that underpin the pricing models of public legal tech vendors. Those stocks never recovered.

On April 10, Palantir's drop added a second data point: even companies selling "AI platforms" aren't immune if a foundation model company ships a competing product directly.

The Macro Overlay

The same Friday brought a headline inflation print that pushed annual CPI to 3.3%, with month-over-month growth of 0.9%. Traders responded by pricing in a 78% probability that the Federal Reserve will not cut interest rates for the remainder of 2026. Higher-for-longer rates hit high-multiple software stocks hardest because their valuations depend on discounting future cash flows at lower rates.

This is the "perfect storm" frame: AI disruption anxiety plus macro headwinds plus crowded positioning in a handful of software names all hitting at once. Any one of those factors might have been absorbable. All three together produced the sharpest software sell-off in five years.

JPMorgan analysts have called the broad software sell-off a case of stocks being "sentenced before trial" — punished for disruption that hasn't actually materialized in most of their revenue lines yet. That framing has merit, but it also misses that markets typically start pricing disruption well before it shows up in reported numbers.

What's Actually Disrupting

The SaaS disruption narrative has specific shapes, and they're worth separating from the general panic.

First, vertical point solutions with thin AI wrappers on top of GPT or Claude look most exposed. If your company's entire product is a prompt plus some UI, foundation model platforms like Cowork are now shipping that capability directly.

Second, workflow software where the work is mostly cognitive — legal review, compliance monitoring, customer service, basic accounting — faces real pressure. Agentic AI platforms can plausibly automate end-to-end tasks that previously required SaaS tools plus human operators.

Third, software with deep data moats, proprietary integrations, regulatory certifications, or long-tail enterprise workflows is much less exposed. Salesforce and ServiceNow can be threatened in principle, but the switching costs are enormous and the regulatory surface area around their data is real.

The problem for the market is telling category one from category three in real time. When in doubt, traders are selling.

Burry vs. The Fundamentals

Michael Burry's Palantir critique is provocative but not obviously right. Palantir's government revenue — a large majority of its business — is tied to long-term contracts with defense and intelligence agencies that are not going to be replaced by a Claude plugin any time soon. Its commercial business has real customer concentration risk, but it also has deployments that are deeply embedded in customer operations.

The Anthropic ARR numbers Burry cited are also worth treating skeptically. $30 billion ARR at Anthropic would make it one of the fastest revenue builds in software history, and while Anthropic is growing fast, the $30 billion figure has been reported with varying levels of confidence. Some of it may be contracted future revenue, not run-rate.

Still, directionally, Burry's argument lands because the market is already primed to believe it. The disruption narrative is doing most of the work, and any high-profile voice amplifies it.

What Enterprise Buyers Should Watch

For anyone making software buying decisions in the next 12 months, the sell-off creates both signal and noise.

The signal: your existing vendors are going to feel pressure to ship agentic features or sharpen pricing. Use that leverage. Renewals and new deals are likely to be more negotiable through the second half of 2026 than they were in 2025.

The noise: individual stock price moves do not necessarily reflect what a piece of software is worth to your organization. A vendor whose stock is down 40% may still be the best tool for your specific job. Don't outsource procurement decisions to the stock market.

What's Next

Anthropic, OpenAI, and Google are all racing to ship more complete agentic platforms, and each new release is likely to trigger another wave of software stock repricing in adjacent categories. Expect volatility in legal tech, customer service software, marketing automation, and expense management through at least the middle of 2026.

The Federal Reserve's next meeting is May 6. If inflation cools unexpectedly or rate-cut odds shift, the macro overlay on the AI disruption story could ease. But the agentic AI piece isn't a macro story — it's structural, and it doesn't unwind if rates drop.

The bottom line: the market is trying to price a transition from "AI helps software run" to "AI replaces software." That's probably too simple as a thesis, but directionally it's what's driving the sell-off. For operators, the right question isn't whether your current vendor's stock is going up — it's whether their product still beats the latest agentic platform at the specific job you hired it for. Run the test.

Sources: CNBC on Anthropic Agent Sell-off | Palantir Drop Deep Dive | Bloomberg on Legal Tech Plunge