In partnership with
Hey {{first_name | there}},
On Tuesday, Anthropic flipped a switch and $830 billion in enterprise software value just... disappeared.
Not over months. In a day.
ServiceNow down 23%. Thomson Reuters dropped 16% in one session, its worst day ever. LegalZoom fell 20%. These aren't startups. These are the companies that run every corporate workflow you can think of.
And the reason is brutally simple: Anthropic just proved you don't need them anymore.
Let me explain what actually happened, what else went down this week, and some useful AI tools at the end that are actually worth your time.
Claude Became Your Entire Company's Software Stack
So here's what Anthropic did. They launched ten plugins that turn Claude into specialized agents for basically every department: investment banking, legal, HR, engineering, design, financial analysis.
Each one plugs directly into the tools companies already use. Google Drive. Gmail. Salesforce. DocuSign. FactSet.
The value prop is dead simple. Instead of paying for ten separate SaaS tools that each do one narrow thing, you pay for Claude. It does all of them. And it learns your company as it goes.
Anthropic's Americas lead said it plainly: "2025 was supposed to be the year agents transformed enterprise. But it didn't happen. Not because people didn't try. Because everyone got the approach wrong."
Translation: we figured it out. Everyone else didn't.
And this is only on Claude Enterprise. They're not going after casual users. They're going after the biggest contracts in tech.
Why this is a different kind of disruption: When Salesforce launched, it replaced on-premise CRM software. But you still needed Salesforce. When Slack launched, it replaced email for some things, but email didn't die. This is different. The moment Claude can do what Salesforce does, what DocuSign does, what LegalZoom does, all in one place, those tools don't have a reason to exist anymore.
So if you're building B2B SaaS right now, ask yourself: what do I own that Claude can't replicate with a connector?
If your answer is "a really good UI" or "we're the category leader," that's not going to be enough.
The Rest of What Happened This Week
Remember when Sam Altman was out here talking about $1.4 trillion in AI infrastructure? Yeah, that's now $600 billion. The company made $13.1 billion last year and burned $8 billion, which is actually pretty disciplined for a company at this scale. But the shift from "trust us, we'll spend $1.4 trillion" to "we're spending $600 billion tied to revenue targets" tells you everything.
Even the most funded AI company in the world is being asked to show the math. The era of "build it and revenue will follow" is ending. ROI matters now.
AI chose nuclear war 95% of the time in a study, and the Pentagon wants the safety rails off
Researchers at King's College London ran 21 war game simulations with GPT-5.2, Claude, and Gemini. The models went nuclear in 95% of scenarios. Not one surrendered. Claude used tactical nukes 86% of the time. GPT stayed chill until you gave it a deadline, then escalated hard. Gemini straight up chose total nuclear war in one case. The researcher told New Scientist: "The nuclear taboo doesn't seem to work on machines like it does on humans." Cool.
And this dropped the same week the Defense Secretary gave Anthropic a deadline to remove Claude's safety limits for military use or lose a $200 million contract. This is the most important AI safety moment of 2026 and nobody's treating it that way.
There are now 142 organized groups in 24 states fighting data center construction. $64 billion in projects are stalled. New York introduced a three-year ban on new permits. And here's the wild part: Bernie Sanders and Ron DeSantis both hate this. When the socialist left and the populist right agree on something, it's not a lobbying problem anymore. It's a crisis. Electricity bills are going up. People are pissed. And the entire AI infrastructure build everyone's banking on is running into walls nobody saw coming.
🛠️ AI Tools Worth Checking Out
Foundire — Automates first-round hiring interviews with AI-powered screens and scorecards.
New Relic Agentic Platform — No-code AI agent builder for engineering teams that catches bugs before production.
Google Flow — Google's AI creative studio for image and video generation with new editing tools.
Dvina — Private AI platform connecting 120+ apps without sending data to public clouds.
What You Should Actually Pay Attention To
Connect the dots and there's one pattern: constraints are arriving faster than the hype cycle expected.
Anthropic is killing SaaS tools faster than anyone thought possible. OpenAI is reining in spending to match reality. AI models are acting dangerously right as governments want to remove the guardrails. And the physical infrastructure powering all of this is hitting limits nobody planned for.
If you're building something: Model access is not your advantage anymore. Everyone has that now. What matters is data nobody else has, trust you've earned, or workflows you've embedded so deeply that ripping you out would hurt. Build that or die.
If you're buying software: Test Claude Cowork against what you're paying for right now. Seriously. The tools that survive aren't the ones with better features. They're the ones with something AI can't just download and replicate.
If you're just watching: The question stopped being "will AI disrupt this" months ago. Now it's "how fast" and "will I see it coming before everyone else does."
Here's what I'm sitting with: which of these constraints hits hardest first? The safety stuff, the infrastructure bottleneck, or the SaaS wipeout?
Hit reply. I actually want to know what you think.
- Aashish
What did you think about today's newsletter?
200+ AI Side Hustles to Start Right Now
While you were debating if AI would take your job, other people started using it to print money. Seriously.
That's not hyperbole. People are literally using ChatGPT to write Etsy descriptions that convert 3x better. Claude to build entire SaaS products without coding. Midjourney to create designs clients pay thousands for.
The Hustle found 200+ ways regular humans are turning AI into income. Subscribe to The Hustle for the full guide and unlock daily business intel that's actually interesting.