In partnership with

Today's news has a theme: money and infrastructure. Meta priced its new AI agent like a luxury subscription, SpaceX and Nvidia started building AI data centers in orbit, and Hugging Face got offered $13 billion and said not so fast. None of it changes what you do today, but it shows you exactly where this is all headed. Let's get into it.

Turn Great Books Into Better Habits.

The best ideas from the best books, distilled into practical insights you can actually use.

Go deeper on ideas like:

  • Atomic Habits: Build better habits with small, consistent changes.

  • The 7 Habits of Highly Effective People: Develop principles for a more effective life.

  • The Meaning of Your Life: Explore what actually makes life meaningful.

With Shortform, you get the key ideas, arguments, and insights from thousands of nonfiction books, all in one place.

Why Shortform?

  • Bite-sized guides you can finish in minutes

  • Key ideas explained clearly and in context

  • Practical takeaways you can put into action

  • New books and ideas to explore every day

Meta Just Put a $200-a-Month Price Tag on Its New AI Agent

What happened: Meta is reportedly rolling out a new consumer AI agent platform called Hatch within the next few weeks, built on its OpenClaw technology, according to Yahoo Tech, citing The Information. Hatch would let a personal AI agent order from DoorDash, browse Etsy, and interact with Reddit and Yelp on your behalf from a customizable dashboard. Internal documents show Meta considered a premium tier priced up to $199.99 a month for expanded usage, with limited WhatsApp trials starting as early as this week. Meta is also planning a bigger model called "Watermelon" for an October release.

Why it matters to you: If AI agents that shop and browse for you become a real subscription product, expect every tech company you already pay for tools to test the same move on you. Watch your software stack closely for "agent" add-ons that quietly show up as a new line item on your bill.

What to do about it: Before you add any paid AI agent tool to your business, ask exactly what task it replaces and what it costs per month, not per feature.

SpaceX and Nvidia Are Now Building AI Data Centers in Orbit

What happened: SpaceX's AI division and Nvidia announced a partnership to run Nvidia's new Vera Rubin computing hardware in space, starting with a satellite system called Starmind, according to Nvidia's own announcement. The plan is to eventually run "gigawatts" of AI computing capacity in orbit instead of on the ground. Nvidia says its new Vera chip completes tasks up to 1.8 times faster than standard processors, the kind of performance jump that makes an orbital data center worth attempting in the first place. No firm launch date was given, but SpaceX's Starmind satellite is the first target.

Why it matters to you: This is years away from touching your business directly, but it's a signal of how much money and physical infrastructure is now being bet on AI demand staying this high. When that many billions move into concrete, satellites, and orbit, it's not a trend that reverses quietly.

Hugging Face Just Got a $13 Billion Offer, and Said Not So Fast

What happened: Hugging Face, the platform millions of developers use to store and share AI models, is reportedly fielding acquisition offers north of $13 billion, according to TechCrunch. The company hasn't named who's interested, and CEO Clem Delangue says Hugging Face is close to profitable and isn't in a rush to sell. Hugging Face turned down a $500 million investment from Nvidia last year rather than let one company hold that much sway over its ownership. That resistance to a single controlling investor appears to still be the sticking point.

Why it matters to you: A lot of the free and cheap AI tools your business might use quietly depend on models hosted through Hugging Face, even when it never shows up in the name. If it gets bought by a single tech giant, pricing and access on that infrastructure could change fast.

Chinese State Hackers Are Using AI to Double Their Attack Volume

What happened: Security researchers say Chinese state-backed hacking groups have roughly doubled the number of attacks they can run by using DeepSeek, a low-cost Chinese AI model, to automate parts of the work, according to Bloomberg. The AI handles reconnaissance and scripting tasks that used to take a human hacker hours, letting the same team hit more targets in a day. Researchers say this pattern, cheap AI increasing attack volume rather than sophistication, is becoming the norm across state-sponsored hacking groups, not a one-off.

Why it matters to you: More attack volume means more phishing emails and more automated break-in attempts hitting small businesses too, not just big targets. Cheap AI does not discriminate by company size.

What to do about it: If you haven't required multi-factor authentication on your email and financial accounts yet, this is a good week to finally do it.

Claude Is Now Watermarking Everything It Writes, and Users Are Furious

What happened: Anthropic confirmed it has started adding invisible watermarks to text and images generated by its Claude AI models, according to TechCrunch and Forbes. The watermark is meant to help identify AI-generated content as detection tools and regulations catch up. Subscribers have pushed back hard, arguing that content they paid to generate and then edited themselves shouldn't carry an invisible tag without clearer disclosure. Anthropic has defended the move as part of an industry-wide push toward labeling AI output responsibly.

Why it matters to you: If you or your team use Claude to draft marketing copy, emails, or content, assume it may carry an invisible marker that some tools can detect later. That's worth knowing before you publish AI-assisted writing as fully your own.

OpenAI Just Cut Its AI Prices Again, This Time Up to 33%

What happened: OpenAI cut API pricing on its GPT-5.6 Sol model by 20 to 33 percent through November 21, according to its own developer community announcement, confirmed by Amazon's AWS Bedrock pricing update. It's the third price cut on that model family in about a month, as OpenAI, Anthropic, and Google keep undercutting each other to win business customers. Output token pricing dropped from $30 to $20 per million tokens for the discounted window.

Why it matters to you: If your business pays for any AI tool built on OpenAI's models, this is a good moment to check whether your vendor is actually passing the savings to you or just pocketing a bigger margin.

What to do about it: Ask your AI vendor directly whether their pricing has changed since this cut. If they can't answer quickly, that tells you something.

A New $103,000 Visa Fee Could Hit Any Business That Hires Skilled Tech Workers

What happened: The Trump administration proposed a new $103,265 fee for H-1B visa applicants after a court blocked its earlier version of the fee, according to The Washington Post and Bloomberg. The fee would apply to new applicants in the capped visa program, which covers roughly two-thirds of computer-related job approvals. The administration's own Department of Homeland Security estimates the fee could disrupt hiring plans for a large share of small businesses that rely on the program.

Why it matters to you: If you've ever considered hiring international talent for a technical role, AI-related or not, this fee changes the math significantly. Even businesses that never planned to sponsor a visa should know the skilled labor pool just got more expensive for everyone competing for it.

Most Young Adults Are Now More Worried Than Excited About AI

What happened: A new Pew Research survey found that 55 percent of U.S. adults under 30 now say they're more concerned than excited about AI, up from just 31 percent in 2021. 73 percent of that age group now believe AI will lead to fewer jobs, up from 61 percent two years ago. The shift shows up across the wider population too: 71 percent of all Americans now expect AI to reduce the number of jobs available over the next two decades.

Why it matters to you: Your youngest customers and employees are trending skeptical of AI, not excited about it. If your marketing or hiring pitch leans heavily on "powered by AI," that framing may land worse with under-30 audiences than you'd expect.

What to do about it: Test how your AI-related messaging actually lands with younger customers before you assume it's a selling point.

The Bottom Line

Here's the throughline today: everybody's placing enormous bets on AI staying exactly as hot as it is right now. Meta is pricing agents like premium subscriptions. Nvidia and SpaceX are ready to put data centers in orbit. Hugging Face is fielding a $13 billion offer instead of taking it. Meanwhile, the people who'll actually use this stuff, especially the ones under 30, are getting more nervous about it, not less. You don't have to pick a side on any of this today. You just have to keep asking what any of it actually does for your business before you pay for it. That's the whole rollercoaster in one sentence.

Enjoying the Ride?

If this issue saved you from reading eight different AI newsletters yourself, forward it to one person who'd get something out of it. New here? Subscribe below so tomorrow's issue lands in your own inbox too.

Talk tomorrow,
Mark Shilensky