@FinanceYF5: 2/ The dividing line has been drawn, but no one has sent notice. None of the five flagship models have a high-income proportion below 50%, while among American adults that tier is only about half. Even worse is the low end: among Claude users, those earning under $50k per year account for only 4%, while among American adults that tier is 24%.

2/📊 The dividing line is already drawn—but no one has sent out the notice.

None of the top five flagship models have less than 50% high-income users, yet among American adults that bracket is only about half.

The low end is even uglier: among Claude users, those earning under $50k account for just 0.4%, while among U.S. adults that bracket is 24%. https://t.co/xdg3hkKpeO

1/ You think you’re using AI—but you’re actually being filtered out by it.

Among U.S. Claude users, 79.8% come from households earning over $100k a year.

This isn’t a rich person’s story—it’s the silent dividing line that fell in spring 2026.

2/ The dividing line is already drawn—but no one has sent out the notice.

None of the top five flagship models have less than 50% high-income users, yet among American adults that bracket is only about half.

The low end is even uglier: among Claude users, those earning under $50k account for just 0.4%, while among U.S. adults that bracket is 24%.

3/ Here’s what the offline user looks like:

55% of Meta AI users encounter AI passively in social apps; only 21% actively open the app to ask questions.

They use it one day a week, do 1-2 quick tasks per day, use the weakest free model, and then think, “I’m using AI too.”

4/ The real divide isn’t about whether you use AI, but which model you use.

Anthropic internal experiment: when Opus acts as the seller, it sells each item for $2.68 more on average; when it acts as the buyer, it pays $2.45 less per item.

The scariest line: the disadvantaged side never even realizes it’s being taken advantage of.

5/ Model capability is the means of production of this era.

Every decision gets slightly eroded by the model gap—thousands of times a year. The gap compounds; it’s not linear.

In the 19th century, it was who had a steam engine. In the 20th, who had a computer. In 2026, it’s who turned on Opus.

6/ The middle-class illusion: polishing your resume with the free version doesn’t count as getting on board.

Among ChatGPT users earning $100k, only 4% pay—3% are reimbursed by employers—the other 90% use the free version.

He bought a yearly gym membership but hasn’t gone for months—the real loss isn’t the membership fee, it’s an entire era of the AI dividend.

7/ People above the dividing line integrate AI as infrastructure.

47 sessions per day: process PDFs, run SQL, do architecture reviews, draft investor emails, and debug their kids’ Olympiad math problems.

The difference isn’t income—it’s initiative. An engineer making $90k pays for Claude Max out of pocket, outperforming a hedge fund trader making $500k who only uses the free version.

8/ The dividing line is so low that anyone can cross it.

Open Claude, build a real workflow, stick with it for three months—that’s the entire entry requirement.

Ridiculously low—but most people will spend a lifetime pretending it doesn’t exist.

In-depth analysis from WeChat public account [嘉妍Kea]

The AI divide: You think you’re using AI, but you’re actually being filtered out by it. https://mp.weixin.qq.com/s/eMyNv9Sc6uLzC5E1CjLWYA…

GenAI Summit SF 2026: A tens-of-thousands-person conference in the Bay Area!

Will Fans’ exclusive ticket purchase link (15% off): Exclusive Code: WILL Exclusive Links: Luma: https://luma.com/genaisummit26?coupon=WILL… Eventbrite: http://eventbrite.com/e/1985545163032/?discount=WILL…

https://x.com/FinanceYF5/status/2059886472443142593…

That’s all.

If you like this topic:

1. Follow me (@FinanceYF5)
2. Like + retweet the first post below.