▶ YouTube Summaries

• https://www.youtube.com/watch?v=f1jlJCNgtOg
• Original title: Layoffs are getting Wild (ft. Big A)

Prime and guest "Big A" (ex-Nvidia/Twitch, now a finance YouTuber) walk through four AI-cited layoffs ordered from least to most egregious. Recurring theme: the layoffs themselves are often defensible — companies should be able to hire and fire — but they feel brutal because the job market is so weak that laid-off workers have nowhere to go. Companies now feel forced to blame AI; saying a layoff is not AI-related (Intuit/TurboTax) gets punished by the market. The deeper worry: replacing senior staff with "AI + interns" risks losing the systems understanding needed to know when AI output is actually bad.

Meta — most defensible, still ugly

8,000 fired a day after leaked audio of Zuckerberg saying it's "not in your strategic interest" for the company to communicate everything. Meta is wildly profitable (set to overtake Google ad revenue ~2026/27), pays best-in-class severance, but is laying off into a dead job market. Big A: the recordings are obviously to train AI on employee work; the "Facebook brand tax" (paying extra to get people to work there) has only worsened. Layoffs are funding AI capex, not direct AI replacement.

ClickUp — the "100x" delusion

Cutting 22% while claiming the business is the strongest ever, promising a "100x organization" and $1M salary bands for big AI impact. Prime's math: 100x means a year's work in 3.65 days — physically impossible for decision-making, and software isn't a factory you can simply scale output on. No company has shown productivity gains remotely like this. Worse, a week before the layoff they released a video of a boss kicking a drink into an employee's face — terrible optics.

Webflow — most egregious behaviour

Employees locked out of laptops at 7am with no email or confirmation; some went 24 hours not knowing they were fired. Webflow (~1,600 staff) had done this once before — firing without notice, illegal in most jurisdictions including at-will California, especially brutal for workers on closed work permits who must relocate. Likely genuinely struggling: no-code site builders are being eaten by Claude/ChatGPT (cf. Wix losing ~50% of its value).

Cloudflare — the model to emulate

~1,100 cut while AI usage is up 600%, but with an industry-leading severance: full base pay through end of 2026, healthcare, equity vesting through Aug 15 including waived one-year cliffs — despite Cloudflare running a net loss last quarter. The one case where words match actions. Open question: they also hired ~1,100 interns, so this may be senior-for-intern cost-cutting, a bet that "AI + intern = experienced dev." Mitchell Hashimoto's anecdote frames the risk: a Ralph-loop agent spent ~$350/4h optimizing a renderer 88ms→1.5ms and looked incredible, but his hand-written version hit 20µs (~75x better). Without systems understanding, you accept mediocre AI output as great — "drinking from a fountain of mediocrity."