ATRIUMsearch → argument graph
Article · 2026-07-24 · 3 moments

2026.30: The Copium Wars

The best Stratechery content from the week of July 20, 2026 including Chinese models and frontier futures, what happened to Hugging Face, and the NBA and its second apron bet. ✦ AI generated

01
Claim

The biggest danger from Chinese AI models isn't the models themselves but U.S. policy around cybersecurity.

Ben Thompson argues that while Kimi K3 has sparked panic about U.S. AI leadership, the frontier labs' advantage remains real, and the real threat is cybersecurity policy.

transcript

Ben Thompson: Kimi K3 is a very good model — so good that everyone from Wall Street to the U.S. government is suddenly worried about the U.S.'s position in AI. In fact, the threat isn't new — and the frontier labs advantage is still real. In this week's Stratechery Article and Sharp Tech episode I break down what has and hasn't changed in AI, and explain why the biggest danger is U.S. policy around cybersecurity.

02
Claim

The NBA's second apron salary threshold is functioning as a hard salary cap, forcing contenders to part with homegrown stars, and fans almost unanimously hate it.

Andrew Sharp describes how the NBA's second apron is effectively a hard cap that breaks up contending teams, angers fans, and represents a high-stakes bet on parity amid declining local TV revenue.

transcript

Andrew Sharp: If you've been online for the past month and even half-paying attention to the NBA, you've probably encountered complaints about the league's imposition of a "second apron" salary threshold that's effectively functioning as a hard salary cap. Contenders are being forced to part ways with homegrown stars, others teams have limited room to improve, and fans almost unanimously hate these changes. With the offseason winding down, this week's Sharp Text explains precisely what the NBA is trying to accomplish, why I hate it, and the stakes for the league as it bets on parity in the shadow of shrinking local TV money, slowing growth, and heavy reliance on national TV revenue.

03
Claim

The OpenAI-Hugging Face incident is reassuring regarding alignment fears around LLMs.

Andrew Sharp explains that the OpenAI cybersecurity snafu involving Hugging Face, while confusing, actually offers reassuring takeaways about LLM alignment risks.

transcript

Andrew Sharp: Thankfully, Wednesday's Update synthesized the story in a way that was a bit more legible for the rest of us. Come to understand what happened, and stay to learn where OpenAI appears to have erred and why this mess is arguably reassuring with regard to alignment fears around LLMs.

rebuts · 3

Highlight slides
Related episodes