
Playbook
Tasteful Skills
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.
Apple may become Intel’s comeback customer — Intel jumped nearly 20% on a Wall Street Journal report that Apple is in a preliminary chip manufacturing deal, part of a broader White House-backed push that already brought in Nvidia and could reduce U.S. reliance on TSMC.
The market is increasingly an AI story, not a broad economy story — The hosts cite Greg Ip’s framing that the AI economy grew 31% while the non-AI economy grew just 0.1%, with the “AI Big 10” now making up about 40% of the market.
The U.S. labor market keeps defying the doom narrative — April jobs came in at 115,000 versus Wall Street Journal expectations of 55,000, with gains in retail, transportation, warehousing, and healthcare showing resilience outside tech layoffs.
There are ‘K-shapes within K-shapes’ in the real economy — Whirlpool cut its dividend for the first time since the 1950s and is down 80% over five years, while Six Flags reported higher revenue, attendance, and customer spending despite being a highly discretionary business.
DeepSeek is trying to buy its way back to the frontier — The Chinese AI lab is reportedly raising $7 billion at a $50 billion valuation, with founder Liang Wenfeng personally contributing 40% of the round as it races to secure compute.
Compute demand is overpowering AI rivalries — The hosts were struck by the SpaceX/xAI-Anthropic infrastructure deal, arguing that despite Elon Musk’s public hostility toward Anthropic, “demand for compute finds a way.”
The show opens on what they call “the great lock-in continues,” with Intel up nearly 20% after the Wall Street Journal reported a preliminary manufacturing deal with Apple. The key framing isn’t just Apple maybe returning in some form after the Apple Silicon split — it’s that the Trump administration and Commerce Secretary Howard Lutnick have apparently spent a year nudging Tim Cook, Jensen Huang, and Elon Musk toward Intel to rebuild U.S. semiconductor capacity.
They zoom out and connect the dots: the U.S. converted nearly $9 billion in federal grants into Intel stock, taking roughly a 10% stake at around $21 a share, and now the stock is around $125. Lip-Bu Tan is cast as the turnaround architect, reshuffling leadership, hiring ex-TSMC talent, betting hard on Intel’s 14A process, and trying to make the pitch every foundry wants to make: if we build it, will Apple, Nvidia, and others actually come?
From there, the conversation shifts from Intel specifically to the larger concentration trade. They cite a Bloomberg-style chart showing the “AI Big 10” — basically the Mag 7 plus names like AMD, Broadcom, and Micron — now accounting for about 40% of the market, which they compare with historical concentration episodes like railroads and Japan.
One host says Greg Ip’s Wall Street Journal piece helped disentangle what’s going on: AI-related growth is surging while much of the rest of the economy is barely moving. They run through the contrast: muted personal consumption, weak housing and business structure investment, but tech equipment up 43%, software up 23%, and everyone still arguing whether that means “obvious bubble” or “look at the revenue growth.”
Then comes the counterweight: the U.S. added 115,000 jobs in April, more than double the 55,000 economists expected. What makes the moment sticky is their disbelief that the numbers keep breaking the online narrative — layoffs in tech feel huge inside the AI bubble, but in a 100-million-plus-worker economy, a 4,000-person cut at a tech company barely registers next to hiring in healthcare, retail, transportation, and warehousing.
The hosts’ favorite example of “K-shapes within K-shapes” is a mini case study in two earnings reports. Whirlpool — a century-old appliance company that had paid a dividend continuously since the 1950s — just cut that dividend, while Six Flags, of all things, reported higher revenue, attendance, and customer spending; their punchline is that you can defer buying a fridge, but you can’t “vibe code Space Mountain.”
That leads into their “barbell thesis” for the AI era: own both the infrastructure and the irreplaceably human, anti-slop experiences. They mention the Ellison family’s exposure to Oracle on one side and Warner Bros. IP like Batman and Superman on the other, and Josh Kushner backing both OpenAI and the San Francisco Giants — weird pairings that start to make sense if AI amplifies both digital abundance and scarcity-premium experiences.
The back half turns into a quick-hit roundtable: DeepSeek reportedly raising $7 billion at a $50 billion valuation, with the founder putting in $3 billion himself; Tyler’s field report from the OpenAI/Elon trial, where courtroom boredom was broken up by text-message drama and even an AI-generated Sam Altman emo song; and finally the SpaceX/xAI-Anthropic infrastructure deal. Their takeaway is simple and very TBPN: in AI, cultural beef matters less than available power and GPUs — compute gets sold, and then sold again.
Share
Keep Reading
The Weekly Echo. The inbox-shaped summary of what mattered.
New editorials announced here.

Playbook
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.

Playbook
Learn how tasteful prompting helps you move beyond generic AI output by shaping context, style, and judgment from the start.

Playbook
OpenAI shipped /goal for the Codex CLI. It turns a prompt into a persisted, self-continuing contract.