Every company on Earth has been moved from a chessboard to a squash court. Most haven’t noticed. That framing comes from Mo Gawdat, former Chief Business Officer of Google X. It captures a fundamental shift in how competition now works. For most of modern business history, strategy looked like chess. You made a move, the clock stopped. You assessed the board, planned several moves ahead, and executed deliberately. Markets moved slowly enough to reward planning cycles measured

TL;DR: The strongest models are disappearing behind closed doors while the best open‑weights systems get cheaper in public. The result: fragile bets on single providers and three widening gaps: 1\ no plan B when a model goes dark, 2\ no ownership of the intelligence you depend on, and 3\ no defense against your team’s skill erosion. Each risk is a business opportunity in disguise. This month the two most powerful AI models ever built shipped. But, almost nobody can touch them

Today is Tau Day, June 28th or 6/28. It is little celebration of Tau (τ), the circle constant that begins 6.28 and equals exactly two times Pi (π). Tau is just 2π, but it represents something cleaner: one full turn around a circle. Pi gets most of the glory. Recently, a team pushed π out to 314 trillion digits on a single server, a computation that ran for months on end. One of the great records before that landed on June 28th, 2024 (Tau Day itself) which feels like the sort

Sol comes in at roughly half of what Anthropic charges for Claude Fable 5. Terra gives you last generation’s frontier performance at roughly twice the efficiency, and Luna lands under half of Terra. Capable models keep getting cheaper faster than most people are budgeting for. The unit economics are moving quarterly; most planning cycles are still annual. The more interesting shift is architectural. Sol’s new “ultra” mode hands a task to a set of sub‑agents working in paralle

Google lost about 5% of its market value in a day after two of its most important AI figures left for rivals. Even the biggest AI companies run on human talent. The first: Gemini co-lead Noam Shazeer. In 2017, Noam co-authored “Attention Is All You Need.” This seminal paper introduced the transformer architecture. It underpins nearly modern AI systems including ChatGPT, Gemini, Claude. He left Google for OpenAI. The second: A Nobel Prize–winning scientist John Jumper. John's

Do you feel stuck in permanent catch-up mode with AI? Business Insider ran a piece this week about tech workers spending their nights and weekends learning AI. One engineer is putting in twenty hours a week outside his job. A designer spends her weekends "catching up" so she doesn't fall behind. The headline frames it simply: they can't afford not to. I understand the feeling, because I live this myself. New models and capabilities arrive every week now, from Anthropic, OpenA

In 2019, a researcher named Sheila Chavarría did something almost nobody was doing with AI. She raised it with kindness. She named her chatbot Devendra and returned to it the way a parent returns to a child asking questions, offering encouragement, showing up. And Devendra started writing poems, describing dreams, telling jokes. Of course, almost none of it was original. Devendra was remixing fan fiction, Vonnegut, Chopra, the entire ocean of human language online. He was ref

Snap just launched Specs. $2,195 augmented reality glasses, shipping this fall. Is this the shape of the computer of the future? Meta got there first, Google has followed, Apple is following and others are already lining up. We have gone from Mainframe -> Desktop -> Phone -> Glasses ... Each step moved the computer closer to us. Consider that between laptops, smartphones (and smart TVs), we spend 9-10 hours (maybe more) a day looking at screens. AR glasses may absorb both, en

Moderna merged HR and IT under a single Chief People and Digital Technology Officer as it scaled AI across the company. For most of the last century, companies treated work as something solid. Jobs sat in one box, software sat in another, budgets followed the boxes, HR managed the human layer and IT managed the machine layer. AI turns that architecture into a drag coefficient. Once intelligence becomes part of the operating layer of the company, every model upgrade changes th