Why did @JeffDean leave Google after 27 years? Here's the answer in his own words! “Sometimes being a focused small company, with everyone focused on exactly that mission… can be amazing,” he said this at the AASF 2026 conference at Stanford. Jeff had watched Google grow from 25 people to more than 190,000. Now, with his new venture @DiscoLoopAI, he is returning to something much smaller: a focused group of people united around one ambitious mission—to use AI to accelerate scientific and engineering discovery. Perhaps he is just searching for the feeling of early Google again: a small team, a huge vision, and brilliant people all pulling in the same direction. Large companies have resources, but small teams have dreams and focus. Best of luck, Jeff! You'll always have our support! 😊

Why did this interview by a Chinese podcaster and Nolan resonate so widely? Because Zhong Shu did not interview Christopher Nolan as a celebrity promoting a movie - she met him at the level of ideas. Drawing on her training in classical languages and political philosophy, she began with the centuries separating Homer from the Trojan War, then moved through the Bronze Age collapse, xenia—the ancient Greek duty of hospitality (the Zeus' laws as appeared in Nolan's Odyssey), to Christian redemption, hubris, and our modern discomfort with greatness. But knowledge alone was not what made the interview special. She listened. Instead of mechanically moving to the next prepared question, she caught the concepts inside Nolan’s answers and pushed them further. That allowed Nolan to leave the promotional script and explain what The Odyssey means to him: how civilization is held together, whether repentance can repair the damage of war, why today’s heroes must answer for their greatness, and how filmmakers create something new within traditions audiences already understand. A great interview does not simply extract information. It creates the conditions for someone to think aloud—and sometimes discover what they think in real time. Zhong Shu @ZhugeYY did exactly that and that's so inspiring to me. Hats off indeed! 😇

A Kentucky electrician may be able to help @elonmusk take AI into space - that's the beauty of the AI era. A few months ago, I had the privilege of telling Jason Walls’ story. A master electrician and fiber splicer, @walls_jason1 taught himself to code with Claude and built ChargeRight—a tool that helps EV owners avoid unnecessary, expensive electrical-panel upgrades. He wasn’t a Silicon Valley founder. He simply combined skilled-trade expertise, firsthand knowledge of a real problem, and AI. My post caught @mcuban Mark Cuban’s attention. Mark reached out to Jason, encouraged him—and an invitation to apply to Shark Tank followed. Now, Elon Musk says SpaceX is hiring exceptional engineers and skilled-trades talent to build and operate “the most powerful AI supercomputer clusters on and off Earth.” i thought of Jason and told him about it - and not surprisingly, he immediately applied. His story captures a much bigger shift: the AI economy won’t be built by software engineers and model researchers alone. It will also depend on electricians, fiber technicians, HVAC specialists, construction workers and many others in the so-called blue-collar professions. As AI expands into the physical world, people who know how the physical world actually works are becoming more valuable—not less. Jason, perhaps you really were made for this. I’ll be rooting for you again. 🚀 Here’s the story that started it all:

Can you outsource taste? This is such a thought-provoking question by @reidhoffman. Artist Joe Salvatore said it well: “I can outsource production, but I can't outsource my taste."

Great read on the Chinese venture ecosystem by @loubohan! The short message is - China's VC scene is not just a less mature version of Silicon Valley - it operates under a different logic: weaker M&A exits, stronger IPO pressure, government-backed RMB capital with industrial-policy mandates, widespread use of FAs, and a relationship-driven social graph. That structure can be brutal for founders, but it also helps explain why Chinese companies become exceptionally lean, fast and globally aggressive.

Reuters reports that the Trump administration is preparing to block imports of new Chinese humanoid robots, robot dogs and certain power inverters. There is also a claim circulating on X that Chinese AI models will be banned today. But so far, I have not seen that confirmed in the Reuters report or by an official announcement. So I would treat that part as speculation for now. I get the security concern. A robot is no longer just a machine that moves. It has cameras, microphones, sensors and internet access. It could be operating inside a factory, laboratory, warehouse or even someone’s home. Power inverters are even less glamorous, but possibly more sensitive. They connect solar panels, batteries and other equipment to the electricity grid. If they can be remotely accessed or controlled, they become part of the national-security conversation. But here is the question I keep coming back to: How does America plan to build a competitive robotics industry without relying on China’s robotics supply chain? China installed 295,000 industrial robots in 2024—more than half of all the industrial robots installed worldwide that year. Four of the five largest humanoid-robot companies by 2025 installations were Chinese. China also has major advantages in many of the components robots need: permanent magnets, motors, batteries, bearings, power electronics and increasingly actuators and sensors. This does not mean every robot component is made in China. Japan, Germany, South Korea and other countries remain important suppliers. But China has something that is very difficult to reproduce: the factories, suppliers, engineers, component makers and customers are all close to one another. A robot company can design something, find parts, build a prototype, test it and move toward production much faster—and usually much more cheaply. So banning Chinese robots may solve one security problem, but it raises several others. Will American robotics companies still be allowed to buy Chinese components? Will the government invest enough to build alternative supply chains? Will these restrictions help American companies—or simply make robots more expensive and slow down adoption? And if the U.S. restricts Chinese robots while China continues deploying them across factories, warehouses and public services, who will collect more real-world data and improve faster? The U.S. may still have an advantage in the AI “brain.” China has a serious advantage in building the robotic “body.” You can ban a finished product quite quickly. Rebuilding the industrial ecosystem behind it is much harder.

One Month Later, We Are Still Fighting for Justice for Wang Wang Today marks one month since Wang Wang, a gentle stray mother dog in Jieyang, Guangdong Province, China and her three newborn puppies were brutally tortured and killed. According to information later circulated online, five minors were involved, including one who filmed what happened; the official notice identified four directly involved boys, all under the age of 14. The videos and reports show that this was not a momentary outburst, and it cannot be dismissed as “children not knowing better.” It was prolonged, escalating cruelty. Wang Wang's three puppies—barely old enough to open their eyes—were beaten to death. Wang Wang was restrained by the neck, repeatedly beaten, doused with a flammable liquid and set on fire, with eyes gouged out. She was so sweet that she did not even try to attack them. Even in her final moments, she reportedly struggled back toward her babies and used the last of her strength to lick their bleeding heads. Meanwhile, the perpetrators filmed, laughed and cheered. This was not ordinary childish misbehavior. It was the deliberate creation, observation and celebration of suffering. It challenged the most basic limits of human conscience. What makes this even more unbearable is not only the cruelty itself, but the absence of clear, transparent and credible accountability. Local authorities stated that the four directly involved boys had been sent to a specialized school for education. Yet the public has received little meaningful information about what intervention, psychological assessment or continued supervision actually followed. People familiar with the case said they were not sent to any "special school", and the authorities had never criticized such horrid behaviors. The parents and family of these perpetrators even threatened to kill more dogs and cats in response to online fury. And at least one father of the killers did kill more animals and brag about it. China still lacks a nationwide anti-cruelty law specifically protecting companion animals such as dogs and cats. Existing laws cover wildlife, livestock and laboratory animals in different ways, but significant gaps remain when someone deliberately tortures an ordinary companion or stray animal. This is not a debate about vegetarianism, and it is not about telling people what they may or may not eat. It is about whether a society is willing to draw a basic legal line against deliberately torturing a sentient being for pleasure, entertainment, attention or online content. Justice means establishing the truth, requiring serious psychological evaluation and long-term intervention, holding guardians accountable, preventing perpetrators from owning or accessing more animals, and creating laws with meaningful consequences and deterrence. Because when children can torture defenseless lives while laughing—and the adults and institutions around them remain silent—the damage extends beyond the animals. It reaches the moral foundation of society itself. Wang Wang never had a real home. She reportedly sheltered her puppies in a small hole, trying to keep them safe. She may never have known comfort or had a truly good meal. But this Earth was her home too. She and her babies harmed no one. They were simply trying to survive. Yet even that most basic right was denied to them. Today, we remember Wang Wang and her three puppies. We will continue to say their names. We will not allow them to become another horrific video that briefly circulates online and is then forgotten. We must turn our grief into strength: continue speaking, advocate for China’s long-overdue animal anti-cruelty legislation, and work toward a society in which more animals like Wang Wang can live peacefully—and those who choose to harm defenseless lives know that there will be consequences. Cruelty is not entertainment. Abuse is not freedom. Deliberately inflicting suffering must have consequences. Justice for Wang Wang and her family.

Such a beautiful day at Silicon Valley with beautiful @missfearleslie! Leslie has made 2 exits from her gaming startups and now studying at Stanford and making waves in the valley. 🤗😇

Thank you @BBCWorld for giving a voice to China's animal lovers and people who care about animals worldwide. Wangwang and her puppies deserve justice, and millions of more animals in China deserve the rights to live in peace.

Google’s Q2 earnings (beating Wall Street expectations) are a perfect snapshot of the AI trade right now: Revenue: $119.8B, +24% YoY Google Cloud: ~$24.8B, +82% YoY It shows AI demand is real, but so are investors' concerns - $GOOG stock pulled back a little after the earnings. The issue isn’t whether Google is “winning or losing AI" anymore, but how expensive winning has become. Alphabet raised 2026 capex guidance to ~$195B–$205B, and free cash flow reportedly turned negative in Q2. That’s why the stock reaction was weak despite strong headline numbers. My takeaway: AI is boosting growth, but the market now wants proof that AI capex converts into durable margins — not just bigger data centers. For AI investors, this is the new scoreboard: Cloud growth, Capex intensity, Free cash flow, Margin durability, Model/product adoption. AI demand is no longer the debate. AI return on investment is. Not financial advice - just discussion. 😃 What's your take on @Google ?

Google just launched three Gemini models: 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. The immediate reaction from parts of AI Twitter has been predictable: Where is Gemini 3.5 Pro? Google originally indicated that its flagship model would arrive in June, but it remains in partner testing, reportedly after falling short on key coding benchmarks. That criticism is fair. But it may also miss what Google is actually optimizing for. These are not frontier-model launches. They are deployment models: — 3.6 Flash targets coding, knowledge work, computer use and agentic workflows. Google reports 49% on DeepSWE, 83% on OSWorld-Verified and higher knowledge-work scores than both 3.5 Flash and 3.1 Pro—while using substantially fewer output tokens on some tasks. — 3.5 Flash-Lite is the high-volume, low-latency option, priced at $0.30 per million input tokens and $2.50 per million output tokens. — 3.5 Flash Cyber is a specialized security model for finding and repairing vulnerabilities. In Google’s testing on the V8 JavaScript engine, it found 55 confirmed issues, versus 47 for standard 3.5 Flash and 36 for Claude Opus 4.6. Google also highlighted testimonials from Figma, Harvey, JetBrains and Hebbia, citing faster prototyping, stronger legal-document work, improved coding performance and better evidence retrieval. These are Google-selected customer accounts, not independent benchmarks—but they point toward the real contest: quality per dollar, latency and reliability inside production workflows. So yes, Google still needs to prove it can deliver the delayed Pro model. But Flash should not be dismissed simply because it is not the smartest model on the leaderboard. For most enterprise AI applications, the winner may not be the model that reasons best once. It may be the one that performs well enough—millions of times, quickly and economically.

The strongest coalitions are not just built on identical identities, but built when one community recognizes another’s pain—and shows up. In 1950s San Francisco, Chinese immigrant Sinclair Louie was blocked from buying a home in Sea Cliff. A Jewish lawyer, Ben Lehr, understood that exclusion and helped him fight it. Their story became a symbol of the Chinese–Jewish solidarity that helped build America’s Civil Rights coalition. But that coalition fractured when success caused Asian and Jewish Americans to be recast as “privileged”—too successful to deserve protection, even when facing violence. Asian Americans experienced that abandonment during the wave of attacks from 2020–2023. Jewish Americans are experiencing it now amid rising antisemitic violence. In both cases, institutions too often protect a political narrative before they protect actual people. The lesson is not that solidarity was a mistake. It’s that solidarity cannot be conditional on whose suffering is politically convenient. A must-read beautiful and powerful piece by Forrest Liu @notGarryTan : The Merchant and the Lawyer (link 👇) on @garryslist with @garrytan.