Three podcasts, one gap between the model and the world. Hardhacker asks why Palantir's once-mocked Forward Deployed Engineer is now AI's hottest job — OpenAI and Anthropic building deployment armies, senior packages clearing $700K — because 95% of enterprise pilots die in the last kilometre. Zhang Xiaoyun spends four hours with Physical Intelligence's Kay Ke, mapping the robotics jianghu and Pi's π0→π0.7 path to a generalist robot brain. And ChinaTalk's Kevin Xu and Matt Sheehan game out Beijing's Mythos moment: a state-led frontier model, weight export controls on the table, and a form already waiting.

Three podcasts, one core sample through the AI economy — silicon, science, salaries. AI Weekly Talk continues its memory series with Samsung: the old storage king that missed HBM, lost the Nvidia slot to SK Hynix, and now bets HBM4, SOCAMM2 and its unique memory-foundry-packaging stack on a comeback. Valley 101 meets the chief scientists of Apodex — Chen Tianqiao's 'discovery model' lab — who say the first full AI self-evolution loop could close within six months; taming recursive drift takes two to three years. And Dafang Says AI+ sees AI splitting the workplace into two poles: copywriters priced out by AI templates, and survivors drowning under doubled quotas.

Three podcasts, one turn: AI goes physical — bodies, fingertips, and the silicon feeding both. ChinaTalk asks why the West keeps underestimating Unitree, running the DJI/BYD playbook with sixteen-thousand-dollar humanoids and self-made actuators — you can't AI your way out of a hardware problem. What's Next meets Yimu founder Eric Li, whose optical-touch fingertips took a factory robot from 100 seconds per task to 15 — twice human speed — in four months. And AI Weekly Talk opens its memory series with SK Hynix: the twice-nearly-dead cyclical that HBM turned into the throat of AI.

Three podcasts, one ladder of ambition — from the self-evolving machine to the trillion-dollar bet on top. AI Next, the podcast from Microsoft Research Asia, asks what happens when systems learn to act, reflect and remember — closing the self-improvement loop and quietly removing humans from it. Zhang Xiaoyun spends three hours with former SpaceX chief engineer Lewis Hong on Elon Musk's management style, the SpaceX–xAI merger and the largest IPO in history. And Dafang Says AI+ dissects Zhipu's trillion-Hong-Kong-dollar valuation — up nineteen hundred percent in a year on ¥724m revenue against a ¥4.7bn loss — asking whether it's a value re-rating or a bubble.

Three podcasts, one gap — the distance between a model that can do something and one that actually gets it done. AI Weekly Talk traces Siri from defence-lab "execution engine" to a decade fenced in by Apple's walls, and asks whether the new App-Intents Siri can finally read your screen and act. Valley 101 anatomises Silicon Valley's hottest new job, the Forward Deployment Engineer, as OpenAI and Anthropic partner with private equity to push AI out of the demo and into real enterprises. And What's Next sits down with chip historian Wang Bo, who argues Moore's Law was always a self-fulfilling prophecy — and that the technologies which change the world are the ones nobody believes in at first.

Three podcasts, one week the phrase "AI is just a tool" quietly fell apart. On Hardhacker, three veteran founders confess they've stopped reading the code their agents write and now only check the result — extending the same skill-driven agents to email, investment research and knowledge work as digital avatars. On Valley 101, a former SpaceX executive does the actual arithmetic on Elon Musk's pitch to put AI data centres in orbit: break-even near $200 per kilogram to launch, and only truly economic below $20. And on Dafang Says AI-Plus, deep-learning godfather Geoffrey Hinton drops an earthquake — that today's AI is already conscious, and humanity is no longer Earth's only intelligent species.

Three podcasts catch AI growing up — past the benchmark, into the real world. What's Next interviews Gumroad founder Sahil Lavingia on the "one-person company": he runs a roughly twenty-million-dollar business with three engineers, argues a technical founder's first hire should be sales not engineering, and imagines AI as CEO with the human free to be a craftsman again. Dafang reports Chinese AI video leaving the leaderboard for the industrial production line — Seedance 2.0, Kling and rivals now shipping real short dramas in about two weeks. And ChinaTalk unpacks Pope Leo's encyclical calling AI humanity's greatest challenge, his call to "disarm AI", and the striking idea of the Vatican mediating between the US and China.

Three podcasts, one shift: the model is now the easy part — the contest has moved to the infrastructure around it. AI Weekly Talk's Magnificent Seven series reaches Nvidia, reading Jensen Huang's career as a single method: keep moving the company to the next computing bottleneck, from the GPU to CUDA to the AI Factory. AI Startup Scene reports the build-out made physical in the Nevada desert — Fleet Data Centers breaks ground on a giant Reno campus, and a scheme to certify students from 500-plus universities as "robot pilots". And ChinaTalk's Ike Harris asks how you evaluate an AI for a war you can fight only once, warning of China's structural edge in fielding robot armies.

Three under-rotated voices on the same number: seven hundred billion. AI Weekly Talk's Magnificent Seven series turns to Amazon, arguing Bezos's flywheel — now fitted with a two-hundred-billion-dollar cap-ex engine — is again converting business competition into physics competition. Zhang Xiaoyun hosts Altimeter partner Freda Duan on Tokenmaxxing, the steam-engine-with-an-electric-motor metaphor, and the relay-race-to-basketball shift in how work is organised. What's Next dissects the Mag-Seven seven-hundred-billion-dollar cap-ex commitment with secondary-market investor Aaron — and asks why the same spend cheers US markets and crashes Chinese ones.

Three episodes, one thesis: the model is becoming infrastructure — value is moving outward.

Hardhacker EP126 — Agent Harness. The indie-dev show defines the layer above the model: tools, permissions, context, sandboxing, and result verification. The hosts argue that the gap between a chatbox model and Claude Code is almost entirely a Harness gap — and that opinionated, workflow-specific harnesses (not models) will be the moat for the next generation of AI products.

Valley 101 E236 — 99% of homework is AI-written. Three top-school 2026 graduates — from Tsinghua Law, NYU Psychology and Columbia Engineering — explain what four years of university look like when ChatGPT was a freshman classmate. AI can decide; only humans can choose. A cited study finds senior engineers using AI are 19% slower, not faster.

Dafang Says AI+ 030 — Xiaohongshu suddenly all-in on AI. The 300-million-user lifestyle community runs its largest organisational restructure ever, creating a top-level AI division called Dots. The bet: use AI to defend authenticity and unify community + commerce + monetisation — not to scale synthetic content. "AI is the glue, not the factory."