The most fascinating watch today is the deep-dive interview with Xpeng’s CEO, He Xiaopeng, on his massive bet on “physical AI” and humanoid robots. He argues that applying digital AI paradigms to the physical world is a major mistake, and details why developing a general-purpose humanoid robot is up to 100 times harder than building an EV company. He Xiaopeng: Robot IRON’s Birth, The Accident, GX, His Big Bet & Swimming in Blood
Today’s absolute must-watch comes from the master chefs at Lao Fan Gu (老饭骨), who deliver a masterclass on creating incredibly juicy, restaurant-quality lamb dumplings at home. It is a highly satisfying blend of elite, precise culinary technique and heartwarming kitchen camaraderie between a master and his long-time apprentice.
In [國宴大師•羊肉餃子】一咬爆汁！只需要幾種簡單的調料，在家做出來比飯店還香～| 老飯骨 by the channel 老饭骨, Chef Ma breaks down the strict rules for a perfect lamb dumpling filling. He recommends using flank or belly meat (腰窩) for an ideal fat-to-lean ratio, insists on chopping the meat by hand rather than buying it pre-minced, and folds in plenty of oil to trap the moisture. A brilliant and critical step shown is waiting to mix the chopped scallions and sesame oil into the filling until right before wrapping, which ensures the aromatics stay fresh and don’t overpower the delicate meat.
If you only watch one video today, prioritize the massive Steam 618 sale breakdown. It details huge discounts on major titles and flags four time-sensitive free games you need to claim to your library immediately before their promotions expire.
For narrative action fans, there’s a new PS5 playthrough titled 繼續007~玩到通關! PS5. The broadcast takes viewers through the entirety of a James Bond title, going all the way to the game’s completion. It’s highly recommended if you’re looking for a cinematic, “playable 007 movie” experience to watch in the background.
why claude, codex and cursor switched primitives (github take note): The front-end engineers behind Pierre Computer Company detail how they built the wildly performant code tree and diff rendering components used by leading AI coding assistants, leveraging vanilla JS, aggressive virtualization, and the Shadow DOM to instantly render 150MB patch files without janking the browser’s main thread.
How the Electrical Grid Is Being Rebuilt for AI | Bloomberg Primer is a fascinating look at how the explosive power demands of artificial intelligence are forcing a massive, global overhaul of the electrical transmission system. It serves as an excellent primer on the geopolitical battle for grid dominance and features experimental hardware like Veir’s liquid-nitrogen-cooled superconducting cables.
CNBC’s deep dive on Why The UAE Walked Away From OPEC is a massive geopolitical story explaining the UAE’s shift toward a post-oil economy and its growing independence from Saudi Arabia following regional conflicts. In U.S. politics, WSJ covers Ken Paxton’s Trump-backed victory over John Cornyn in the Texas Senate primary, marking a major win for MAGA loyalists (Trump-Backed Paxton Defeats Cornyn: What’s Next in the Texas Senate Race?). For Chinese speakers, Jason at “美投侃新闻” offers a brilliant historical perspective on the current stock market euphoria in 我们该离场吗？, analyzing legendary trades by Stanley Druckenmiller and Bill Ackman to ask if investors should exit the market now.
How I deleted 95% of my agent skills and got better results — Nick Nisi, WorkOS This is the most practical talk in the batch, explaining how to tame LLM non-determinism by abandoning open-ended prompt instructions in favor of a strict TypeScript state machine that forces agents to cryptographically prove their work.
The single best video this week is “Reverse engineering a Viking VOIP phone protocol with Claude Code” by Boris Starkov from Eleven Labs. It provides a stunning, high-signal demonstration of an autonomous agent sniffing traffic and rewriting persistent memory to brute-force a hardware device, proving exactly how capable models have become at executing complex, multi-step engineering tasks.
This week was heavily dominated by the maturation of AI agents, moving beyond basic text chat into structured, sandboxed integrations via the Model Context Protocol (MCP) and full GUI automation. We are witnessing a fundamental shift in daily workflows, with the terminal increasingly being bypassed in favor of IDE-embedded browsers and autonomous models generating massive, risky pull requests that demand stringent human review. Underpinning this is a ruthless optimization of infrastructure, spanning from Google splitting out specialized training and inference hardware to SpaceX aggressively cutting data center build times down to 66 days.
If you want to bring the comforting, savory aroma of a Japanese diner straight into your kitchen, you must watch MASA’s guide to a tofu puff and pork belly sukiyaki set meal. It is the perfect blend of homestyle warmth and accessible cooking techniques that will instantly upgrade your weeknight dinner rotation.
In 在家重現日式定食屋的幸福感！油豆腐豬五花壽喜煮定食, the always-reliable channel MASAの料理ABC shows home cooks how to assemble a complete Japanese set meal. The focal point of the tutorial is a simmering, sukiyaki-style main course featuring rich pork belly and flavor-absorbing tofu puffs. Even with a few garbled auto-captions, MASA’s visual walkthrough of the heating and simmering process remains incredibly appetizing and easy to follow.
If you only have a minute today, the official Minecraft channel’s teaser for a surprise broadcast is absolutely worth your time. It’s short, hype-inducing, and gives you all the essential dates and times for their upcoming TwitchCon Rotterdam stream.
The Minecraft team just dropped a very brief but exciting announcement video: THERE’S A BONUS MINECRAFT LIVE?!. Clocking in at just over a minute, it confirms an extra Minecraft Live show happening on May 30 at 11:30 AM EDT / 5:30 PM CEST directly from TwitchCon Rotterdam. It promises exclusive insights and sneak peeks, making it a must-watch event that you can catch on YouTube, Twitch, TikTok, or Instagram.
Frontier AI at Home — Alex Cheema, EXO Labs Alex Cheema cuts through the AI hype to focus purely on local hardware inference, explaining the memory-bandwidth bottlenecks of auto-regressive decoding and demonstrating how to cluster Apple Silicon and RTX GPUs using Thunderbolt 5 RDMA to run 1-trillion parameter models locally.