We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Nvidia: what investors told people to watch and read
Chamath Palihapitiya promoted the new All-In episode with just "pod!!!!!!" . The Besties take apart Jensen Huang's exit from Nvidia. Topics include the spy allegations and who leaked the memo, the market impact and the AI trade, China hawks and export controls, the effect on open source, and who should be Nvidia's next CEO. A final segment covers Anthropic's IPO road show . Episode
People were also asked to read and sign letters, not just read them:
- Amjad Masad: "Read the Stand with Jensen letter and consider adding your name" . Letter
- Garry Tan signed the same campaign: "Jensen deserves due process, not whatever is happening right now" .
- Brad Gerstner (Altimeter) shared a letter sent to Nvidia's board. He said it sets out "the minimum acceptable standards the Board owes shareholders" during the crisis . Post
Jason Lemkin shared a new 20VC x SaaStr takeaways piece, "Jensen Resigns, Sequoia Pulls Back to the Bay, and Anthropic's IPO Draws 6x Demand." He pulled out three points: Anthropic's book is 6x oversubscribed, Sequoia is winding down evergreen, and "the AI trade only cracks if demand cracks" . Article
Garry Tan: his own harness-eval repo
Tan shared his open-source g-harness-evals repo "and what it found" . A Chinese-language summary he quoted lays out the results. The test held the model and tasks fixed and swapped only the agent harness: Claude Code, Codex, OpenClaw, Hermes Agent and QM. On 20 tasks, pass rates bunched between 71% and 79%, and the confidence intervals overlapped heavily. Cost per solved task varied much more: $0.44 for OpenClaw, $0.22 for Claude Code and $0.17 for Codex .
The repo also tested Tan's own gstack skill pack. The skill text loaded every time, but the model ran the skill's steps 0 times in 46 runs, and those runs cost twice as much . Tan discloses that QM is a YC company, YC invested in Hermes, and gstack and gbrain are his personal projects . Repo
Sarah Guo: an inference engine built from scratch
Guo shared @saltyph's project with the comment: "the future is generated, continually learning inference engines in open source!" . It's an inference engine for DeepSeek V4.1-Flash written in plain Rust and CUDA, with no PyTorch or SGLang. The author reports about 1,740 tokens/sec decode on 4×B300 and a 1M-token prefill in about 12 seconds. The code is on GitHub . Post
Eric Schmidt: Genesis
At a CHM Live event marking SRI's 80th anniversary, the interviewer asked whether Genesis went out of date before it was published. Schmidt co-wrote the book with Henry Kissinger . His answer: "Henry laid out the next 20 years in a way that no one else did." He said Kissinger saw the arrival of a nonhuman intelligence as an epochal change, one that raises questions of coexistence, what it means to be human, and alignment, and that this was "all written before the common understanding" . Video
Shorter picks
- Tim Sweeney pointed to his GamesBeat interview with Dean Takahashi, which also features Epic EVP Marcus Wassmer. It covers "breakthroughs in game development efficiency" and the future of engines and tools . Interview
- Jack Dorsey pointed to @Hello_World's "Buzz Plugins Megathread," calling plugins "huge, and where the magic is" . The thread presents Buzz as a "network harness" in which each team controls its own workflows, tools and rules without waiting on a roadmap . Thread
- Paul Graham quoted Joshua Reynolds on self-directed learning: "Few have been taught to any purpose who have not been their own teachers" .
An All-In speaker praised AlphaSense’s summary of Anthropic’s IPO as “excellent.” A speaker on the show called Will Man’s tweet “very insightful”; it relayed whispers from “safe” group chats and people in Washington that “the old king’s murder” was complete and they were days from the throne. The discussion placed the tweet alongside Anthropic’s planned IPO and Nvidia losing its founder and CEO.
Eric Schmidt endorsed Genesis, his book with Henry Kissinger, for its account of the next 20 years, which he said no one else had laid out as well . As he described it, the book argues that a new nonhuman intelligence marks an epochal shift and raises questions of coexistence, human identity, and alignment .
Balaji recommended an analysis by @gemchange_ltd as a starting point for identifying which Nvidia board members opposed Jensen Huang and pressing them to resign, as part of his effort to secure Huang’s reinstatement.
Elon Musk told followers to “Try Grok Bot,” linking to @AnatoliKopadze’s post. The post promotes a prompt for turning what you read, watch, and think into a self-organizing knowledge base, and points to a full guide.
Paul Graham shared Joshua Reynolds’s observation that instruction forced on learners is received “lazily” and “ineffectually,” while people who learn purposefully often teach themselves—highlighting the value of self-directed learning. Post
Jack linked to and endorsed @Hello_World’s “Buzz Plugins Megathread”, calling Buzz plugins “huge” and “where the magic is.” The thread describes plugins as Buzz’s most significant new update, giving teams control over their workflows, tools, rules, and integrations while connecting their setups across the network.
Jason Lemkin shared a new 20VC x SaaStr article, “Jensen Resigns, Sequoia Pulls Back to the Bay, and Anthropic's IPO Draws 6x Demand: The Top Takeaways from 20VC x SaaStr,” at SaaStr. He highlighted that Anthropic's book is 6x oversubscribed, Sequoia is winding down evergreen, and “The AI trade only cracks if demand cracks.”
Sarah Guo (@saranormous) shared @saltyph’s open-source inference engine for DeepSeek V4.1-Flash, saying the future is “generated, continually learning inference engines” in open source . The linked post says the engine was built from scratch in Rust and CUDA, without PyTorch or SGLang, and is available on GitHub .
Elon Musk shared @AJamesMcCarthy’s video post of his drive to work, which McCarthy described as looking like a scene from a sci-fi film; Musk captioned the share “Made on Earth by humans.”
Garry Tan shared his open-source g-harness-evals repository, describing it as his harness-evaluation repo and linking it at GitHub. The benchmark it presents found that coding-agent setups had similar task pass rates (71–79%), while cost per successfully solved task ranged from $0.17 for Codex to $0.44 for OpenClaw.
Aravind Srinivas highlighted Jeff Grimes’s post and video showing a photorealistic Pantheon recreation, emphasizing that rendering was distributed across 48 sandboxes and completed in three hours.
Chamath enthusiastically recommended an @theallinpod episode, posting “pod!!!!!!” with a link to it. The episode post says “The Besties” break down the claim that Jensen is out as Nvidia CEO, covering spy allegations and a memo leak, market and AI-trade impact, China and export controls, open source, and possible successors; it also lists Anthropic’s IPO road show.
Brian Armstrong promoted a children’s book made by his team (title not specified) as a way for parents to help children get a better start with money; he said each copy includes $50 in bitcoin to offset the book’s cost and help start a child’s investment portfolio. Coinbase described it as its first children’s book, using a story to help parents explain that money is changing. Coinbase’s announcement
Jason Lemkin shared a new The Agents video with @miadia (YouTube link); its teaser highlights that they canceled Canva because their agents never used it, renewal rates for $100K+ sponsors rose from the 20s to the 70s, and an agent co-sold a six-figure deal with Wix’s Base44.
Elon Musk replied “True” to Sourcery’s linked post, which includes a video and statements attributed to Max Levchin criticizing socialism and wealth redistribution . This signals agreement with the post’s message, but Musk gave no further reason or explicit recommendation to watch the video .
Garry Tan recommends his idea-minder skill, which he says lets a personal agent recursively improve as it is used: https://gbrain.io/s/idea-miner-what-to-build-rrhjgv
Elon Musk pointed followers to a broadcast for Starship Flight 15 by sharing a SpaceX post labeled “Watch Starship Flight 15”; he said the flight was about four hours away. Watch the broadcast.
Jason Lemkin invited people to a live conversation about AI agents for GTM with Granola’s Head of Sales, encouraging attendees to bring tough questions; the event signup page is here.
Jason Lemkin directs readers to a free guide to AI Sales Agents; the guide’s creator and a reason for recommending it are not stated. He also links to more information about “our AI agents” at SaaStr.ai/agents.
Tim Sweeney shared Dean Takahashi’s GamesBeat exclusive interview featuring Sweeney and Epic EVP Marcus Wassmer, highlighting discussion of game-development efficiency and the future of engines and tools. The linked post frames the interview around Sweeney’s prediction of an AI revolution in gaming, with 10x productivity and entirely new genres if developers get behind it.
