Michael Margerum
1.7K posts







🚨 BREAKING: 84 TanStack npm packages were compromised in an ongoing Mini Shai-Hulud supply chain attack, adding suspected CI credential-stealing malware. Socket flagged every malicious version within six minutes of publication. This is a developing story.





A former Goldman Sachs executive just said something on camera that should terrify every lawyer, doctor, and analyst on the planet. His name is Raoul Pal and he used to move billions on Wall Street. He was asked one question: "How disruptive will AI be?" He said it is the single greatest innovation in human history. Greater than the internet or the electricity. The only thing he compared it to was the splitting of the atom. But here is the part nobody is ready for. He said knowledge is now worth zero. Think about that for a second. Why do lawyers charge $800 an hour? Scarcity of knowledge. Why do consultants bill Fortune 500 companies millions? Scarcity of knowledge. Why did your parents tell you to get a degree? Scarcity of knowledge. AI just destroyed that entire model. A teenager with ChatGPT can now draft legal contracts, build financial models, write code and analyze medical scans. No degree, decade of experience and no six figure student debt. And the numbers already prove it. Employment for workers under 25 in AI-exposed jobs has dropped 13%. Wall Street banks are planning to cut 200,000 jobs. 30% of U.S. companies have already replaced workers with AI. 18 million entry level American jobs could disappear entirely. But here is the part that should keep you up at night. The junior roles where people learn, make mistakes, and develop real judgment are being automated first. Which means the experience that AI cannot replace is the exact experience young workers can no longer get. Pal says humanity now faces a binary choice. Merge with the machines or reject them. There is no middle ground and this is not about robots in factories. This is about the collapse of the entire knowledge economy.



Pretty incredible that this is running 100% locally on 2 x 512GB M3 Ultra Mac Studios connected with a Thunderbolt 5 cable consuming ~400W. This is possible now because of a perfect storm of: - Models: Really good Chinese open models. - Hardware: Apple Silicon with unified memory happens to be perfect for sparse MoE LLMs. - Software: Low-latency RDMA over Thunderbolt 5 on macOS 26.2 reduces latency by 100x to single digit microseconds. This enables @exolabs to scale up with tensor parallelism. Today this requires $20k of hardware for frontier AI but the cost is being driven down on all fronts: models, hardware and software. M5 Ultra is expected soonTM, and should have ~50% more memory bandwidth than M3 Ultra and >4x FLOPS with tensor cores (Apple calls these Neural Accelerators).



















