Big day in the AI world. Here are some important highlights for you all: Satya Nadella's full discussion at the All-In Summit was released on Youtube today: > Nadella advocates for a multi-model world, emphasizing that interoperability and open-source checks are essential for a healthy ecosystem. > Microsoft's capex is designed to serve third parties, not just 1 or 2 customers (such as Anthropic and OpenAI) and calls it silly if any hyperscalers are investing just to serve 2 customers. > Copilot is strong with 30+ million users. > Nadella supports the idea of a third-party tester for AI models. > 60% of cost is racks and chips. Company is renting quite a bit of chips due to shortage of supply. Goal is to build more and lease some. > Goal is to have diverse offering of chips. > Nadella hopes to see GDP growth from AI - need to see at least 7-8% GDP growth that is broad based. Dario and Jensen spoke at Dreamforce 2026: > Dario used an analogy of a car company facing a safety incident to explain his approach to AI safety. He emphasized a three-step plan: improving one's own records, setting shared industry standards, and incorporating an international component > Jensen says safety is important and needed. > Jensen notes massive shift in AI - closed models are doing well. Number of tokens have increased by 25x, with a lot of it in the last 6 months. Open models went from 30% in the beginning of the year to 70% in market share. Sachin Katti of OpenAI spoke at AI Infra Summit: > The big statement: "The way we will make sure frontier models are safe and aligned is by spending compute. So the counterintuitive point is that we’ll need even more compute to make sure future models are more safe and aligned" - more compute is needed. Contradicts the idea of an investment slowdown as said by many over the weekend. > Nvidia's Rubin racks gave a 3x performance improvement over Blackwell just right out of the box, then ran Astra as agent to optimize its own inference on Rubin chips and over 72 hours, the model produced an additional 2x throughput improvement. Mark Zuckerberg posted an article on how AI safety can be done: > Mark makes an important statement; "Meta delayed shipping Muse for several months to focus on safety and security. We didn't call for everyone else to do this before we would. We just did it as part of our day-to-day work because it was clearly the right thing for people and for us. I'm proud of the security foundations we've built." - Mark is suggesting that Anthropic and OpenAI can initiate their own safety protocols without having to tell the world they would go extinct by end of the decade. Just do your part. > While Meta plans to offer free versions of its AI tools to billions of people, he notes that for users who want to use more compute, there will be a dynamic auction mechanism. This auction mechanism is intended to guarantee that users get the lowest possible price for the intelligence and compute they use, while ensuring that the overall capacity is dedicated to what people collectively find most valuable. > To keep humanity in control, the vast majority of AI compute and intelligence must be purposefully directed by individuals toward achieving their personal goals and aspirations. This requires a lot more compute for Meta, labs, and hyperscalers. "This means that Meta, other frontier labs, and clouds committed to giving people a greater balance of power must collectively build out a sufficiently large amount of compute such that we can allocate enough to recursive self-improvement to remain competitive while still committing the significant majority towards people's individual goals."