Honestly I don't think the world he describes will obviously exist in the near term, and he is probably mostly eager or even desperate for his major investments into AI infra to pay out. He is probably especially nervous about his competitive edge in AI after the recent Google launches that take real shots at OpenAI's presumed moats. I hope AI will continue to get more useful, but for it to be precise and reliable enough to replace CRUD apps that reliably store and update thousands, millions or in edge cases even billions of records/fields/data points (he is probably nudging at Salesforce and Oracle's institutional investors here since they are competing with Microsoft in CRM and enterprise infra) will probably require at a bare minimum a new model, or even a new system of models that don't exist yet, and a few years to hone and scale it. The idea of SMEs and end users using excel with python and copilot as a database UI with interactive tools to manipulate and visualise the data is kind of a platonic ideal of Excel as a product. Having tried to use it that way, I can confidently say we're not quite there yet even for power users who are mediocre developers (like me), let alone the average SME manager who will ideally be the ones helping shape these tools for their businesses. It will probably take more than a couple of years to incorporate all of the popular functionality of excel in a way that is performant at scale and intuitive for the majority of relevant users. His vision is pretty cool, and I respect his optimism, but I have a hard time seeing a world where B2B SaaS is obsolete in less than 5 years. I think B2B SaaS will probably be able to offer useful functionality that Excel can't for at least that amount of time, and probably cost competition (assuming they can offer the same or comparable features) for a while. How long that extends the horizon probably depends heavily on the product, the target market, how chip tech evolves, and execution.