Building off [[Focus and No Change]] When building something, we start with a hypothesis underpinned by critical assumptions + conditions to test and validate. The [[probability]] of success for these hypotheses depends on the customer & market winds blowing in the assumed favourable direction. If the things you need to change don't, then that could kill the hypothesis. However, when the hypothesis is disruptive, we have to bet on the change and influence it. We are obsessed with what's the trend. What's new and changing. We forget that things that last can also be built on a deep understanding of **what will not change.** On that front, there is a better question to ask, that's not clouded by the latest fad/shiny trend > what's not changing in the next decade? Betting on things that are "stable in time" and "evergreen" in nature leads to focus and timeless quality. ### **"Here to stay"** covers what's not changing. - **Performance** - Irrespective of industry, customers will still want the "best performance" for their applications, mostly operationally linked and of a mission critical nature - max value, min costs. Performance might lose out to superior UX, base functionality and regulation restrictions. - **Applications** - Customer will still want deep tech to be "applied" for it to be useful. They will want the lowest switching costs for upgrades and the easiest interfaces for accessibility. It needs to work for all user types for it to win here. Performance Return will be balanced w/ Operational Risk. - **Models** - Customers will still want the best formulations that effectively use the available data and enable decision making through unleashing algorithms. Specialised domain knowledge will win here, given the commoditisation of general models and expertise. - **Data** - Customers will still want their data to be of high quality and accessibility. With data being the moat for multiple [[Ventures]] powered by #foundationalmodels - **Tools** - Customers will still want the tools that empower them more freedom to create, have control over the customised IP they develop for their use cases and to automate. Main question being how do we increase our productivity per unit and our leverage. - **Compute** - Customers will still want the best compute (irrespective of quantum / classical) at the cheapest price. Customers will have the willingness to pay more for better latency, reserved capacity, together with application specific low latency infrastructures. ----