arsalan: yeah. i think also we get enamored with simple things permit the latest one is how many parameters do we have? 5 billion, one trillion? it reminds me of the old days where everybody would argue who had this faster chip for your computer. it's not really the bottleneck, and if you look at large models that are one trillion parameters, underneath they are much smaller specialized models. so we are seeing more organizations go to the model where they are saying the smaller model, this book for my actual need, and people are starting to care about how fast can i answer questions and the cost. because those big models are generally slower and very, very expensive not just a dream, but to answer a question with. databricks, the core is how do we drive down the costs and customization with smaller models that are much more accurate that maybe some of these smaller models. annabelle: annabelle: let's talk about the business of databricks, you were last valued around the $43 billion mark. is that wh