(This post is about Turi, but I will occasionally refer to it by its older names Dato and Graphlab.)
Turi got acquired by Apple for $200M (or so it is rumored). Reactions on the internet go from hearty congratulations to folks saying this is another example of the “AI bubble”. And of course, there are ill-informed, get-rich-quick comments like this:
Building machine learning tools, and selling them successfully, is a very challenging business. Dato did both extremely well. I still remember going to NIPS 2010 in Vancouver (last NIPS before they changed locations), and seeing demos from Carlos Guestrin’s group on their projects “graphlab” and “graphchi”. The demo went something like this: We have a massive webgraph here on this mac mini, and I’m going to start running PageRank on it using graphlab, and before I finish explaining what it does the computation is finished. And they were pretty concise in their explanation too. It was super cool, I thought esp. since I was working on scaling random walks-based ML algorithms, and knew how hard it was to pull off a trick like that at scale. This was before the AI hype, and all, when ML researchers would themselves as ML researchers and not “AI”.
Since that demo, I’ve been a big fan of Graphlab, and followed them closely. I saw them grow from a small university project to a proper startup with actual paying clients. I also got to know Danny Bickson (co-founder of Turi) and signed up to be the “Graphlab/Dato evanglist” in SF. I have recommended their products to some of my clients, where I found it perfectly solved their need. Every one of them was happy not just with the product, but also with Dato team which is super-responsive to customer questions. Hard work. Years of toil. Building things brick-by-brick. Serving your customers. That’s how you create value. Of course, having people like Carlos and Danny goes a long way. I’m very happy for the folks at Turi (/Dato/Graphlab), and it’s a big win for Apple. Very well deserved!