You’re argument here almost makes sense to me. Where I struggle to agree is in how you perceive it. It’s not that they don’t know how to leverage it, Google for example is by definition a machine learning company that obtains revenue through ads. In fact AI and machine learning have provided incredible value to many businesses already, and not just the Facebooks, Google’s and IBM’s of the world. The reason they kept MapReduce secret for so long, is because surprisingly enough, just about anyone with a computer and some programming knowledge can replicate it. AI isn’t the same. Not only can the outcomes be counter intuitive at times, but most of the research that has made Machine Learning popular has emerged within the last decade. The algorithms aren’t new, but the uses are. The point of open sourcing your tools is that fact that they not only need to be battle tested, but Machine Learning experts are incredibly rare. If you want to be the king of AI, you must convince researchers and hobbyists alike to use the tools your company produces. This is the same reason why companies like Nvidia donate massive clusters of GPU’s to AI competitions. We are still learning what they can do, and keeping that learning to a very small echo chamber of experts within your walls is actually counter productive to progress. The capability of A.I. is growing at a breakneck pace that is almost unprecedented. Machine Learning(aka Big Data) has always been AI, that hasn’t changed, only people’s realization of what is actually going on in companies that leverage Big Data has.