I thought ML needs huge datasets? This is great for GitHub's Copilot as there is plenty of high quality, production, open source code they've used without authorisation from the creators (and possibly proprietary code).
But this dataset just doesn't exist for hardware development on the internet. Almost everything on the net is hobbyist stuff. Great for hand building 10 on a desk but absolute rubbish for a (even horribly) manufacturable and COMPLIANT product.
Further: an anecdotal, but to me very telling, one liner from the video "the component is "not use for new designs" but we can ignore that because there's 260,000 of them". I have personally sniped larger quantities than that from digikey. "Not for new design" means you don't use it in a new design. PERIOD. And ironically, for something like a Murata cap, they've probably literally changed one letter in the PN because they've made a tiny process or recepie change and its more a "move everyone to the new iteration" process.
I do sincerely wish them success and maybe they can carved out a small bit of the hobbyist market but that's probably it.