I’ve spoken to many $1B+ revenue data companies in the past weeks, and these are some thoughts on where this is all headed and various types of players that will emerge: 1) The facilitators. Think companies that offer data anonymisation technology, ‘stock x for data’ to build a universal quality standard for data evaluation so labs don’t need to waste time with low quality data and deals can happen quicker, escrow tech for buyers and sellers to put cash and data into and then exchange in order to not get scammed, banks for financing data vendors, etc. I actually think this is quite overlooked and there’s a lot really cool stuff that will emerge. 2) The crypto-style data broker. One man operation. Operating out of a tax heaven. Great 6 month opportunity to make cash. Not all to fulfilling, but demand is there. My two cents on this if you’re considering this: act as an ‘advisor’ to the sellers. Build relationships. Almost like an investment banker on the sell-side. Build a website that depicts trust and advisory, not large scale marketplace that doesn’t care. 3) Large market-place websites. I think there will be a couple that do very well and will make a small % margin on a large volume of deals. This is less of a venture capital play and will likely flourish better if you bootstrap, seen as you don’t need outside capital. Classic problem of getting this off the ground is the chicken and egg problem to get enough sellers and buyers in the first place. Here I would optimise for going viral on social media to achieve that first unlock. 4) Couple deeply routed players for each type of data that completely verticalises (own deep supply roots; maybe even buy companies) and be in constant communication with the buyers (labs, enterprises) to understand their needs. Here you go the next step and don’t just sell the data raw, but utilise reinforcement learning to increase margin by 10-100x. For this you need some good technical talent and capital to get it off the ground, but comes with the opportunity to build something long-lasting.