"Other" is showing that this new market is still being fully explored for opportunism reasons and not stuck on a monoculture model.

Think of it like Netscape Navigator and Internet Explorer being the two competing browsers for the first ten years of the Internet for most people, "Other" looks to be the Firefox, Chrome and Opera alternatives - but in the AI/LLM domain - which will take the lion's share of the market once it settles down and those who can compete succeed whilst most of them will falter. By settles down I mean into a regular churning of information until another paradigm shift happens.

With models now being able to be easily run on in-house systems, rather than be outsourced to servers which not only prioritise workloads, but have access to propriety data, it would seem that we're seeing that AI/LLM's are being seen as useful but are also being kept away from being pimped out as often. Anonymised drudge goes to outsourced servers (The cheaper the better) with sensitive stuff either being done in-house or being used with the more specialised, and expensive, servers used at the moment.

The Substack in question for those who want to view the findings in a little more detail.