Probably because your optionality is so low. If you invest the time to learn this and then don't get a $350k job, what's your recourse? Furthermore, the applications of KDB+ in broader industry are few, so you won't come out of finance with a valued skill. The practitioners of this obscure skill are therefore bimodal in their distribution of pay: those who get a job working with KDB+ and those who don't and wonder why they wasted their time. I think a better relatively obscure language would be a functional language like OCaml or Reason or Rust.
Contract work pays around 2x the equivalent salary, and maybe around 1.5x after you account for lack of paid holidays, sick days, employer pension contributions, and downtime between contracts (although this can vary a lot depending on your particular circumstances)
The article's answer is obvious: it's an extremely niche language and people learn it when they need to, using the proprietary system paid for by their employer. Supply meets demand perfectly, and people learn it when they are paid to.
Can confirm the gist of the article but I wouldnt say its the "hottest" coding language in finance. It is largely used in electronic trading for realtime and historical DB's that algo's interface with to determine pricing models. Also true that experienced KDB Devs and KDB DBA's are rare to come by.
There are two kdb consulting companies in Northern Ireland, FD who own Kx, and AquaQ. Working here in London, you can’t move for FD and AquaQ consultants. I wouldn’t be surprised if they saturate the kdb job market, as both companies are on a massive hiring spree.
Curious if Whitney agreed to a non-compete when he formed Shakti[1], or if Shakti is supposed to complete gracefully against Kx Systems when it's finally ready.