I've been dabbling with OVH and it feels very pricey and fragile. Has a very lipstick on a pig approach to whatever they used to be doing before piling into cloud.
I ran a test yesterday on ChatGPT and co-pilot asking first if it knew of a specific paper which it confirmed and then to derive simple results from which it was completely incapable of. I know this paper is not widely referenced (ie few known results in the public domain) but has been available for over 15 years with publicly accessible code written by humans. The training set was so sparse it had no ability to "understand" or even regurgitate past the summary text which it listed almost verbatim.
I entered into WASM assuming it was simple and minimal but there are a lot of runtime edge cases that are quite surprising. I built an interpreter that passes all the non-SIMD test cases from the official spec. Would be interesting to clean-up the spec to be a real minimal execution only core. There is a real barrier to entry for producing new lightweight implementations.