Nope. Enterprise you pay for seat to access all of the enterprise features and then you just pay for tokens as you go. Vast majority of their actual revenue comes from enterprise and their revenue is just api pass through to the model providers.
How is it irrelevant? Operating a single node would be $10k+ per year. That makes no sense for a large corporation to adopt. They’re better off rewriting the entirety of Akka.
16 core machine is 32k per year. Assuming you have a hundred machines (which really isn't that many) that is 320k per year. Which pays for a single engineer to get you off Akka.
$2k per vCPU is 16k for a pretty standard 8 core ec2. The ec2 itself might cost ~$300 a month or $3600 a year. Where the hell did they come up with this pricing model lol.
This comment is really buried in the discussion. $2000 per vCPU for a library is absolutely insane pricing, unless you're running Akka on a single bare metal server, this is a non-starter for anyone.
Really unsure what Lightbend is trying to do here, this seems like the death of Akka.
There is like nothing where that would be cheap. Even if you looked for something that has very high IPC and clock rate.... the CPU with that will have at least 4 cores if not more.
The commercial license is $2k per vCPU LMAO, aka an 8 core node would cost $16k annually to operate. This is about to be an absolute shit storm, don’t see how this will make a lick of financial sense for corporations to pay this.
I have limited experience, but had some licensing costs in my budget over the years, and $2k/vCPU seems to be at the lower end of such systems. The lever some companies pull is to let you pay for test and developer systems. Then $2k/vCPU gets expensive very fast (from another comment it sounds that you only pay for production systems).
It can add up quickly if you use a couple of libraries and also have to license the OS and perhaps other software. Some proprietary software companies can be quite aggressive with what they consider as cores, e.g. if you have a large VMWare Cluster they consider every single CPU of that cluster for license costs, not just the ones the software runs on - because you could potentially scale the system up or down or do a live migration.
I'm not sure how that would even work with cloud providers, at the worst you would have to pay for the max number of cores you have used in a year (however, it's not even clear what a core is in that context with SMT and preemptable cores).
