> What is the difference between this and having to sell the cloud access and supporting the people who buy a subscription
Knowledge/training.
If you're shipping a brand new hardware arch, exposed as raw hardware, then you're on the hook for training everyone in the world and fixing all their weird edge case uses.
I.e. are you willing to invest in Intel/AMD/Nvidia-scale QA and support?
If you're exposing a PaaS (or even IaaS), then you have some levers you can tweak / mask behind the scenes, so only your team need be experts at low-level operations.
For a fast-paced company, the latter model makes a lot more sense, at least until hardware+software stabilizes.
> I.e. are you willing to invest in Intel/AMD/Nvidia-scale QA and support?
At least in our experience, the first line of support is from the chassis vendors. You don't go to a store and buy MI300x. You buy them from someone like SMCI/Dell, who provides the support. Of course, behind the scenes, they might be talking to AMD. Even those chassis companies often have other providers of their gear (like Exxact) as another line of defense as well.
In the case of Groq, it would have been death by 1000 cuts to have to support end users directly, especially if they are selling small quantities. It is much easier to just build data centers full of gear, maintain it yourself and then just rent the time on the hardware.
Knowledge/training.
If you're shipping a brand new hardware arch, exposed as raw hardware, then you're on the hook for training everyone in the world and fixing all their weird edge case uses.
I.e. are you willing to invest in Intel/AMD/Nvidia-scale QA and support?
If you're exposing a PaaS (or even IaaS), then you have some levers you can tweak / mask behind the scenes, so only your team need be experts at low-level operations.
For a fast-paced company, the latter model makes a lot more sense, at least until hardware+software stabilizes.