Wrong. He isn't even running on cloud anyway. For this type of workload, bare-metal is 10 times a better option, both in terms of performance and costs.
The "problem" this dev is having is what AWS was built for.
He could get his OS of choice, customize it with his server software, then copy the AMI to various regions, spin up dedicated instances as he needed them, use what he needed in various regions and direct players to their closest server. Turn on and turn off servers as need arises and they are close to players. After his game settles down to predictable traffic, he can move to some bare cost ISP. He would not have to call anyone and within a hour or so he could have a game that gave a much better experience.
What kills a flavor of the month game like this is being laggy and having an awful experience.
The only thing I think would be an issue is the bandwidth charges. I wonder why his game creates so much traffic?
The game is very latency sensitive and it makes complete sense to use a cloud and distribute the servers based on the user load : locations and that's what he plans to do as mentioned in the article.
He could get his OS of choice, customize it with his server software, then copy the AMI to various regions, spin up dedicated instances as he needed them, use what he needed in various regions and direct players to their closest server. Turn on and turn off servers as need arises and they are close to players. After his game settles down to predictable traffic, he can move to some bare cost ISP. He would not have to call anyone and within a hour or so he could have a game that gave a much better experience.
What kills a flavor of the month game like this is being laggy and having an awful experience.
The only thing I think would be an issue is the bandwidth charges. I wonder why his game creates so much traffic?