Might competition increase? The competition is already so high with almost all the tech giants already spending so much money and resources on their AI technologies and they hold the largest market share in the AI industry because they are doing their best to provide their users with the best models and tools they can use at the lowest prices possible. I know we can't really say whether they are going to make it big in the AI industry or not, but to be honest, I think it's not really a great idea to get into a market that is already so much saturated with so many giants holding the top positions.
They are not going into AI to build the technologies, they want to build HPC infrastructures for computing customers and they might get some profits if they do it well and its a means of diversification and joining the AI race..
Though the giants are already in the market, the market is big enough and expanding quickly for them to make pretty good profits.
What people do not understand about the AI world is that you do not have AI for just "all things" in a single place. You could use Claude for coding, for example, but it is shitty at imaging; you could use nano/gemini for images, but it is shitty at chatting; you could use ChatGPT for chatting, for example, but it is shitty at coding. And these are just the three together; none of them probably have any AI expertise in... HPC infrastructures?
I have no idea what that even is, but yeah, if you want to build an AI for a niche thing, you would be able to do it. A friend of mine- well, not a friend, but someone I know online- is building a football scouting AI; it literally is putting up data from many places- videos, images, data- all combined into a single location, and with AI you could literally sort through millions of people from around the world.