That's a valid point, and to be clear, I'm not claiming Oracle's making a wise choice. Among other business risks with the plan is that customers who've been burned by Oracle abandoning the specific product line they needed might not be quick to start using Oracle for all their AI stuff. Like I don't think anyone likes using Oracle's products, and if you're starting your company's AI plan from scratch, that might be an excellent time to consider not-Oracle as your provider/partner/contractor/whatever.

However, if your dead-end investment has a 20% ROR, and you think your AI investment will have a 1000% ROR, you'd kinda be foolish not to throw every possible resource at the new venture. I'd bet that a lot of the people maintaining Product A could be switched to the AI project. It's going to need lots of people supporting the networking, CRUD APIs for various things, sales, tech support, legal, etc. The product itself will be different, but much of the underlying support might look the same between the two. Why "waste" someone maintaining Product A's CI/CD pipeline when they could be helping the AI project move faster?

And again, I'm not arguing that they're right. It's more that if they're completely convinced that this is the future of their company, then that could be a rational, defensible decision. There's a lot of "ifs" in all this, to be sure. And in any case, don't make the mistake of anthropomorphizing Larry Ellison.