The reasons for this shift in budget away from funding Free Software and the NGI initiative seems to be an allocation of more funds for AI, leaving internet infrastructure by the wayside. Meanwhile, the EC has thus far declined to comment to share its official reasoning for striking this funding from its budget.
Investing into AI seems/feels more speculative and inefficient. I think you can get a lot more value by investing the same into actual, practical projects. Training AI, and training it well, is very expensive. And the gains or results are not necessarily even predictable, let alone certainly useful or used.