What are your thoughts on full switch programmability in networking gear?

In the context of modern networking, the concept of full switch programmability is often touted as a game-changer. However, its practical application in production environments raises questions. Do you think it truly enhances network performance, or is it just a buzzword? Additionally, how do you feel about the competition between on-path cores in DPUs versus off-path models? Is there a real advantage to one approach over the other? Let's discuss the implications of these technologies in the industry.