For two reasons:

1. This requires explaining why only some kinds of information processing are privileged to be conscious, which seems rather arbitrary.

2. There's the question of levels of abstraction. Which information processor is conscious? The physical CPU, the zeroth VM, the first VM, the second VM, etc.

3. And there's the question of interpretation. What is computation? A CPU is "just" electrons moving about. Who says the motion of these 10^12 electrons represents arithmetic, or string concatenation, or anything else? The idea of abstract information processing above the bare causality of particles and fields is in itself a kind of dualism (or n-alism, because Turing completeness lets you emulate machines inside machines).

Saying everything is conscious is also dualism. it's saying that every kind of computation (or perhaps every physical substrate) has another dimension of properties that aren't physical/structural and don't interact with the physical/structural world. So it's not an explanatory boon but rather an extravagance.

The 'where is the consciousness' question is interesting but not really a hard problem. The issue can be solved by being clear about what purpose does consciousness serve then locate where that need is realized. Consciousness is about information integration and broad access as a substrate of decision making. Recursive integration identifies the where. But thinking in terms of nested VMs is sort of missing the point. The point is to trace the flow of information and find the points of broad integration. This may involve multiple substrates. Identifying a single thing as being conscious is a mistake. The consciousness is the most narrowly specified causal dynamic that grounds the information integration.