For mostly-self-sufficient organisms in earth's environment, the versatility of carbon appears to outperform silicon on every metric that counts.
The molecular complexity of a single human lung cell still absolutely dwarfs that of even the most modern CPUs we are able to manufacture (apple to oranges, but true).
To reiterate, The belief that evolving machines have to match the kind of evolution we're subjected to is illogical. Machines wouldn't be there without us and we wouldn't have what we have now without evolving our machines.
We're just way more capable of high complexity.
But also don't forget that evolution is not determinist : although you are probably right that silicon based life cannot exist, it's not because it doesn't exist that it is impossible
What are you even measuring with complexity there? We don't go out of our way to make CPUs more complex, either chemically or in terms of circuits, their complexity is just what results from what we want them to do. What we want them to do isn't (yet) "reproduce" or "eat", unless you count the entire larger industrial ecosystem of which they are vital element as well as being a byproduct, as even this paper is in the hypothetical.
Looking at the abstract and skimming the rest, it is suggesting something involving a much bigger system than a silicon chip, but it's still more akin to how a virus hijacks the molecular processes of the actual factory (i.e. the cell it infects), or primates using bones as tools if you prefer a macro-scale analogy.