People always underestimate the market scale of "legacy chips". TSMC has 32% of its money coming from the planar transistor business (28 nm or older) and considering the skyrocketing demands they had as the market leader on cutting edge nodes, that number is absolutely massive.
However, as time goes by, the gap between legacy and cutting edge might either be more and more prominent or will diminish. Critical computing infrastructure, especially ones that needs brute forcing (AI for instance) is expected to see a divergence, but we never know if that's ever true as someone might come up with a workaround by using less transistors (though apparently unlikely). If true, we're not even sure if it's an ever growing trend either.
On the other hand, budget consumer electronics can stay at the 7 nm generation in a long time. Any computer with the 16 nm generation process is still deemed fast these days, in comparison to say what 32 nm was to 90 nm (e.g. Sandy Bridge vs Prescott). The world should always be ready for China's slippery industrial strategies.