It's a bit like saying that tracking a patient's weight on a scale directly causes a 960% drop in their blood pressure; data tracking can inform, but it's not the primary driver.
We know that real health improvements come from lifestyle changes or medications, which might contribute 80-90% to the outcome.
Collecting data on model growth certainly gives us a baseline to compare against, but its direct causal influence on expansion is probably closer to a 5-10% factor, not the main engine.
For instance, if I track a child's screen time, that data helps inform a conversation, but it doesn't reduce their actual screen time; that takes a parent setting new daily limits.
The market demand for these models and developer innovation are the heavy lifters here, likely accounting for the bulk of that incredible expansion.