There's lots to this if the other practical of this don't sink it.
Lots of times, say at night, the wind blows but there's no demand for the electricity. You could store it in batteries but then you pay an efficiency cost of conversion into and out of the battery. If the battery isn't local to windmill or eventual use point you pay the transmission line cost. And the batteries have to be large ( expensive).
The servers convert the surplus electricity to useful products directly with no inherent inefficiency over any other server doing the job that is powered off the grid. So in that sense it's perfectly efficient. The data product is inexpensive to transport over Internet.
At times when the windmil is not producing power at all the data center could in principle be powered by sending power in the opposite direction back to the windmill site from the grid. You'd need a transformer there I suppose but not a new set of transmission lines. So lower maintainence and fixed costs for the power distribution system to the data center
But you do need a data product whose production can shut down when the windmill is producing power for the grid at max capacity. That's easily conceivable as computers can also be gridded around the world and their computations moved from center to center ( eg night on other side of planet ). Or you pick a data commodity that can be episodically produced. I shudder to mention Bitcoin but at least for the time being that's actually practical --- module your feelings about whether bitcoins wasteful work model makes sense. Another one would be high compute loads for jobs like designing proteins, scheduling airlines, training AI models, massive fluid dynamic simulations , weather modeling, etc.... and all sorts of scavenger computing jobs.
A bonus for this model is that you could build out wind power stations in advance of power demands ( like before you decommission your coal and nuke plants ) and have a use for them . If you built out enough excess capacity then you also solve one of the reasons we have these traditional plants -- the need to have enough power when winds are insufficient. With enough excess capacity and well geographically distributed then wind fluctuations won't ever need much base load backup power from coal