For many years news outlets have chosen to not report exit polls until after polls have closed. They don't want to influence late voting turnout. The theory is that voters will stay home if exit poll results show a significant lead for one candidate, even though exit polls are unreliable. The last time the major TV networks used exit polling to call a race was in 2000, when several networks announced Gore had won Florida even though polls would remain open in the western panhandle for eleven more minutes (and of course, Gore would go on to famously lose Florida). The "panhandle problem" caused more than one conservative pundit to accuse the networks of potentially damaging Bush's chances in Florida, since even slightly lower turnout in the Republican-heavy panhandle would have hurt Bush more than Gore. The networks seem to agree with those pundits, and have since held their exit poll reports until after all polls in the state have closed.
In recent years early voting has grown more popular. In Colorado especially voters have been taking advantage of new mail-in voting.
Interestingly, news outlets are not at all reticent when it comes to reporting early results. These early results aren't from exit polls, and they don't show how people actually voted, but they do show the party affiliation (or independence) of all early voters. So for example today AP reports a big lead for Republicans in Colorado's early voting. It seems to me that early voting results have as much potential to undercut turnout as exit poll results, even if early voting results only show party affiliation. Only half of expected Colorado votes have thus far been received. How many voters will forgo voting, believing Republicans have already won?
Early voting reports are too much like exit poll results, there's too much chance of undercutting turnout. As early voting becomes more popular, the problem will get worse. News outlets need to stop reporting them. If they continue, the states need to stop publishing the party affiliation numbers.