The Congressional Budget Office (CBO) is not expected to release a score for the GOP's Obamacare "replacement" bill until next week, but the Brookings Institute has produced its own estimates based on past CBO reports.
According to Brookings, at least 15 million people will lose their coverage under Trumpcare, or "Ryancare," if you prefer.
There is significant uncertainty about exactly how CBO will model these provisions and how it will expect the various provisions to interact with one another. Nonetheless, we conclude that CBO’s analysis will likely estimate that at least 15 million people will lose coverage under the American Health Care Act (AHCA) by the end of the ten-year scoring window. Estimates could be higher, but it’s is unlikely they will be significantly lower.
What should immediately stand out is the Brookings estimation that "possibly millions more" could lose their coverage, meaning 15 million is a conservative or low-ball estimate.
I suspect the number could be significantly higher than 15 million unless states that expanded Medicaid under Obamacare develop their own ways to maintain coverage. But that seems unlikely in some cases. Large, wealthy states like New York and California may have the resources to maintain coverage, but small red states that accepted Medicaid expansion did so primarily because they're poor and needed the federal money to prevent local healthcare systems and rural hospitals from collapsing.
There's no way a state like Arkansas can maintain the same level of Medicaid coverage, which is probably why Senator Tom Cotton is no fan of Paul Ryan's approach.