First, it turns out that white voters make up a larger share of the electorate than people tend to think. So many more, in fact, that there is a fairly stable path to victory for Trump even if he does no better among minorities than Mitt Romney, and even if he suffers among college-educated whites (more on that later).

Second, these pundits have been hasty in assuming Trump would not do well with minorities. Right now, he is trending to not only meet, but to exceed, Mitt Romney’s 2012 performance. If he manages to surpass his predecessor by a few percentage points with whites or key minority groups (and especially if he manages to do both), then he’ll likely be classing up the White House come 2017. And right now, things are looking pretty good.

At the moment, roughly 10% of African Americans have a positive view of Trump, and another 15% are undecided between Trump and Clinton. At first blush, this does not sound great—winning between 10-25% of the black vote would still mean he’s overwhelmingly unpopular with African Americans. However, given that Mitt Romney won a dismal 6% of the black vote, Trump stands to exceed his predecessor by a wide margin. This is a big problem: as Van Jones pointed out, for Hillary to win in November, black turnout would have to at least match 2012’s numbers, and she would have to win 90% these votes. Right now, it’s looking like she might not hit either of these benchmarks.

Although Trump is staunchly anti-immigrant and routinely blames China for America’s decline, he is trending to meet or exceed Mitt Romney’s numbers among Asian voters.

What’s perhaps more surprising is that Trump may even outperform Mitt Romney among Hispanics. Despite his depiction of most Mexican immigrants as rapists, criminals and job stealers, or his commitment to build a giant wall between the U.S. and Mexico (and make Mexico pay for it), or his plan to deport undocumented Americans en masse, 23% of Latino voters support Trump, with another 15% undecided—putting him on-track to meet Romney’s 27%. Even if he just manages to hold his current margin, it will be very bad for Hillary Clinton; if he does manage to match or surpass his predecessor with this key demographic, Trump will almost certainly be president.

A Coalition of Antipathy

After Mitt Romney’s humiliating loss in 2012, the GOP unveiled an “autopsy” of what they believed went wrong. According to their reckoning, the demography of the U.S. was trending away from their core constituency, and towards that of their rivals. If the Republican Party wanted to have a viable shot at the White House going forward, it would need to increase its share of black, Latino and millennial voters by increasing outreach with these groups and softening their tone on contentious social issues. Trump’s strategy is basically the opposite of that, but here’s why it may work:

People on both sides of the political spectrum tend to talk and think about minorities as a homogenous block. It is assumed that if a candidate says something bad about immigrants, of course Latinos will be upset. If a politician is bigoted against Muslims, of course black people would disapprove. If a public figure makes insensitive comments about trans citizens, of course women would take umbrage. Turns out, this is false—and Trump has been able to peel off many minority voters by appealing to their aversion towards other minority groups.

For instance, many African Americans are critical of immigrants. In fact, many Latinos are unsympathetic towards undocumented immigrants, even from Latin America: they are concerned about border security; they often feel they’ve “earned” their place in America, and that others should do the same. There are large numbers of black and Latino evangelicals who harbor Islamophobic sentiments; there are even some Hindu Indians who are supporting Donald Trump because of his anti-Muslim rhetoric.

Moreover, it is likely that Trump will win a larger share of the LGBT vote than any Republican candidate has. This is not only for his unorthodox positions on gender and sexual minority issues, but also in part due to his hardline rhetoric against “radical Islam”: LGBT and feminist movements have long been complicit in anti-Islam fervor—a trend which may be exacerbated in the wake of the ISIS-inspired attack on Pulse in Orlando (despite the many voices from the LGBT community who aspire to push back against these reactions).

As a matter of fact, Trump’s anti-Muslim proposals are actually supported by a majority of Americans. Indeed, many of his espoused views enjoy much wider support than most seem willing to believe or acknowledge—cutting across gender, class, ethnic, and even party lines.

Does Trump Have a White-People Problem?

Here, the critical reader may object that despite Trump’s surprising potential among many minority groups, he has some significant vulnerabilities with white voters. There are a few ways to respond to this:

First, while Trump’s approval rating is lower than Romney’s was among white females and college graduates, this is largely offset by gains with male white and/or non-college educated voters. And even among women and those with degrees, what is particularly striking is that Hillary Clinton does not seem to be benefiting much from Trump’s alleged unpopularity. For instance, Trump is down by 10 points relative to Romney among college-educated whites, but Clinton is only up 3 points relative to Obama’s 2012 performance. Or more strikingly, Trump is down relative to Romney by 9% among white women, but Clinton is only up relative to Obama’s numbers by 1% in this demographic—despite being the first viable female presidential candidate in U.S. history!

Moreover, there are reasons to believe that Trump’s actual support among women and college-educated whites may be under-represented in polls. One strong piece of evidence to support this conclusion are exit polls from the primaries, wherein Trump supporters tended to be both wealthier and more educated than the average American.

Finally, while much is being made about the supposedly deep rifts in the Republican Party over Trump, it is important to distinguish the Republican establishment from its voting base: the entire Trump phenomenon is reflective of a deep cleavage between the two. The fact that the likes of Paul Ryan, Mitt Romney or Lindsay Graham have trouble endorsing Trump indicates virtually nothing about how most Republicans feel.

In short the rhetoric about Trump’s vulnerabilities with white voters seem overblown; he may well win the white vote by a larger margin than any candidate in recent history. And while his numbers with minority voters may be objectively low, he seems poised to secure more than enough support from black, Asian and Latino voters to deny Clinton the White House.

Yes, Trump is a parochial demagogue—but this does not guarantee that minority voters will turn out in high numbers or sufficiently rally behind Clinton. If Democrats take minority voters for granted (as they typically do), they will lose in November.