I'd say the opportunity is more limited in investment banking, but there is still demand on the asset management side. Investment banks are the ones that created the now infamous CDOs and other complex mortgage-related structures.
Asset management using quants is done by mostly by hedge funds but also some traditional (a.k.a. "long only") investment managers (think mutual fund managers). Note that investment banks have traditionally traded their own money (known as proprietary trading), but that's happening less now because their a) many of their traders have done terribly and b) they don't have the same volume of money to trade.
To add one more wrinkle, most sizable investment banks also have asset management units where they take client money and invest/trade it for them, and/or create hedge fund structures/strategies to do the same.
I still see a lot of new job postings for PhD math people by asset managers, and occasionally investment banks. Try theladders.com or eFinancialCareers.com to get a flavor for what's out there. Another broad site that works very well for jobs if you can figure out the right search terms is indeed.com.
As another poster said, just because you specialize in one field doesn't mean you can't do something else with it. Lots of the math applies different places. If you really wanted to go whole hog you could do a math PhD in anything and then get an MBA with a finance concentration.
Note that Taleb is a smart guy, but he does have a product he's selling. He has many valid points, but there are firms still making millions or even billions of dollars using the statistical models he vilifies. Over the longer term there's too much skew/kurtosis/etc. from the human element (finance is a social science), but over shorter time periods there are lots of situations that work just fine with the kind of models you'd be able to build. Just always beware of leverage!