The SMC itself is not fast enough, per say. A NOP instruction takes 170 ns. You still need an external circuit to precision delay the reset, but the circuit is not that complicated. I've noticed the pulse width is not that important on my Falcon, the moment where the pulse begins is.
Regarding security, and how MS might fix the glitch vulnerability, here are my thoughts:
The IBM processor as provided by IBM is not really a secure processor itself, nor should it be. The security layers are provided by microsoft.
The mistake MS made was in not doing anything to prevent hardware attacks, as Tmbinc noted back in 2007. That includes glitches on power, clock and reset. You don't have to worry about voltage glitching on a big processor because of capacitance.
Clock glitching is tough to protect against. Reset glitching is not. Regardless of whether your system needs security or not, even if your system uses asynchronous resets inside, you always buffer and extend the reset to ensure clean resets going to the asynchronously reset flops. Ultimately this was Microsofts responsiblity, not IBM. You never just pass an external reset signal directly into your logic where security is a concern. MS had ample opportunity to heed Tmbinc's advice and correct this each time they spun the CPU, they did not.
We might still see a properly buffered reset on future CPUs and it wouldn't be that hard or expensive for MS. If they are like most ASIC designers, you sprinkle unused logic gates around your custom logic. This is done so that if you need to fix anything, you use the existing (though unused) logic already there, and you only change the metal layers (the wiring) not the silicon (the logic).
Where as changing the logic (silicon) mask can cost a million dollars and take 6 months to spin, changing the metal interconnect often costs tens of thousands, and could be spun in a month. Not to mention you don't waste leftover bare dies that haven't had metal layers added.
Even cheaper of course is to just change the PCB. If MS thinks they can reliabliy fix the problem with a new PCB, we'll see yet another one after Corona, rather than a new CPU. But, that's still avoiding the original problem with the reset.
Now of course MS engineers should KNOW ALL OF THIS, they're not stupid, and I'm sure someone is saying to their managers, "I told you so!". I suspect any future hacking depends on what MS managers decide, not what MS engineers know or don't know.