What would you do if you wanted to dump a capacitor holding 100 joules at 1000 amps in a millisecond, briefly dissipating 100 kilowatts?
Most IGBTs are not equipped to deal with pulse currents like this, and I don’t think you can parallel them the way you can MOSFETs, due to current hogging by the hottest device; the IXGX320N60B3 costs US$22 and is rated to dissipate 1700 watts itself, and IXYS doesn’t publish a datasheet for it. Similarly for triacs.
MOSFETs like the SIHB33N60E-GE3 are maybe more promising: for US$6 you can switch 600 V and pulses of 88 A with 0.1Ω with a 150-nanosecond turn-on delay plus rise time, and you can safely parallel them. (88 A is almost three times their maximum continuous current of 33 A.) So if you put a dozen of them in parallel (US$70) the datasheet claims you can get them to control a kiloamp pulse. (Probably a good idea to add enough series inductance to keep the pulse current from going higher than that.) The Infineon IPA60R099C6XKSA1 is US$8 in quantity 1 and rated for 112-amp pulses, so you’d still spend US$70.
I wonder if a simple mercury-wetted reed relay would be a better choice.