Making a CRT is easy. But to make a CRT that has 1GHz bandwidth, has less than 1% non linearity over the full brightness range, and which will give a similar brightness irrespective of the beam speed, and where you have amplifiers which have a 1GHz bandwidth and both DC stability and still can drive a 200pF load at 1GHz with 400Vpp and do so linearly is hard.
Nah, they never had it that hard. Even in the toob days (no shortage of deflection voltage there), they moved to distributed deflection -- it's just so much
better. And more sensitivity means less voltage, means less power needed and no need for matching between a row of tubes (high impedance outputs) and the transmission line. Practically just drive the thing with a cathode follower, and you're there!
Probably the big crusty electrostatic picture tubes needed voltages like that, but scope tubes have always been in the 200Vpp range max. Which is 100Vpp per plate, so you only need a 150-250V supply to do a good job of it.
Combined with distributed deflection and the deflection-enhancing shield mesh, I think they got down to something like 20Vpp in the 475, and less in the faster and special purpose units. All made with normal (30-60V?) transistors, hybrids and ASICs.
Since scopes long since moved away from high vertical deflection voltages, the only remaining application for high voltages at high bandwidths -- were actually high resolution CRTs, because the video bandwidth is upwards of 100MHz (e.g., >163MHz pixel clock for 1600 x 1200 x 85Hz refresh), and the cathodes must be driven with about 50-100Vpp. Late model CRTs used monolithic chips; possibly some of the best discrete transistors ever made were Sanyo parts, used in the early and mid model Trinitrons (and other high res CRTs), a typical example being 1GHz fT, 100-160Vceo and 200mA Ic (e.g., 2SC3995 I think?).
Today, such transistors are all but unheard of. A shame, because they're still handy for niche applications.
There are RF parts available for voltages like that, but it's usually because you get more power at higher voltages. There are some industrial 13.56MHz (and such) applications where a bus over 100V is desirable, so there are some parts made for those sorts of applications. There's some interest in SiC and GaN for faster power switching applications as well (pushing 1 or 2 MHz, nothing crazy), but it's still very early.
Tim