I see that Linear Technology (LTI) has introduced a new op-amp that they are calling a 140 V device.
The specs look pretty darned good — single supply operation of 140 V supply or ±70 V where you need bipolar operation. Rail-to-rail output, so you can take full advantage of the available high-voltage supplies. Input leakage/bias current is as good as many low-voltage devices — typically 3 pA. The input offset voltage is nothing special at a maximum of 1.6 mV, but tolerable considering the device. Gain bandwidth product (GBW) is 10 MHz and slew rate is 19 V/µs. Low frequency noise is spec'd at 3.5 µV p-p; so pretty good for the likely applications.
Those applications are typically as a pin driver in automated test equipment (ATE), as a piezo-driver, or possibly to buffer and amplify the output from a DAC. Piezo devices are often used as actuators and occasionally as transformers. Audio applications are a possibility too.
Conceivably, this op-amp could be used in a servo-amplifier. Combined with suitable power FETs, you'd have a dandy high-power amplifier. I did a design like that to drive a 90 VDC motor a few years back. I had to jump through hoops to come up with a design. I had to level shift and amplify the output of a conventional ±15 V op-amp up and down suitably to drive the output stage (and it had to remain stable — a desirable feature). This portion of my design used a handful of bipolar NPN and PNP transistors.
Of course, if I did that same design now, I'd probably use an H-bridge topology and a PWM drive. But that would be specific to a motor drive. The pin driver and piezo-actuator applications will likely remain purely analog in the foreseeable future. Further, they are not high-power applications, so they are a good match for this op-amp.
The LTC6090 is available in the standard SO-8 package and in a TSSOP package; the TSSOP package includes guard pins adjacent to the input, output, and supply pins. Why do you care about guard pins? If you are sending low-level signals to the op-amps high input impedance inputs — and considering that right nearby there will be some very high voltage — you will surely care about this feature.
And just one more detail about the packages. They have a copper slug that can be soldered to a copper plane on the PC board to help pull heat out of the package.
With a ±70 V supply and a possible maximum output current of 10 mA, power dissipation must be considered. The quiescent current for the device is just under 4 mA. Small quantity pricing is around $3.50.