Amazing! How would anyone go about making a diy version of an optically isolated probe, both on signal and power side? If not the full cable length opto isolation, then perhaps with optoisolators?
I vaguely remember seeing a Linear technology application note. It was about making linear characteristics for an optocouple. So they had an equal optocouple in the feedback, and it was used on the transmit side to correct for any nonlinear characteristics. I don't know how linear optical cables and fiber optic transmitters are, but that's one way of doing it. Or maybe they are doing it completely differently, since 10Gb and even 100Gb optical networks existed or a while, they might just send it all digital.
I've worked on such a device a decade ago. The entire acquisition system (battery powered) is connected using an optical (ethernet) network link and the data is send pre-formatted over the network. But this is for use in high voltage labs where they use it to measure into the MVolt / tens of kAmpere range and float the units to tens of kVolt.
The biggest trick is to get this techology to become cheap. You can probably get a long way using a battery and SFP + FPGA + ADC / DAC
For a "low" frequency system in the digital domain, there are Delta-Sigma modulators that output bitstream, that practically can directly be sent to a SPF and on the RX side it can be turned into an analog signal with an 1 bit DAC. How well this would work in practice, I don't know, probably DC accuracy needs calibration, and Delta-Sigma typically stops being practical at 10MHz. Maybe it's possible to push this a bit with very high speed opamps and comparators. In theory an optical probe with this would be super cheap, but also not very userful due to the limited frequecy.
For high speed, maybe it's possible to skip the FPGA, if one has a good match of interfaces. There needs to be an ADC, probably Flash ADC, generating data at that very high 100MSPS-1GSPS speed paired up with a SERDES. With the RX side, having a similar SERDES and a DAC. I think the difficulty could be finding such a SERDES, as most of them seem to be only targeting video interfaces.
I think if someone could put together a 50-100MHz probe without going crazy in the BOM and development costs, there is a market to grab.