The two standards are compatible, the port end hasn't changed, just the connector. The shorter sense pins are just designed to help detect an improperly connected cable, so that the sense pins only connect when the power pins are definitely connected.
Buildzoid and others have covered the design issues of the 12VHPWR cable and especially it's horrific implementation on the 5090FE well enough[1] that I don't think it's worth going into too much detail here. For some god forsaken reason, they decided to just dump all of the voltage lines in parallel and then run a single shunt resistor to it, so if a conductor fails and the load becomes improperly distributed, there's no way for the card or the user to know until it catches fire. It's hard to come up with a reasonable justification for this.
But just so we're clear, there are 2 reports of catastrophic failures with the 5090 already, which should be even more alarming considering how few 5090 actually exist right now. The other failure didn't involve third-party cables.
Of course, if used improperly, you can burn through basically any cable, and any cable can fail... but when the failure rate of a specific cable is so high above the rest, it raises many questions. If a specific model of aircraft seems to have an oddly bad problem with pilot error, you can't just shrug that off. In my opinion, consumer computer equipment is the same. It shouldn't light on fire unless you've done something horribly wrong. And even if you do something horribly wrong, the hardware should at least be designed in a way that gives it a chance at failing gracefully first. The connectors that 12VHPWR replaced were specced with good safety margins and previous NVIDIA cards were designed to ensure current was balanced across voltage lines.
It is unclear why NVIDIA didn't see the issue with the 12VHPWR last generation and put some serious effort into fixing the problem. If they continue recklessly like this, there is a non-zero chance that the 12VHPWR connector is only retired after it finally causes loss of life.
Buildzoid and others have covered the design issues of the 12VHPWR cable and especially it's horrific implementation on the 5090FE well enough[1] that I don't think it's worth going into too much detail here. For some god forsaken reason, they decided to just dump all of the voltage lines in parallel and then run a single shunt resistor to it, so if a conductor fails and the load becomes improperly distributed, there's no way for the card or the user to know until it catches fire. It's hard to come up with a reasonable justification for this.
But just so we're clear, there are 2 reports of catastrophic failures with the 5090 already, which should be even more alarming considering how few 5090 actually exist right now. The other failure didn't involve third-party cables.
Of course, if used improperly, you can burn through basically any cable, and any cable can fail... but when the failure rate of a specific cable is so high above the rest, it raises many questions. If a specific model of aircraft seems to have an oddly bad problem with pilot error, you can't just shrug that off. In my opinion, consumer computer equipment is the same. It shouldn't light on fire unless you've done something horribly wrong. And even if you do something horribly wrong, the hardware should at least be designed in a way that gives it a chance at failing gracefully first. The connectors that 12VHPWR replaced were specced with good safety margins and previous NVIDIA cards were designed to ensure current was balanced across voltage lines.
It is unclear why NVIDIA didn't see the issue with the 12VHPWR last generation and put some serious effort into fixing the problem. If they continue recklessly like this, there is a non-zero chance that the 12VHPWR connector is only retired after it finally causes loss of life.
[1]: https://youtu.be/kb5YzMoVQyw