Nvidia has reportedly put the development of another flagship GPU "above" the RTX 4090 on hold at the moment, according to Internet Insider, because it allegedly melted power supplies with its exorbitant power consumption.
According to a very reliable source from the GPU insider "Moore's Law Is Dead" (MLID), it is not said to have required 800 to 1000W (as some speculated in the run-up), but even with a consumption of 600-700W it is said to have repeatedly triggered fuses during operation and caused power supplies to blow.
The last prototype is said to have been four (!!) PCIe slots wide, which is why it was more practical to "hang" the mainboard on the side of the card instead of letting the card reside on the mainboard. To do this, it needed two of the new 16-pin power connectors at the same time.
In short: Allegedly, the time is simply not yet ripe for such a flagship (with 48 GB memory, which should also run with at least 24 Gbit/s and could then perhaps additionally push the power consumption towards 800W). And that is why Nvidia is said to have at least paused with the planning, even if many machine learners would already blindly buy such a card today even beyond a 3,000 euro price limit.
In any case, the current flagship RTX 4090 does not yet use the AD102 GPU to the fullest extent, which is why there is still some space "above" it in the product portfolio for an RTX 4090Ti / Titan.
Perhaps Nvidia is simply waiting a little longer for fast, economical memory. Or for concrete AMD specifications of the next competitor generation. Or maybe Nvidia just wants to see how the markets are developing. After all, in the current crisis situation, consumers' money is not as loose as it has been in recent years.