Nvidia's rumored plan to create a "mining" gfx card segment

I see. That is a good point actually, and you are correct that it is a very volatile market.

It seems like the best way to tackle this, would be to try and predict the periods of boom and bust in the cryptomining card sales?

As for the capability of producing enough GPUs, I am not sure about AMD, but Nvidia definitely seems to have the resources for that. Its only a matter of allocating the budget for it.

I remember a while back, some representative of AMD was interviewed, saying that they aim to increase graphics card production for their next gen cards to come in 2018.

https://www.polygon.com/2018/1/30/16952964/amd-radeon-gpu-supply-shortage-cryptocurrency-mining

I see. So what you're basically saying is that there's no way Nvidia can predict what the market is going to be like a year later, so its more or less just a gamble.

Your idea to cripple mining cards' performance in gaming to sort of force miners to go for mining cards is good, but perhaps they can do this more subtly.
For example, they can make their gaming cards perform slightly less well in mining, and their mining cards slightly worse in gaming (than they should, given the hardware). This could be done on the driver level, so it will work across the board for all cards (even older gen).

However, I dont think that would be necessary at all, if their gpu's are designed for the intended usage.
For example,[strike] a mining card wouldnt need a high Vram capacity. Therefore, those cards would have low vram, in order to reduce the cost and to make them less suitable for gaming.[/strike] Edit: Actually, thats not true, in some cases even more than 4GB of vram might be needed.
Also, if the gpu itself is optimized for mining and not gaming, and this is done on the hardware level, having the capability to do more calculations per second than a gaming gpu, then there's no need for any driver restrictions on the gaming cards themselves.
The gaming gpu's can be more optimized to handle a large memory buffer, and this would necessitate a high bit memory bus, etc.

One other thing to consider, is that the miners look for cost/performance very much. Its the only thing that matters. So they just wont buy a gaming card in order to mine on it, since its not a good value for money.
Nvidia doesnt actually need to cripple the mining performance of gaming cards. They just need to make their mining cards more efficient, i.e. offer more performance in mining than the $ equivalent of a gaming card.
This, coupled with the fact that the mining cards wouldnt be such good performers in gaming, would mean that someone needs a Mining rig for mining, and a Gaming rig for gaming.

Some people might complain about this, yes, but I think in the long term this is good for everyone, including Nvidia themselves.

Yeah, you're making a good point.
Since its incredibly difficult for Nvidia to predict what the market is going to be and how many gpus will be required, miners will inevitably turn to gaming gpus.

Perhaps if the new gaming gpus simply cannot mine, then the problem will be solved. And while that will make miners unhappy, I think that simply because there will be a demand for it, someone is bound to make more mining gpus to satisfy that demand.

Honestly, it seems like an extreme solution, and Nvidia wont do something like that.
However, I'd say that if the problem isnt eventually solved, then both Nvidia and AMD risk losing a ton of their market share to consoles and perhaps even Intel, since Intel is hard at work in creating their very own debut discrete graphics card.

Both AMD and Nvidia, need to consider their strategies going forward with this, because we've already seen gamers turn to consoles when graphics cards turn to be overpriced.
At the end of the day, a good gaming PC will always be more expensive than a console, but when its price is double or even triple that of a modern console, then gamers tend to be turned off and go for the console.