Intel Corp (INTC)

[Unkwn]

He never said that. That's pure nonsense. But my point was about immigration.

No, I think what he said was this:

When Mexico sends its people, they’re not sending the best. They’re not sending you, they’re sending people that have lots of problems and they’re bringing those problems with us. They’re bringing drugs. They’re bring crime. They’re rapists… And some, I assume, are good people.

and this, I believe

I will build a great wall – and nobody builds walls better than me, believe me – and I’ll build them very inexpensively. I will build a great, great wall on our southern border, and I will make Mexico pay for that wall. Mark my words.

Here is a list of more of such quotes.

And yes, this discussion doesn't belong to this board, I admit. I am not a US citizen, so I have no vote. Still, I can tell you the world is watching closely who is about to be elected as the world's most powerful man (or woman). One last word: The candidate isn't to be mixed up with the party.

Now back to topic, at least from my side:

Kaby Lake is coming later this year, and there is no doubt that 3D XPoint is going to play a big part in the future platform. Using 3D Xpoint-powered Optane SSDs may change our perception of what fast storage means, but utilizing it as memory has the power to change the entire computing paradigm.

You really have to be careful with that. The problem with 3D Xpoint, as with basically any NVM, is, that it wears. That's a fundamental difference to what DRAM is like. That doesn't even take the performance into account, where 3D Xpoint (or PCM, whatever it is) is still much slower than DRAM, where DRAM today already isn't really fast enough for modern processors (that's why this utterly complex caching is needed). I see no way 3D Xpoint will replace DRAM, at least not in higher performance applications, including mobile. For most controller applications, it will be too expensive I guess.

I see a market for 3D Xpoint as NAND replacement, but only when costs are at least in the same range as NAND, which it isn't. It may be interesting for some high performance applications in datacenters, but SSDs are already pretty fast these days.

The big question in my opinion is this: How is it going to scale in the third dimension? Will it scale better in terms of cost/density than 3D NAND? Then it might have a great future, but just not yet. I guess/hope that's what Micron and Intel are really targeting at and hence its name with the 3D included. We really don't know yet (and Samsung/Toshiba are certainly working on it as well).

Judging by the addition of persistent memory instructions in ARMv8.2 and the support it looks like Cavium will put in for it in ThunderX2 it is something that is generally considered as one of the big coming things and not just by Intel. But if they are planning for it then they would be depending on Micron to provide it rather than Intel unless Intel goes directly into the memory business.

Intel already is directly in the memory business. They are producing NVM chips and they sell their own SSDs. It's just that they cooperate with Micron.

Yeah, commodity NAND...look at Micron's enterprise value -- approximately $18 billion. That is what the market values Micron's entire memory business at, including DRAM, NAND, HMC, etc. Intel's memory business isn't realistically ever going to be as valuable, let alone more valuable, than Micron's.

$18 billion divided by ~5 billion shares, means that a Micron-level memory business could add a whopping $3.6/share in value. A half-as-successful business would maybe add $1.80.

Don't count on the memory business driving any material upside.

Yes, memory is a commodity, but it is a commodity in the hands of just 4 significant producers worldwide, with extremely high barriers for entry. Just imagine worldwide steel production to be in the hands of just 4 producers! What would that do to pricing in the long term? What benefit would they (esp. Samsung) have to start a price war and try to kick out the remaining few, which just would increase international anti trust pressure on them?

Regarding the market potential: Don't mix the current memory market with the one in the future. SSDs are about to replace magnetic disks and that is a huge potential if you consider growth in data centers and replacements in PCs. Also IoT may be a big memory consumer in the future, as those devices get more and more complex (and integrated flash/SRAM simply isn't enough anymore). In my opinion, the memory market is going to grow strong in the future.

I'm personally less worried about the hardware and eventual availability of silicon than I am about software support.

"If you build it, they will come" may work for Kevin Costner but in computers it has proven to take the better part of a decade in some cases.

Fully agree! A third type of data storage, NVM (flash), fast NVM (3D Xpoint) and volatile (DRAM) memory needs a lot of software changes and won't happen overnight. A good application could be as a fast write buffer in SSDs, but that is already managed well by DRAM buffers (though it has issues with sudden power drops) and the market volume will be pretty small. I don't see anything revolutionary here.