Micron-Intel 3D XPoint split: It's not you, it's m... nah, it is totally you
Breakup lays bare Chipzilla's failures
Intel's Optane XPoint product sales have been disappointing and Micron is facing a collapse in sales of its XPoint chip output to Intel. Intel is, of course, the sole route to market for XPoint chips from the jointly owned Intel Micron Flash Technologies (IMFT) fab in Lehi, Utah, in the United States.
After the association is ended, the two will pursue their own XPoint developments and sell their own products. Micron's QuantX kit is due to hit market late in 2019.
Several inferences could be drawn from this:
- Intel failed to sell sufficient Optane 3D XPoint SSDs.
- Optane XPoint drives do not have sufficiently faster performance compared to NVMe SSDs to make their cost worthwhile.
- Intel's emphasis on the gaming market rather than the far larger enterprise SSD market was a mistake.
- The need to have x86 processor-level technology to use XPoint storage-class memory was another error because that technology has been slow coming.
- A basic marketing error Intel made was to wildly over-promise XPoint performance (1,000x faster than NAND) and then under-deliver.
- 3D XPoint endurance superiority is secondary in a performance-led market.
We're told that gen-2 XPoint development will complete in 2019, implying that the Lehi fab will be producing the chips from then on. Micron and Intel will then pursue independent XPoint developments, which we can refer to now as being gen 3.
Once either has completed its gen-3 development, it will need to manufacture those chips at its own facility.
As Intel and Micron are separating over 3D NAND development, we could be looking at the end of the IMFT joint venture. One of the two will need a whole XPoint fab.
This has been Intel's failure: XPoint is stuck at two layers when NAND is hitting 96; Optane technology is both capacity and performance-limited; it has been mis-marketed by Intel from the start; gamers getting a better gaming experience from XPoint cuts no ice with enterprise IT buyers and system sellers.
Optane's performance advantage is seemingly so minimal that Samsung can make a competing Z-SSD product by souping up NAND technology.
We might conclude gen-1 XPoint is not a good fit for enterprise IT and the timing of the split - before gen-2 XPoint chips are even produced - is interesting.
It seems to The Reg's storage desk that, if gen-2 XPoint product sales at Intel disappoint, Chipzilla could abandon the whole product or look to sell it off. That is what happens when a company believes its own marketing. ®
Sponsored: Beyond the Data Frontier