Intel's NAND flash was overshadowed by their Optane (not included in the sale) and lagging behind Hynix and Samsung, so I guess this isn't such a bad move. The cynical take is that the CFO running Intel is attempting a bunch of standard corporate restructuring strategies to delay the effects of their slowing technical innovation. It is definitely concerning from a US technical superiority standpoint to consider that this same CFO may soon try to sell off some of their more competitive fabs.
>Intel's NAND flash was overshadowed by their Optane
It is important to note Optane is manufactured in partnership with Micron. The partnership is that Intel still owns part of the IP, but actual manufacturing with guarantee capacity is still done by Micron. And judging from previous Micron Investor notes, Intel hasn't been keeping up with their Optane Sales and buying agreement. So that means Intel will no longer be doing any NAND / Optane manufacturing by themselves.
Few other point worth noting,
Intel has one of the best Enterprise / DC SSD range on the market. And they are extremely popular in wholesales channel as well as with System vendors. Intel has been using these SSD as weapons / incentives by giving away SSD with enough Xeon CPU volume. One of the few reasons why AMD EPYC hasn't been gaining much ground at all in Enterprise segment.
The NAND Plant is in China. Wroth pointing out in the current geopolitical climate.
NAND is a commodity market. It doesn't swing up and down like DRAM in the old days mostly due to less players in the market and HyperScaler's volume creating enough buffer. But generally speaking Intel doesn't like commodity business. May be not commodity market per se, but Intel doesn't like margin fluctuation.
Overall I think it is a good move. Most of the time I am very cynical of CFO running companies. But I give Bob Swan some credit because so far things seems to be moving in the right direction. While the pace might not be fast enough. ( Mostly due to inertia ) But at least it is better than BK's era.
There is a lot of flash makers, way, way more than DRAM makers. And there is tons of flash going to things other than PCs, unlike DRAM. Regular DDR DRAM only goes to high-end desktop PCs, and servers, with few laptops. Everything else has switched to the LP-DDR line even if there is no explicit need for energy saving (LPDDR4 may actually cost you less than regular DDR4 these days, especially given the extra for DIMM/SO-DIMM, through hole sockets, assembly cost, and wasted PCB space).
The real shitshow starts when the CFO becomes the CEO (as is usually the case), so Intel is pretty much dead as is US tech innovation on this scale (until VCs do what China does and start pouring money into hard stuff, not just fluffy software).
The challenge even when "VCs do what China does and start pouring money into hard stuff, not just fluffy software" is that like Feynman's There's Plenty of Room at the Bottom observation, there is plenty of innovation room in the intricacies of "low-end", "low-margin" manufacturing. Divorcing that operational, get-your-hands-dirty aspect from your business closes off insights that arise from the ongoing software-eats-the-world wave still building up steam around the world.
The more you cram together the people networks, culture, knowledge, tooling and practices of how to create all this stuff into a single ecosystem, the more they start making Burke-like Connections. Thinking you can Jira-innovate your way to dominance on an ever-narrowing technology stack is architecture astronaut'ing on a strategic corporate scale.
If Amazon had outsourced their IT operations to an IBM, would their internal culture still have de novo evolved the nascent devops principles back then into AWS?
It isn't what Intel gives up now by selling this business that will hurt their innovation. It is the options that suddenly get a lot more expensive to investigate in the future for future innovation. Technology innovation is evolution: extremely messy, more dead ends than successful adaptations, enormously energy-intensive from lots of majority-repetition, and highly organic. You can certainly optimize around the edges to make it slightly less jagged-edge, but innovation sparks at the tiny interstices, usually not at the top.
It’s natural for large tech companies to enter the harvesting business when they can’t innovate anymore. Look at IBM. They just can’t innovate, so they do all kinds of financial engineering on the back of a robust customer audit program.
Eventually that money gets recycled into funding earlier stage companies. (If a company buys back stock, the old investors have to put the money somewhere)
Unpopular opinion: Intel may be slowly dying, but this is fine and eventually happens to all ageing corporations. Looking at the technical quality of their non-x86 products, most of it was not great, so it is not surprising to see the rot spread.
> Looking at the technical quality of their non-x86 products, most of it was not great, so it is not surprising to see the rot spread.
What's a damn shame is Intel's StrongARM / XScale division in the early-to-mid-2000s. Methinks the world would be very different today if they continued investment in that sector - and we'd have more than just Qualcomm and Apple to contend with.
> Intel may be slowly dying, but this is fine and eventually happens to all ageing corporations.
The sad bit is, it will lead to consolidation at the hand of few players which would be worse for customers. Intel used to have good products in flash space.
I'm not sure I agree it was lagging behind. They were definitely not first to market with 3D-NAND but my gut feeling is that was because of the engineering focus on Optane. AFAIK they and Kioxia (toshba) were leading the charge on PLC memory.
All of that is ignoring the fact that there is absolutely a benefit for Intel to have vertical integration in this area. They make countless products that rely on flash storage - from NICs to switches. The second they have to rely on an outside party they have to pay markup for that company's margin and they're at that company's mercy from an R&D perspective.
This whole thing feels penny wise, pound foolish to me.
Indeed. I was briefly foxed as well, given Andy Grove talks about Intel bailing on the memory business back in the 80s right at the start of High Output Management.
Does Intel only sold the NAND manufacturing parts to SK Hynix or it also includes the SSD controllers IP? Because most of Intel SSD performance came from the controller not the NAND flash themselves, in fact Intel usually lag behind in NAND flash manufacturing.
It will probably include the controller IP. However, none of Intel's IP will be transferred until 2025. That should give them time to disentangle any of the IP that's shared between the NAND SSD controllers and the Optane SSD controllers, and give Hynix time to decide whether it's worth continuing development of both their existing enterprise SSD controllers and Intel's designs.
A tinge of sadness. I remember reading reviews of the X-25 like a decade ago and being so excited (it was this anandtech review https://www.anandtech.com/show/2614). To me it was the dawn of the "common SSD". Just now I've looked at a bunch of timelines of SSD history, and I suspect the X-25 was similar to many other people. There was certainly a lot of SSD activity prior to the X-25 being released, but it seemed like the only customer facing brand that had an offering before it was OCZ (you can see it in the review comparisons... you can also marvel at the old 10k RPM HDDs... how's that for a nostalgia trip).
Intel pretty much stopped being relevant to the consumer SSD market as soon as the SSD market as a whole broadened enough to support the development of client/consumer-specific SSD controllers. Once Intel could no longer compete by stripping a few features out of an enterprise SSD design, their consumer SSD products had no real advantage over other brands. (Minor exception: Intel's consumer QLC NVMe drives have consistently had very good retail prices.)
If you mean the 660p/665p, I think even those SSDs are not worth it. The Kingston A2000 uses the same SM2263 controller, is about the same price, but uses TLC NAND instead of QLC.
Oh totally. By the time I actually had my own money to buy an SSD (so like 2013 or so), Intel was completely out of the running (I went for a Crucial). As you said, I remember looking at some reviews last year and being surprised that Intel was in the running at all in any category.
Note: once upon a time, Intel's fab strategy involved new fabs making CPU's (x86) for a couple years, then being repurposed for memory. This meant that their memory fabs were practically free (i.e. they were building them for CPU business anyway), which helped them compete on price in memory.
I'm sure they still find some use for their no-longer-brand-new fabs, but this kind of examination of each business line in isolation doesn't take into account synergies like that. I can easily imagine Intel doing one "makes good business sense" decision after another, until they are no longer able to manufacture competitively at all.
Oh, wait, maybe that already happened a couple years ago...
Everyone seems critical of this sale as some sort of imploding canary for their impending financial doom but if they don't see a lucrative or competitive future for this business than I don't think divesting from it is a bad idea, using the proceeds for buy backs effectively makes their business smaller whilst retaining the same inherent value. It goes against Wall St's grow-at-all-costs philosophy, but personally I think sitting on a war chest waiting for the next big opportunity is better than pouring investments in a low margin business playing catchup from 5th place.
Personally I believe Intel needs some big wins as they're basically getting their lunch eaten from all sides. Fortunately they're still highly profitable which affords them a lot of opportunity for investments, if they can't do that their future may be dependent on others successes of their intelcapital.com arm, which worked well for Yahoo! who at the end, most of their value came from their Alibaba investment.
I guess it harkens back to their first few years, when they decided to sell off their DRAM business? Although, not so clear now whether it means they're on another course to greatness...
Yes, I was thinking of the same precedent. IBM got out of memory at about the same time, I believe, and for the same reason: Too low margins to make continuing to fight against the tough competition worthwhile.
>Although, not so clear now whether it means they're on another course to greatness...
This is the $64,000 question. Intel got out of the memory business when it was clear by 1985 that CPUs were its future. I didn't write x86 because Intel itself wasn't convinced of this; see iAPX 432, i860, Itanium, and StrongARM/Xscale. All failed to give Intel the backup/alternative to x86 that it has repeatedly sought.
It is not an exaggeration to say that Intel has failed, or at least failed to distinguish itself, in every single market it has entered since IBM chose the 8088 forty years ago. Besides the above alternative instructional sets/architectures and flash memory, antivirus (still perhaps the most mystifying move in Intel history), servers/motherboards, and GPUs come to mind. The most successful non-x86 business for Intel is ... Ethernet cards? Meanwhile, it's embarrassed itself with discrete GPUs at least three separate times. A few years ago buying Nvidia would have been a savvy move; now it can no longer do so, and it's not impossible to imagine Nvidia buying Intel in the future.
Even within the one market Intel does dominate, what saved the company 15 years ago amidst Itanic and Pentium 4 was 1) AMD Thunderbird being just as inefficient as Pentium 4 and, more importantly, 2) an Intel Israel skunk works project to improve on the Pentium 3. There is no such out-of-the-blue miracle this time.
Readit News
It is important to note Optane is manufactured in partnership with Micron. The partnership is that Intel still owns part of the IP, but actual manufacturing with guarantee capacity is still done by Micron. And judging from previous Micron Investor notes, Intel hasn't been keeping up with their Optane Sales and buying agreement. So that means Intel will no longer be doing any NAND / Optane manufacturing by themselves.
Few other point worth noting,
Intel has one of the best Enterprise / DC SSD range on the market. And they are extremely popular in wholesales channel as well as with System vendors. Intel has been using these SSD as weapons / incentives by giving away SSD with enough Xeon CPU volume. One of the few reasons why AMD EPYC hasn't been gaining much ground at all in Enterprise segment.
The NAND Plant is in China. Wroth pointing out in the current geopolitical climate.
NAND is a commodity market. It doesn't swing up and down like DRAM in the old days mostly due to less players in the market and HyperScaler's volume creating enough buffer. But generally speaking Intel doesn't like commodity business. May be not commodity market per se, but Intel doesn't like margin fluctuation.
Overall I think it is a good move. Most of the time I am very cynical of CFO running companies. But I give Bob Swan some credit because so far things seems to be moving in the right direction. While the pace might not be fast enough. ( Mostly due to inertia ) But at least it is better than BK's era.
Dead Comment
The more you cram together the people networks, culture, knowledge, tooling and practices of how to create all this stuff into a single ecosystem, the more they start making Burke-like Connections. Thinking you can Jira-innovate your way to dominance on an ever-narrowing technology stack is architecture astronaut'ing on a strategic corporate scale.
If Amazon had outsourced their IT operations to an IBM, would their internal culture still have de novo evolved the nascent devops principles back then into AWS?
It isn't what Intel gives up now by selling this business that will hurt their innovation. It is the options that suddenly get a lot more expensive to investigate in the future for future innovation. Technology innovation is evolution: extremely messy, more dead ends than successful adaptations, enormously energy-intensive from lots of majority-repetition, and highly organic. You can certainly optimize around the edges to make it slightly less jagged-edge, but innovation sparks at the tiny interstices, usually not at the top.
Eventually that money gets recycled into funding earlier stage companies. (If a company buys back stock, the old investors have to put the money somewhere)
The US still has Apple, Nvidia, Tesla, etc. All doing world-leading stuff. I would say US tech is far from dead.
this is pretty unlikely. big government projects (.mil, .nasa) are better at kick starting the hard stuff.
america has gotten so fragmented though that I doubt we'll ever have a moon-shot scale project ever again.
VCs won't do that, it's not their business model. The only way out is government funding - the way it has been for decades.
In China the state pushes for own hardware and they dont care about patents.
What's a damn shame is Intel's StrongARM / XScale division in the early-to-mid-2000s. Methinks the world would be very different today if they continued investment in that sector - and we'd have more than just Qualcomm and Apple to contend with.
The sad bit is, it will lead to consolidation at the hand of few players which would be worse for customers. Intel used to have good products in flash space.
All of that is ignoring the fact that there is absolutely a benefit for Intel to have vertical integration in this area. They make countless products that rely on flash storage - from NICs to switches. The second they have to rely on an outside party they have to pay markup for that company's margin and they're at that company's mercy from an R&D perspective.
This whole thing feels penny wise, pound foolish to me.
I'm sure they still find some use for their no-longer-brand-new fabs, but this kind of examination of each business line in isolation doesn't take into account synergies like that. I can easily imagine Intel doing one "makes good business sense" decision after another, until they are no longer able to manufacture competitively at all.
Oh, wait, maybe that already happened a couple years ago...
Personally I believe Intel needs some big wins as they're basically getting their lunch eaten from all sides. Fortunately they're still highly profitable which affords them a lot of opportunity for investments, if they can't do that their future may be dependent on others successes of their intelcapital.com arm, which worked well for Yahoo! who at the end, most of their value came from their Alibaba investment.
[1] https://newsroom.intel.com/news-releases/nand-memory-news/
>Although, not so clear now whether it means they're on another course to greatness...
This is the $64,000 question. Intel got out of the memory business when it was clear by 1985 that CPUs were its future. I didn't write x86 because Intel itself wasn't convinced of this; see iAPX 432, i860, Itanium, and StrongARM/Xscale. All failed to give Intel the backup/alternative to x86 that it has repeatedly sought.
It is not an exaggeration to say that Intel has failed, or at least failed to distinguish itself, in every single market it has entered since IBM chose the 8088 forty years ago. Besides the above alternative instructional sets/architectures and flash memory, antivirus (still perhaps the most mystifying move in Intel history), servers/motherboards, and GPUs come to mind. The most successful non-x86 business for Intel is ... Ethernet cards? Meanwhile, it's embarrassed itself with discrete GPUs at least three separate times. A few years ago buying Nvidia would have been a savvy move; now it can no longer do so, and it's not impossible to imagine Nvidia buying Intel in the future.
Even within the one market Intel does dominate, what saved the company 15 years ago amidst Itanic and Pentium 4 was 1) AMD Thunderbird being just as inefficient as Pentium 4 and, more importantly, 2) an Intel Israel skunk works project to improve on the Pentium 3. There is no such out-of-the-blue miracle this time.
Of all of the others, StrongARM/Xscale seems obviously the biggest missed opportunity. Threw it away one year before the iPhone.