Intel and the Danger of Integration

The thing is, though, that tying design and manufacturing together has been really good for Intel. Historically Intel's performance in terms of driver current and so forth at a given process node has been better than its competitors and it has reached those nodes more quickly. Intel manufacturing paid for that lead partially by accepting more restrictive design rules for transistor layout than a merchant silicon shop could ever get away with because they knew they only had one customer and one design they had to make work. With Intel's numerous missteps at 10nm that lead is over and the manufacturing debacle is taking down the design too but there's no reason to think Intel would be starting from such a dominant place if they had split up. If I were to go back to 2013 I'd tell them "10 nm is too early for cobalt interconnets!" rather than "Split up your design and manufacturing".

And really you could also look at the problem as the design team being too wedded to x86 and unwilling to consider other ISAs just as much as the manufacturing problem. Intel could have done an ARM processor for the iPhone but they wanted to make a go at an x86 mobile chip instead. And while the x86 tax is small potatoes when you're paying for a deep OoO pipeline it's pretty noticeable in the world of in order processors or graphics processing units.

I understand the point that Ben Thompson is making: that a vertically integrated Intel lost out in the mobile revolution because it wouldn't unbundle it's manufacturing from its processor design. But its current crisis (not referring to the CEO's affair) has nothing to do with vertical integration. It has everything to do with Moore's law scaling coming to an end. This gives fabs like TSMC and increasingly SMIC from China the ability to catch up. Unless there is a breakthrough in technology -- molecular transistors or some such, Intel's ability to dominate will end soon.

Good article, but it misses plenty.

One has been marketing that damages its own brand, particularly the "Atom" phenomenon. Intel had such an obsession with phones that it just had to make x86 processors that were too weak for phones, never mind anything else, full of bugs, etc.

It might be a classic case of "boiled frog" because few have realized that ever since Intel switched to PCIe, Intel chips have been starved for PCIe lanes. There has been no point to SLI in a consumer configuration because there just is not enough I/O bandwidth to support two graphics cards. In fact, Intel hopes you will settle on "integrated graphics" (the same as on a phone) so you won't have a reason to buy a computer instead of a phone.

Optane is a story like 10 nm. Just add tone-deaf marketing that makes a potentially revolutionary product seem like nothing.

Probably the thing that has hurt the PC industry the most is the slow transition to SSD. Given a choice to a brand new computer that has an i7 chip and an HDD which will show you a spinning cursor most of the time, or an eight year old computer with a new SSD which can boot before you get old, the choice is obvious.

If Intel had said something like "an i5 chip has to come with an SSD" or if Microsoft had required SSDs for the Windows 8 launch people would be like "Wow this is much better than what I had before." As it is, Intel has mandated a Meh experience for a long time and the tech press has let them get away with it. (It's murdered the tech press too since now all Tom's Hardware talks about is RGB lighted fans and the really cool monitors that might get released someday...)

The problem with Intel, as I see it, is an inability to get new designs to production, and a low tolerance for market growth versus market sustainment. They’ve been a disaster at acquisitions, being totally fine with wasting billions on failed outside investments without any program for internal ventures. The latest focus on “Intel is now a data company” is just complete denial of the reality.

But the margins are awesome, and I’d reject any notion that they should erode that with fab services. They need more of the high-volume, high-margin Intel-designed chips that have been their hallmark. Everything else is a distraction.

Some of that depends on identifying new forms of computing. Therefore, it makes sense to do technology experiments, but lately Intel has focused on market experiments, where they are not really developing new technology, but building internal organizations with sales, marketing, and product management that dwarf engineering.

This is a problem not just because it diverts resources, but because nobody really knows what Intel is anymore. Engineers don’t feel like they are working in an engineering company anymore.

Intel’s margins are great, and despite the negative prognostication, their market position is still dominant. Even the engineering talent at Intel is still some of the best, although the management is somewhat deliberately not their best representation. They Just need to focus on being an engineering company again, ignore sales, ignore acquisitions, and invest in their own people to do the experiments that will lead to the next ubiquitous high-margin chip.

I really think this article fails to make its point about what exactly intel stands to gain from say, manufacturing processors for Apple. It trades in its major competitive advantage, advanced processes (dispite the propaganda intel is far from out of the game process advantage wise) coupled tightly with its processor designs that ensure margins that would make most companies blush, for low margin hypercompetitive contract work.

Intel is under absolutely no real threat of being unseated as the company that powers virtually every server in existence, throwing that away to chase contract mfging work dosent make any sense, even if some other companies do it well. The arguement it makes would be more like intel going back into DRAM production, and abandoning x86.

It is not like Intel has not repeatedly tried.

Atom is their attempt. As are more advanced MCP51 micros.

Intel's problem is that their CPUs were performance focused and as such didn't scale down as well - plus vendors didn't have to care as OSes were portable enough while software has to be rewritten anyway, so going Intel brought no compatibility advantage. This is the same reason why Microsoft has trouble penetrating mobile market.

On one hand, this article seems to make a lot of sense. OTOH, INTC is up 40% in the past year. It seems likely that there are quite a few parties out there who probably don't agree with this analysis.

Intel's biggest threat is the decline of WinTel, not its inability at making stuff for consumer products.

Intel has legions of top tier IC designers, they can make Apple A10 or Qualcomm 845 class chips as weekend projects. What they don't have is determination to abandon the proven WinTel model business, which is still an extreme cash cow.

Intel embracing RISC-V would be a super-bold move. Given the support we can already see from the long list of RISC-V foundation members, it's bound to succeed. In the next couple of years it will probably wrestle a few royalties away from ARM.

Beyond that, I think it will end up a legit heavy-duty apps processor for Android phones. It might start out on the mid/low-end phones, but manufacturers will absolutely love the idea of cutting out ARM from their BOM.

I don't think, right now Integration is something wrong about Intel. It is how they miscalculate every thing every time for the past years, even before BK. Planning / Vision, and Execution, I am not sure if there are anything else other then these two attribute that is important to a company, and Intel lacked both.

Intel used to win because they have an effectively monopoly in PC market. The high price, high margin and high volume CPU drove investment in tech and their Fab. That was possibly the only item I know other then iPhone that had those three factors together. When the Smartphone revolution started, everything changed. The scale changed, TSMC now produce possibly close to 5x volume then Intel do at Intel Fabs, and their leading node 7nm will likely reach a similar quantity to Intel's 14nm in 12 months time.

TSMC now has 2X+ the Total addressable market in leading node compared to Intel. So while new leading node investment cost continue to increase, TSMC also had increasing market size to help spread itself. If you note Intel's mainstream CPU die size, the trend has been going downwards over recent years, i.e They are optimising for cost.

Intel could also have expanded their Scale. They could have kept their integration and make more chip, expanding their Fab advantage in both scale and technology. Although they did but it was either late or taken far too long. Their GPU is only looking to launch in 2020, as a first gen product I doubt they will make much an impact and volume. Imagine they had a dGPU prepared 4 years ago, when the crypto wave hit, they ride the wave and gain money earned from Crypto, or they would have a dGPU for a market prices that is cheaper then what ever Nvidia or AMD's bumped up prices. Their 14nm process would actually be miles ahead of 28nm or 16nm that those GPU had.

It took them 8 years, 8 years! from acquiring Infineon modem to actually produce a modem in their own Fabs. Next time some M&A talks about synergy one should ask them, how long does it take. Now they have the contract for iPhone modem and could produce it themselves instead of TSMC, they face a new problem that no one has asked so far. How is Intel going find additional capacity to produce an additional 100M modem for Apple over the next year? My guess, the original plan of Intel was the move to the much delayed 10nm will finally be fixed in 2018, leaving some room for the Modem. The modem is much smaller in size, so it is more like 40M unit to Intel's median chip size. But that is still roughly 20% additional unit Intel has to chip on their leading node.

And if you had read some rumours about Intel delaying, or renaming their chipset originally going from 22nm to 14nm. That is a possible reason why. May be they shouldn't have defer Fab 42's construction in the first place.

I am pretty sure BK's departure has more to do with his performance then whatever they wrote in that PR. And may likely has something to do with Apple. I wouldn't be surprised if Apple has been unhappy about Intel's lying and execution. Remember when Apple started to design their "thinner" MacBook and MacBook Pro, it was obvious they had 10nm and 7nm Intel's CPU in mind. First was suppose to happen last year and the latter schedule for next year. This two year delay, which Apple may or may not know, put Apple in an awkward position. Their custom foundry, were suppose to attract Apple to built an integrated SoC with Intel Modem inside, which is now even less likely to happen when Intel has not been as open about its problem. When Apple contracted Morris in TSMC, they managed to bring up a new Fab running for Apple in less then 6 months. It was a big bet on both side, billions of dollar invested. Could you imagine Intel doing that? These relationship takes time to built, and TSMC right now has literally zero fault.

I do agree Intel now faces another turning point, last time they had Andy Grove, now I am not sure who they have hire to get this behemoth running. The task is enormous, and lots of Intel's best engineers and executive have left the company. The only one who may have a slim chance to completely transform Intel is Pat Gelsinger, if Andy Grove saved Intel last time, it will be his apprentice to save Intel again. Unfortunately given what Intel has done to Pat during his last tenure, I am not sure if he is willing to pick up the job, especially the board's Chairman is Bryant, not sure how well they go together. But we know Pat still loves Intel, and I know a lot of us miss Pat.

Well Microsoft is about to be fully integrated if they make their own CPUs for their PCs like Apple is about to do (and sorta like how they designed their own XBox processors)... Let's see how that plays out for them and if this anti-vertical-integration theory holds.

Does this mean we can have more than 16GB in our laptops now?

Also, is "stratechery" named based on the old Celebrity Jeopardy skits on SNL?

i love the article's general quotation about disruption:

... what makes disruption so devastating is the fact that, absent a crisis, it is almost impossible to avoid. Managers are paid to leverage their advantages, not destroy them ...

as other firms innovate, those advantages become obsolete practices so leveraging them means taking on more risk, not less risk.

Does anybody know a good technical history of Intel?

I teas “Inside Intel” and there was some good stuff but seemed too superficial.

I think the most common issue would be neglecting constant values.

> it was already clear that arguably the most important company in Silicon Valley’s history was in trouble: PCs, long Intel’s chief money-maker, were in decline, leaving the company ever more reliant on the sale of high-end chips to data centers;

Where PCs were replaced, they were succeeded by laptops, and for the most part they use Intel processors.

I’m not so sure that not choosing to produce ARM based chips for the iPhone was such an obvious miss for Intel. Why would they want their business subject to the whims of a single company over which they exert zero control?

Just as an example, a couple of months ago Apple seemingly out of the blue, announced they would no longer be using Intel chips in future laptop models and the markets barely noticed. Were they more reliant on PC sales that would not have been the case.

Perhaps they are way behind AMD with respect to the latest chip designs, but even that is far from proven.

I think this article is attempting to make a company with a very complex history and product lineup sound simple.