My memory is that Intel bought StrongARM from DEC as an eventual replacement for the i860/i960 line which was sold for controller and real-time usage. The volumes and margins never matched Intel’s PC business, but did give Intel sales a broader footprint. Intel designed XScale as a follow-on, better targeted at volume applications like consumer, PDA, and comms, and eventually smartphones (a few big wins in RIM/Blackberries). Seems like that wasn’t enough - I’m guessing each new socket was still a painful battle against Qualcomm, Freescale, etc., with volume and margins that didn’t stack up against the core PC and growing server business which was now under threat, hence the sale to Marvell.
Intel CEO Highlights the Company’s Top Three Mistakes
- Thread starter Daniel Nenni
- Start date
My memory is that Intel bought StrongARM from DEC as an eventual replacement for the i860/i960 line which was sold for controller and real-time usage. The volumes and margins never matched Intel’s PC business, but did give Intel sales a broader footprint. Intel designed XScale as a follow-on, better targeted at volume applications like consumer, PDA, and comms, and eventually smartphones (a few big wins in RIM/Blackberries). Seems like that wasn’t enough - I’m guessing each new socket was still a painful battle against Qualcomm, Freescale, etc., with volume and margins that didn’t stack up against the core PC and growing server business which was now under threat, hence the sale to Marvell.
That's what's on the wikipedia page, but I'm skeptical of the accuracy of that page. No mention of supercomputing, e.g. the i860 Paragon system, VLIW aspects of the i860/960 CPU architecture (yuck), or a very good for its time floating point subsystem. As a colleague of mine once said so eloquently, "I only use wikipedia for subjects I don't know much about." The memory of the tone he used made me chuckle a bit thinking about it.
I would a agree that the capabilities were very different between i860/i960 -> StrongARM -> XScale, but the role wasn't - give Intel other products to sell in volume for other apps beyond PC and eventually server, ostensibly to help amortize process development and fab costs. But Intel never evolved or discovered a killer market for their "second processor architecture" that looked good enough to keep the continued investment going and instead, drifted along trying different things, until AMD heated up the central race in PC and server, and they pulled the plug.
MKWVentures
Moderator
I worked on the Xscale line for multiple years. At a couple points the worlds leading smart device was running Intel mobile processors (Blackberry, Treo). The divestment had nothing to do with PCs or AMD. The Atom program followed this and was its own problem. message me and we can discuss how Intel spent BILLIONS on mobile before the Iphone was even planned. FYI Graphics issues were and are the same root cause. It has nothing to do with being to focused on PCs or x86.
IMakeICsAtHomeJK
New member
If Intel never failed on those above plans, then they wouldn't have been a company taking on enough risk in the first place.
To directly answer the question though: (i) GPGPU could have made a difference with regard to targeting 'casual' consumers of graphics-centered software; (ii) Intel's focus on manufacturing "top-level" semiconductors undoubtedly produced internal distraction; and (iii) Intel choosing not to orient themselves towards the growth of mobile phones still leaves the question of whether or not they would have been competitive against Qualcomm, Samsung, et alia.
That story tracks. I agree that the mythological ending will likely superimpose upon Intel's.
I have an iMac (Intel, 2015), Macbook (Intel, 2015), and two identical HP laptops (Intel, 2022) for GNU/Linux. The chips are awesome. But, there is Apple Silicon now. Intel may still continue to produce more laptop-appropriate energy-efficient chips than AMD, however, if I was building a GNU/Linux-machine tomorrow, it would be running Ryzen for sure.
These are not failures because of lack of effort on Intel's part. They were soundly beaten by competitors who were better equipped to deliver in markets like mobile and GPU.
Intel is looking inwards here and saying "If only we had been more committed to mobile or GPU" ignoring the fact that companies like Qualcomm and NVidia had built up wide moats in those industries where it's not clear to me that there would have been a way for Intel to be successful regardless of how much effort they applied.
I also want to add that if Intel turns around it's because PG forced the company to look outwards and is saying "TSMC is better at doing a lot of things then we are, so we should use them for those things" and "other companies make great chips, maybe some of those chips could made better combined with some of our technology". This is how an ecosystem works, not everyone succeeds at everything but companies end up specializing in different areas of expertise which can be combined into great products.
Intel is looking inwards here and saying "If only we had been more committed to mobile or GPU" ignoring the fact that companies like Qualcomm and NVidia had built up wide moats in those industries where it's not clear to me that there would have been a way for Intel to be successful regardless of how much effort they applied.
I also want to add that if Intel turns around it's because PG forced the company to look outwards and is saying "TSMC is better at doing a lot of things then we are, so we should use them for those things" and "other companies make great chips, maybe some of those chips could made better combined with some of our technology". This is how an ecosystem works, not everyone succeeds at everything but companies end up specializing in different areas of expertise which can be combined into great products.
Last edited:
"We ended up not winning it or passing on it, depending on how you want to view it. And the world would have been a lot different if we'd done it," Otellini told me in a two-hour conversation during his last month at Intel. "The thing you have to remember is that this was before the iPhone was introduced and no one knew what the iPhone would do... At the end of the day, there was a chip that they were interested in that they wanted to pay a certain price for and not a nickel more and that price was below our forecasted cost. I couldn't see it. It wasn't one of these things you can make up on volume. And in hindsight, the forecasted cost was wrong and the volume was 100x what anyone thought."
It was the only moment I heard regret slip into Otellini's voice during the several hours of conversations I had with him. "The lesson I took away from that was, while we like to speak with data around here, so many times in my career I've ended up making decisions with my gut, and I should have followed my gut," he said. "My gut told me to say yes."
After Paul Otellini: Intel Lost the iPhone Battle, But It Could Win the Mobile War
Paul Otellini shows off the chip-circuitry lining of the coat hanging in his Intel cubicle as an iconic bunny man keeps watch (Alexis Madrigal). Forty-five years after Intel was founded by Silicon Valley ...
Intel was willing to be a loss leader in their first couple years of NAND development. I don't know if X-point was ever profitable. And famously intel lost billions on their most recent/biggest mobile push in a textbook example of too little too late. They didn't have the foresight to see that a product like the iphone would be huge and let it slip between their fingers. Granted if our timeline repeated, Apple would have definitely bailed due to the 10nm and 7nm delays. But before that time intel would have had wafer volumes from apple that would have only been surpassed by core and xeon. Who knows maybe in that timeline IDM 2.0 would be an easier swap and Apple would be planning to come back to intel.
Last edited:
It's not a matter of never failing, it's a matter of not being able to see any significant wins from programs other than x86 CPUs.
This comes down to competition for company resources. x86 CPUs are the most profitable product, when you are allocating limited resources do you put them into making products with a 50% gross margin that are going to sell in high volume, or do you put them into products with 5-10% gross margin and you don't really know what the volume is going to end up being? When times are good there is lots of money to spend on new products, ventures, and acquisitions, but then whenever there is a recession and a company is forced into austerity it's a different story.
Even today, Intel is being forced to exit a lot of businesses. Maybe one of these markets that was exited will suddenly take off and become the next big thing and PG's successor will look back and say something like "If only we didn't get out of networking chips"
There's probably no such thing as a 10% gross margin product at Intel, even these days. The program would never get funding in the first place. The problems I've seen in big companies, including Intel, that usually result in failed initiatives are senior management changes which cause a change in priorities, politics (it was so-and-so's program and they've moved on), unreasonable expectations about how quickly a new program can ramp and gain market share (especially chip development) or with insufficient budget, trying to break into new markets with only incrementally better products, and appointing engineering managers who are not really qualified to do their jobs. In other words, all of this is stuff that's done when the company is not fully motivated to succeed in a new market. Expanding into a new market takes a strong sense of urgency. The bigger the company, the more urgency that's required to succeed. Oh I forgot one... acquiring a start-up that is a small team which operates with lightweight development processes and then bringing it into Intel and forcing the team to use Intel's development processes. (This last point is true of so many big companies.) That last one was for you, @cliff.
Agreed. As for networking, the Barefoot Networks case is very odd, even for Intel. What a mess.
Intel has committed this sin many times, without seemingly learning anything from the string of failures.
Today I think Intel is looking at TSMC margins and thinking they can do the same. Probably not, I'd be surprised if IFS ever makes money as a stand alone BU. I think Intel's foray into becoming a foundry will fail, but ultimately that failure will lead them to do what they should have done a long time ago - exit the foundry business altogether and become a fabless company.
Thank you.
Agreed.
I'm on a different page. I think IFS can make money, though probably not margins as fat as TSMC's. TSMC took years to get where they are, and the vast majority of their capacity is in a relatively low cost geo, obviously. On the other hand, IFS has the advantage of several in-house high volume and high margin projects that will be biased towards choosing IFS, if IFS can meet the minimum technology, quality, and volume requirements. Assuming Gelsinger stays CEO for another five years or more (which is likely), that's probably enough time for IFS to get established and profitable, though not really a TSMC peer. Of course, no one is a TSMC peer.