ARM Disrupting Intel with its Business Model | Hacker News Reader

ARM Disrupting Intel with its Business Model

fernstrategy.com

20 pointsmjfern22 comments

This seems to me a case of mis-attribution of success.

The article says "Intel’s dominance in the global microprocessor market is being threatened by the rapid rise of ARM Holdings.".

ARM Holdings might be "rising rapidly" today but it was founded in 1990. They've been doing ARM processors with the exact same business model for the past 20 years.

The reason they're successful today is simply because they have superior technology in an area (power consumption) that is important for a quickly growing mobile markets.

Arguably ARM was a better architecture than x86 from the beginning but it didn't matter: after Windows/x86 got majority of the market share, the switching cost for a different processor were too big and Intel was doing a good job driving x86 in ways that mattered for that market (performance).

Only when a completely new market emerged (for PDAs, iPods and cell-phones) that was free from requirement to support legacy code, ARM found it's place because it delivered one thing much better than Intel with their existing line of x86 processors: power consumption was dramatically lower. It also happens that this was a core requirement for mobile devices.

We can't go back in time and see how ARM Holdings would hold up if they were doing their own manufacturing as well, but in my opinion the business model has a minor impact on their current success.

The most important thing is that they have a product that is superior to competition.

In the article, I discuss the importance of ARM's technological advantage over Intel in the area of energy efficiency, especially for portable devices (see paragraph 2). My key point is that even if Intel can match (or beat) ARM from a technical standpoint (e.g., clock speed and energy efficiency), the company may still get disrupted due to ARM's unique business model. ARM's business model provides a benefit for OEM customers in the areas of cost, customization, and supplier-customer relations.

I recognize that it's not just ARM's business model that poses a threat; the threat stems from a combination of ARM's business model, its current technical advantage (in energy efficiency), and trends in computing, with a shift away from PCs (and the Wintel standard) to connected devices running Android and iOS. To your point, the shift away from the Wintel standard has significantly reduced the barriers (i.e., backward compatibility) to the adoption of the ARM architecture for OEMs.

(Note, ARM's recent "rapid rise" in the processor market is illustrated, in part, by the company's stock market performance, with a market cap that has grown about 5x over the last two years.)

I wonder if "disruptive" is really the right term here. As you point out, Intel dominates in market applications such as servers, desktop PCs, and netbooks -- and it's not at all clear to me that Intel's business there is threatened in any way by ARM chips or business model.

Where ARM has succeeded is in new markets (phones and tablets) that both rely on low-power operation and are relatively immune to legacy code or chip architecture issues. As far as I'm aware, Intel's never been successful in those spaces and even took the strategic step of dumping its ARM (Xscale) operation.

My point being, if Intel's secure in established markets and uncompetitive in newer markets, that's not a technological disruption in the classic Christenson sense. It's merely the consequence of earlier strategic decisions to not go after newer low-power application markets. Where would Intel be, for example, if in the early days they had dumped the microprocessor business to focus on their core business of memory chips?

edit: typo

The risk for Intel (and x86) is that ARM-based processors appear to be moving up-market into servers and datacenters, a sign of technology disruption.

- Calxeda (formerly Smooth-stone) is developing servers based on ARM-based technology. (http://www.calxeda.com/)

- In November 2010 it was revealed that Dell is prototyping ARM-based Cortex-A9 servers.

- In January 2011, Nvidia announced Project Denver, with "plans to build high-performance ARM based CPU cores, designed to support future products ranging from personal computers and servers to workstations and supercomputers."

- Microsoft recently announced that Windows 8 will support both x86 and ARM architectures.

- In a February 2011 analysts presentation, Warren East, CEO of ARM Holdings, suggested that ARM is seeking to expand into the PC and server market: "There's a blurring between computers and smartphones, and ARM's success in smartphones is helping us get into computing...Cortex A processors support multiprocessing and that delivers the high level of performance required by server applications." (http://www.theregister.co.uk/2011/02/01/arm_holdings_q4_2010...).

In October 2010, I posted a related article, "The End of x86?," that examines the potential disruption of x86 by ARM using Christensen's disruptive technology framework: http://www.fernstrategy.com/2010/10/21/the-end-of-x86/

Hmm... Calxeda and Dell are at the prototype stage and haven't shipped yet. In the Register article, ARM admits they have "0%" market share in PCs and servers. So the data for ARM in those markets essentially boils down to "boy, those markets would sure be profitable" and "just wait until 2020!"

There's a lot more to the server market than power savings. The Register article also notes: "It is a pity that ARM is not moving the quad-core, 40-bit Cortex-A15 to market faster and talking about either 64-bit kickers or how it will manage virtualized 32-bit applications on multi-core ARM servers such that no one cares that they are 32-bit applications."

As to Microsoft, I can remember Windows NT shipping for DEC Alpha, MIPS, and Power PC. Just because Windows 8 will work on ARM-based netbooks doesn't justify extrapolation to the PC and server markets. And even if Microsoft did (theoretically) support ARM in those markets, well, they actually shipped NT for Alpha, MIPS, and Power PC, and where are those architectures now?

My read of history is that if ARM ever became a real threat to Intel in PCs and servers, the company would simply lower profit margins in those markets to starve the competition. But hey, maybe this time things really will be different.

The reason they're successful today is simply because they have superior technology in an area (power consumption) that is important for a quickly growing mobile markets.

But lets not forget ARM has been successful for a long time in this space. It's not like when the iPhone came out then ARM started selling. From Wikipedia,

"ARM licensed about 1.6 billion cores in 2005. In 2005, about 1 billion ARM cores went into mobile phones.As of January 2008, over 10 billion ARM cores have been built, and in 2008 iSuppli predicted that by 2011, 5 billion ARM cores will be shipping per year.[8] As of January 2011, ARM states that over 15 billion ARM processors have shipped."

So two years before the iPhone they were licensing more than a billion cores per year. That's more cores than Intel sold this year, five years later.

ARM is more high profile now though because smartphones are more high profile, but ARM has been doing this for a while now.

The Newton MessagePad (circa 1993) used ARM and StrongARM cores.

This analysis is pretty superficial.

Yes, ARM is ahead on low-power devices, and yes Intel can (and probably will) catch up, especially as ARM moves towards more powerful processors (dual core, etc). This article did acknowledge that at least, which most seem to miss.

It does go into a few of the advantages of the licencing model ("this model enables OEMs to customize integrated chips that conform to different form factors", etc), but it didn't discuss the disadvantages this model brings.

For example, Intel has a pretty significant lead in building chip foundries. AMD hasn't been able to compete successfully at the high end because of that, and the ARM licencing model doesn't generate the profits needed to go head-to-head with Intel in foundry and high end research.

This is a real problem for ARM. To quote Ars:

First, there's simply no way that any ARM CPU vendor, NVIDIA included, will even approach Intel's desktop and server x86 parts in terms of raw performance any time in the next five years, and probably not in this decade. Intel will retain its process leadership, and Xeon will retain the CPU performance crown. Per-thread performance is a very, very hard problem to solve, and Intel is the hands-down leader here. The ARM enthusiasm on this front among pundits and analysts is way overblown—you don't just sprinkle magic out-of-order pixie dust on a mobile phone CPU core and turn it into a Core i3, i5, or Xeon competitor. People who expect to see a classic processor performance shoot-out in which some multicore ARM chip spanks a Xeon are going to be disappointed for the foreseeable future.

It's also the case that as ARM moves up the performance ladder, it will necessarily start to drop in terms of power efficiency. Again, there is no magic pixie dust here, and the impact of the ISA alone on power consumption in processors that draw many tens of watts is negligible. A multicore ARM chip and a multicore Xeon chip that give similar performance on compute-intensive workloads will have similar power profiles; to believe otherwise is to believe in magical little ARM performance elves.

http://arstechnica.com/business/news/2011/02/nvidia-30-and-t... (read this whole article - it's very good).

That leaves the low-to-midrange to ARM. As has already been noted, Intel is quickly becoming power-competitive there, and is very, very good at producing chips cheaply.

I'm not saying ARM isn't an important competitor, but I am saying the whole "ARM will eat Intel from below" narrative is naive. It's going to be a fierce competition, and it's not at all clear that ARM's business model really is an advantage at all.

Meanwhile, Intel can hedge it's bets and play at the licencing game, too. For example, they have experimented with licencing Atom manufacture to TSMC (the same ARM foundry mentioned in the article): http://www.pcworld.com/article/160496/intel_opens_up_the_ato...

Might also be interesting to add that Intel had a very nice ARM platform (XScale) which was sold to Marvell. They had the full ARM license deal, not just the ability to put ARM cores on a micro like everyone else in the world.

So I would argue that Intel has a pretty good understanding of ARM strengths and their business model, they just think that they can go the distance and come up with something competitive. Even given the fantastic growth of ARM, Intel is about 100 times bigger so it's probably more useful to compare Intel to the processor manufacturers that use ARMs in mobile platforms.

There's a danger that Intel will continue to chase high profit sectors and stagnate technically due to lack of competition there. The gap between what is technically achievable and what we get today in x86 processors has been widening for years and is not slowing down.

That's a general problem with large companies, they often can't afford to pivot and succeed in lower profit sectors, even if its clear that the high profit sectors are going to stagnate and shrink over time.

Exactly. What they typically do is acquire someone who can. In Intel's case I'd doubt they would acquire ARM because they would probably have to continue to license the technology to everyone else (I don't really know what I'm talking about here). However, they could acquire someone who has the "low power CPU technology" they would need to compete. If you think about who that someone could be nobody really comes to mind. Maybe they could buy back Marvell. Qualcomm, Samsung, TI are clearly out of the question.

> However, they could acquire someone who has the "low power CPU technology"

The other possibility is that Atom will slowly become more competitive. Weirdly, I think the most important thing for that would be a solid Android-on-x86 story, so device manufacturers don't have to do so much custom GUI work.

I think the key point here is that Intel is one company while ARM is many. All feed each other improving competing designs, proving technologies and exploring their markets in ways a single company could not do with volumes that dwarf (and have been consistently dwarfing) the PC market.

As humongous as it is, Intel simply can't compete against the combined power of ARM plus its myriad of licensees. Eventually, from the diversity of the ARM ecosystem would emerge a competitor for Intel's (and AMD's) x86 lineup. Once set in motion, it was a matter of time.

Nice article. To add, I think the problem is with intel's strategy and their technology than the cost.

Intel expected notebooks to be more dominant and their roadmap showed. As soon as Kindle started taking off with ebooks, Apple realized they could do better than Kindle with iPad.

By selling xscale unit to Marvell, Intel did a mistake.

Total OEM cost for netbook was in the range of 10-20 vs just the processor in ipad is claimed to be $30 (after everyone taking their cut. Everyone includes ARM-Samsung-Packaging compnies)

This is just a small point, but the article says that Intel will "close the energy efficiency gap with ARM" in 2011-2012 with the release of the latest Atom processor. What it overlooks is the fact that it won't be competing with current ARM chips. Nvidia and Samsung and TI aren't going to just sit there waiting for Intel to catch up; they'll be improving their designs and moving to smaller fabrication, just like Intel is doing.

Real disruption won't come until ARM create (edit: design) a real general purpose, mainstream 64-bit processor.

ARM holdings licenses. I think Nvidia may do what you're talking about.

Isn't that what he mean? ARM would design, and license, a general purpose processor.

Perhaps. He's changed his post since I replied.

Unless the "general purpose processor" market is disrupted itself. It might turn into a niche. Maybe SoCs are the future. (64 bit is in the works though)

it is absolutely not clear that separation of "general architecture designer", "specific chip design customizer" and "foundry" is a winning business model. In my view Intel has been successful because they've owned all three pieces and thus have ripped the benefits from, sorry for the profanity, the synergy.

Intel don't have competitive product in that specific market that ARM is so successful. Once/if it gets it, will see then. I'm not saying that it will win - it can go in any direction, but we'll be really able to compare Apples to Apples then.

Reply on news.ycombinator.com