Otellini predicts ARM's Transmeta-like fate
Intel's Paul Otellini claims that ARM has no chance in the fight for desktop and server market share, predicting a slow demise like Transmeta and its Crusoe chips.
Speaking to AllThingsD, Otellini rubbished ARM's long-term chances in the PC and server market. Conjuring up the names VIA Technologies and Transmeta, low-power chip specialists whose market shares sat at around 0.1 per cent until the latter folded in 2009, Otellini suggested that ARM's assault on the desktop, laptop and server markets is over before it's even begun.
'I happen to be around long enough to remember [VIA and Transmeta],' Otellini told AllThingsD. 'People come and go, and we’ve never had an exclusive, if you will. And, overall, the best chip has won.'
In addition to comparing ARM's chances against his company to those of VIA and Transmeta, Otellini took the opportunity of the interview to badmouth Microsoft's Windows RT - the Windows 8 offshoot aimed at tablets running on ARM architecture system-on-chip (SoC) processors - by bringing up the well-established lack of compatibility for legacy x86 applications.
Otellini's comments, which are hardly the first Intel's CEO has made against both Windows RT and ARM itself, reveal a company which is eyeing its Cambridge low-power rival with concern - and well it might: despite Otellini's assurance that x86 and x86-64 are the One True Architecture, interest in ARM outside the mobile ecosystem is growing. Google has recently launched an ARM-based Chromebook laptop manufactured by Samsung, which sold out within a single day of being listed for purchase, while AMD's SeaMicro subsidiary is expected to announce a new many-core ARM-based microserver early next week.
The only market left for ARM - or, to be precise, its multitudinous licensees - to poke its nose into is the desktop, where it has been relegated to niche products like the RISC OS-based A9home. With growing interest in low-cost development boards like the Raspberry Pi, however, that could well change - and then Intel would have real cause for concern at the lucrative budget end of the market.
Previous Article
22 Comments
Discuss in the forums ReplyBack in the mists of time - OK, it was the 1980s - two companies ruled the roost in the UK home microcomputer market: Sinclair and Acorn. Acorn had got where it was by producing mid-priced but very powerful computers designed by engineers; Sinclair by producing bargain-basement and significantly less powerful systems designed by accountants. Both filled a niche, and both were pretty much obsessed with wiping each other out. (As an aside, there's a good reason for that: Chris Curry quit Sinclair when Uncle Clive wouldn't support his MK14 microcomputer project, leaving to found Acorn with Hermann Hauser.)
That obsession basically killed both companies: Sinclair decided to attack Acorn by going up-market with higher-price and more powerful devices, culminating in the incredibly unsuccessful Sinclair QL, or Quantum Leap; at the same time, Acorn went downmarket with the Electron, aiming to knock Sinclair's Spectrum off the bargain-basement shelves. Both companies failed miserably, and when the Great Microcomputer Crash hit went bankrupt.
Now, to be fair, there were other factors at play here: Acorn had overestimated demand for the Electron, and ended up with warehouses full of 'em and no capital; Sinclair ploughed millions of his company's cash into the C5 electric vehicle project, an abject failure that sold fewer than 12,000 units. The slump in the market, of course, affected all companies equally. Nevertheless, it seems likely that had Acorn stuck to its roots and likewise Sinclair, the two companies would have been in a better position to weather the financial storm that was to follow.
Why the history lesson? ARM and Intel are in positions very much like Sinclair and Acorn: ARM produces low-cost low-power chips built to a definite budget, while Intel offers higher-power and higher-priced parts for users who require extra oomph. Both are market leaders in their own fields, and both are attempting to muscle in on each others' markets with a dramatic change in tactics: ARM is going upmarket with the Cortex-A15, adding in features like virtualisation extensions that take up valuable silicon and increase the cost of the chips, while Intel is going down-market with its low-power Atom system-on-chip designs.
ARM (a company born from the ashes of Acorn, funnily enough, with a name that was once an acronym for Acorn RISC Machines) isn't wasting money like Sinclair did, however, and Intel has enough money to spare that if it wants to it could just go ahead and *buy* ARM - but the similarities between the two battles are striking.
Hmm. Should have pitched that as a feature, really. Bah!
Still worth a punt, you could flesh it out a bit more and stick it up as a feature. There was a TV documentary that covered this a few years back now that was interesting, but most of the non-UK readers won't have seen it, (and probably a fair few UK readers missed it as well). Interesting stuff, and there are parallels to be drawn.... (Though I'm sure both current companies would deny it vociferously!)
P.S. i dont like intel much but i recognise how good they are at turning sand into money.
Thats the 2 main players in the mobile phone space at this moment, Where arm is prevailent, Getting a chip for tablet implementation is probably easier due to the battery not been as big as an issue.
Arm is basically a licence company as it does not have its own FAB, Nor does it see alot of the profits from its chip, If it was in intels position it would be making billions of dollars just from samsung and apple instead tsmc and samsung make the billions.
I think that about sums it up. I think he's scared and trying to show some bravado.
My own hunch is that ARM's business will slow down (well to be honest, it's low margin SoC designs anyways, and it basically just makes money from licensing) within the next 5 years. But truth be told, it won't fail. Many more units will be using ARM based components for a long time. Hell, most televisions and monitors use ARM based logic boards.
I'd say we still have at least 5 years of massive growth ahead of us in the mobile market and for after that? well companies like Dell and Sony are currently releasing Ultrabook / Tablet hybrids, if the blurring of the line between pc / mobile continues ARM very much has a chance to take a serious chunk of market share from Intel, who in my opinion have simply been caught napping.
Of course there are also potential problems, like for example the brutal price war currently waging between chinese smartphone manufacturers like Xiaomi, Huawei and so on which is likely to put massive pressure on component pricing, or Intel waking up...
But if ARM can keep putting out competitive products for the next couple years they should be established well enough to be able to survive the inevitable market saturation and the "good enough" effect that will come eventually.
The price wars would harm mostly the manufacturers more than ARM itself as they sell the licenses as far as I know. As long as they keep advancing at a steady pace, ARM is secure. But buying power wise, ARM is puny compared to Intel who have their own fab plants. Market saturation or not, ARM's competition is with the Atom SoCs in the coming years and while Intel has the margin cost covered (R&D and fabrication) the majority of SoCs are still ARM powered. Once market saturation has reached a high, the value of smartphones will lower. It'll be common enough. It won't mean sales will be harder to drive, but lower. Upgrade cycles slower.
Intel hates to admit they're wrong - Intel is a former ARM architecture licensee. They had a joint venture with Marvell, XScale, which they then sold out to Marvell in '06.
If Apple can make what is alleged to be the most kick-ass ARM processor on the market (A6), imagine what Intel could do if they threw a few billion dollars at it.
As to trying to outcompete with x86 to Apple and Samsung, never going to happen. Maybe other OEMs, but not these two (and the rest don't really matter, do they?) What makes Apple's dick hard is control - Apple has had the A6 in the works for years - you don't just pull what appears to be a processor laid out by hand out of your ass - Apple's been working toward this for a good, long while. As for Samsung, Intel can't compete on price. Given that Samsung produces their own Exynos processors that probably cost $5 apiece, how can Intel compete? Samsung will be happy to buy Clovertrail and those that follow, but there's no way they would give up ARM to become dependent on Intel.
Intel needs to muscle their way into mobile, but it's not where they're going to make the bulk of their money. Traditional PCs and servers aren't going away, and we're a good long ways away from ARM having marketshare larger than AMD in the server room, let alone Intel.
I think it's time for Otellini to shut up - he's beginning to sound increasingly desperate, when Intel's position in the market has never been stronger (remember, in their 'weak' quarter, they 'only' had $13bil in revenue, not the $14bil expected.)
It both grants access to i.p. (and excellent i.p. at that) and makes the licensee intimately involved in the evolution of that i.p. It has gone soo sooo far because it is cheap to get into, incredibly *good* at providing low-power processing power and it has put the manufacturers at the very heart of the success of the future of the 'franchise' as a whole.
It's more idea versus a single manufacturer's product, Intel belonging to the latter, with all the freedom of manufacture, form and implementation that provides without the gun-to-your-throat near-total submission to that one all-powerful supplier otherwise entailed.
So it isn't company A versus company B as such, it's one manufacturer versus ....... every other manufacturer and what they have invested in and nurtured up to that point, with much simpler (and cheaper) licensing costs.
Not to mention the hardware itself is exceptionally good at providing processing power at the most miniscule of power consumption levels. Lets not forget that with A15 coming soon, Arm might very well leapfrog Atom, even on one-size-fits-all bulk processes....
Then again, the PowerPC architecture was once chided as "better" than x86. Look at where it is now. Gone from the desktop market, and only in the game console market.
I'll agree with Parsnip mainly because you've been right most of the time anyhow. That said, Intel doesn't really have a vested interest in mobile as it seems to be. I mean thus far they've shown a rather halfhearted effort. Sure the recent Atom chips (in phones) are rather impressive but still, they've taken this long. And sure, one could argue that the improvement of their onboard GPUs means they're moving to the mobile market.
I'm not too sure about that to be honest. It just seems they want to outplay all the Integrated mobile laptop chips. Take out AMD's APUs seems to be more the order of the day than actually making a mass produced mobile phone/tablet SoC. After all, margins on those units are rather low. ARM's just IP anyhow. It's Qualcomm, Samsung, Apple and TI, the manufacturers that Intel would be gunning for if Intel decided it would compete with ARM.
All this said, more mobile units in a cloud means a higher need for servers. And guess what? I'm pretty sure Intel Xeons are going to be selling pretty nicely when the time comes. That and IBM will also be making a pretty penny.