Here comes Intel, again, making a run for the mobile phone market. First time it tried buying start-up telecoms IC companies; second time it started making ARM chips (X-Scale); now, it's going the big company route - a new chip (Atom), a new OS (Moblin), a big acquisition (Wind River), and a big company collaboration (Nokia). Third time lucky?
These are the sorts of things execs can do. What execs can't do is engineer a low-powered, low-cost core. But low-power, low-cost cores are what phones need.
We've seen it all before. Top execs think they can mandate success by deals and strategies but, at the end of the day, the best engineering solution wins.
After the glory of the press conference, what is Nokia going to do for Intel? Nokia exec vp Kai Oistamo wouldn't even confirm that Nokia would put Atom in its phones.
And what will
It looks a bit like execs thrashing around trying to convince the world they have a potent strategy for getting into mobile phones.
Without such a strategy, Intel's reliance on the commoditising PC industry probably means the company will go ex-growth with consequent further punishment of its share price.
It's all a long way from Intel's roots when its top execs thought that making a better chip was the way to be successful.
The problem with Atom is that it represents massive overkill for mobiles, both in processing power and especially power consumption (both active and standby) where ARM (for example) is getting on for an order of magnitude better -- and always will be due to the different chip architectures.
The basic difference is that ARM was designed for phone-type applications from the start (low power and high integration levels with just enough MIPS for the job), Atom is a stripped-down PC CPU with way too many MIPS and far too much power even before you add on any extra chipset needed (even if this is integrated in future).
It's like being given the task of getting one person from A to B through traffic as quickly and economically as possible; ARM is a motorbike, Atom is a Hummer with two wheels removed and stabilisers bolted on...
Nobody wants a mobile phone with fantastic multimedia capabilities if the price is either being the size and weight of a brick or a battery life closer to netbooks than a phone.
Ian
Thanks Ian, sounds spot on to me
Ian: While I agree with most of what you say, your claim of "way too many MIPS" is very puzzling. The 800MHz Atom used in the Atom 4-5" MIDs isn't incredibly faster than the 600MHz Cortex-A8 in the iPhone 3G S or the Palm Pre. And it's much, much slower than the 1GHz Cortex-A9s available both in 1-core and 2-core variants in SoCs like TI's OMAP4, NV's Tegra2, or ST-Ericsson U8500.
Intel wants you to think Atom is really fast, that ARM can't match their performance and certainly not their software portfolio. This is completely insane and misleading. Their performance is miserable in the 2010+ timeframe and Moblin will soon become very unattractive compared to Android. The only good news is I'm impressed by what I've seen of Moorestown's power efficiency - but it's still much worse than ARM, so why bother when all your other advantages are gone?
Android, by the way, is the ultimate ARM trojan horse because it's all Java, so the same apps can work on x86 and ARM with no porting whatsoever in theory. Given that it seems there will soon be both x86 and ARM netbooks running Android (with the first probably being x86), this will do a lot to legitimize ARM in that market in a way many insiders don't even seem to be realizing.
We don’t know what Nokia may have signed up to with Intel in terms of taking future product, but I reckon it’ll be minimal considering the level of Intel’s long-standing desire to get into small form factor devices. I reckon this is a marriage born out of desperation for both parties, and one in which Nokia can’t lose, because ultimately they can walk away from anything which doesn’t cut the ice. Keep in mind the challenge facing any semiconductor vendor in this market. They have to slap a complete integrated working bug-free proven solution – hardware, software, middleware, apps, the whole caboodle - down on the table in front of an OEM first, before they’ll get even a sniff of a Purchase Order.
Many moons ago Nokia collaborated in exactly the same way with Texas Instruments to build the next generation high-end/Smartphone architecture, and thus was born OMAP™. OMAP was a runaway success for TI and has powered the majority of Nokia phones ever since. The trouble is times are hard and the end user market still expects more and more for less money. With a guaranteed volume load from Nokia, TI isn’t particularly motivated to spend a whole load of R&D dollars that it hasn’t got to fulfil any fanciful ideas that Nokia may have to replace its own incumbent product. On the other hand, for Nokia it's nigh impossible to reduce the BOM of their handsets significantly when tied in to a sole source supplier for the key semiconductor platform over whom they have no commercial leverage. Even before the current recession started to bite Nokia must have been smarting (hehehe…another great pun) like hell. R&D isn’t getting any cheaper, and they not long ago bought out the rest of Symbian and its entire cost base, with the intention of giving the software away for free. It’s no surprise they're currently culling all those Symbian employees left, right, and centre, with more planned in the year ahead.
So Nokia were left with an interesting conundrum - where are the huge R&D dollars going to come from in the mobile space to develop a next generation platform, whilst at the same time getting them out of their current single source supply trap? It’s a tough call, and beyond any chip company currently battling in the handset market, because they’re all still blithely selling silicon by the square millimetre and continually eroding prices as they ride the slippery slope of competition to ultimate oblivion.
Roll up! Please step forward any large semiconductor company with deep pockets, and loads of dosh in the bank to invest. This is an ideal opportunity for anyone sufficiently foolish and arrogant to believe they can ascend to a dominant position in the handset space like they did once before in another market, in another time. Candidates exhibiting an almost fanatical desperation to be in high volume small factor devices will be viewed most favourably.
Big Softie, six years ago Aldo Romano, then head of ST's telecoms business told that me ST paid for 300 engineers to work exclusively for Nokia. If these other supply alliances which Nokia has forged - with Broadcom, TI, Qualcomm and Intel - require the same sort of obligation, just think of the amount of buckshee engineering resource Nokia could be taking advantage of!
I'll side with "Big Softie", Nokia has just found a new deep pocket to dip into, I think we can expect some "white papers" and lots of ego building praise to flow in Intel's direction. In exchange we'll see boat loads of cash returning. Real products will be in very short supply. The joke is that in the perverse world of stocks both companies will prosper...
Meanwhile the future for mobile is being played out by QCOM and Mediatek, QCOM wants $5 royalties per phone yet MTK is selling whole multimedia cell phone chip sets for $5. What is more disturbing is that MTK has a serious competitor hot on its heals(from what I've heard, better phone, better Multimedia, better roadmap, better price) so even MTK is hurting.
Do the numbers, and you'll quickly see that Atom is a PR exercise.
Robert, I agree with you and Big Softie. Nokia's playing a clever game by getting its co-development collaborators like ST-Ericsson, TI, Qualcomm. Broadcom and, now, Intel, to do loads of free R&D for Nokia for which Nokia can give a quid pro quo - or not - as it sees fit.
I agree with Robert and Big Softie in general terms. The new deep pocket idea is not one I'd considered, but it makes sense. However we still need to set the context - new deep pockets for what?
Nokia seem to be trying to build much more value in their software capabilities - to be honest I've been watching their new strategy development for some time and it still does not quite add up to me...but I'll accept they are way smarter than I am. I can see what they are doing, but not where the new business is for them.
Intel are in a much more obvious problematic position. Their key traditional market growth drivers are going away fast, and the only market for significant growth for them is Mobile. They are desperate and have cash. So they are buying their way into the mobile silicon sales they hope.
So at a 10,000 foot level their seems to be strategic alignment. Except its not an equal deal. Intel has to succeed in this sort of partnership. Nokia can take it or leave it, ARM or Intel, what do they care? That puts them in a very strong position in that relationship.
The big Intel challenge as far as I can see is the same one that effectively halted their previous embedded/mobile attempts - lack of margin in the target business. They don't know how to operate in such a fiscal climate, their fat margins are inculcated throughout their organisational culture.
So to me the questions comes down to - "can Intel make Mobile a good margin business for silicon/hardware providers?" or "Can Intel develop a low margin business culture?"
Geoff, I don't think they can - that's why Atom is being off-loaded to foundry.