Intel enters wearables:"The Intel Enigma"

[witeken]

And Intel's margins are only set to improve as the foundries haven't figured out how to move below 28nm without increasing the costs of the chips whereas Intel has done exactly that.

Everyone in ultramobile is going to have this problem except for Intel.

That's astonishing. Broadcom is admitting the cost per transistor is getting higher, while Intel is simply marching on. Even more, what the foundries call 5nm is actually Intel's 7nm node, which they can do without EUV. The person is probably a bit too pessimistic about Moore's Law, but it's telling something nonetheless.

Is that why SoFIA is on 28nm TSMC for 2014-2015 and not Intel 22nm or 14 nm ??

Have you forgotten? 28nm SoFIA was a clever idea from some person at Intel that has increased the TTM of SoFIA substantially, even though it wasn't long on the roadmap. Intel is still busy porting the Infineon assets to its own fabs, so there's no other option. It's really time to market.

 

[NTMBK]

Did you check the price on wearables? Its smartphones all over. That means 60% margins on CPUs.

Prices are high, but where are the sales? I still haven't seen a single smart watch, and I work in a company of affluent tech geeks.

 

[witeken]

Prices are high, but where are the sales? I still haven't seen a single smart watch, and I work in a company of affluent tech geeks.

It's a new market.

http://www.myce.com/news/intel-reports-breaktrhough-in-ssd-costs-to-release-10tb-ssds-73553/

 

[AnandThenMan]

Hotlink image doesn't seem to be working.

 

[Arachnotronic]

Is that why SoFIA is on 28nm TSMC for 2014-2015 and not Intel 22nm or 14 nm ??

First SoFIA is being done on 28nm for time to market reasons (i.e. modem IP already on 28nm), not cost reasons. Intel has stated that it will see a cost benefit when it moves SoFIA to 14nm.

 

[tential]

Prices are high, but where are the sales? I still haven't seen a single smart watch, and I work in a company of affluent tech geeks.

I've seen a number of them actually each person who I know who has one has purchased multiple ones already.

 

[witeken]

Intel certainly has had more than expected problems with 14nm. With all the delays and extra costs, and delayed projects, it may take them a long time to make 14nm cheaper than 22nm if you include all the associated expenses.

14nm will be cheaper in Q3'15. Just think about that... Those chips in H2 have been in the fabs in H1 and Intel's 22nm's yield is in the 90s, yet they're going to bring costs to where Haswell was 2 quarters ago or is even today. Not stellar, but not too bad either. Samsung's 20nm+FinFET yield, for comparison? ~20%.

Intel is accustomed to margins around 65%. If they were to go with that for an ARM competitor that would mean each Intel mobile processor would be over $50. The Average ARM processor is $20. For Intel to get to $20 their margins would have to drop to near zero. And that's being very generous to Intel it assumes they can make a line of processors that will compete with the various ARM offerings.

It doesn't matter what the margins are: Intel is stating very clearly that they're pursuing the market, and they're going to be profitable, probably in 2017.

 

[witeken]

Hotlink image doesn't seem to be working.

The Tremendous opportunity slide on this page: http://www.myce.com/news/intel-reports-breaktrhough-in-ssd-costs-to-release-10tb-ssds-73553/

 

[Arachnotronic]

Intel is accustomed to margins around 65%. If they were to go with that for an ARM competitor that would mean each Intel mobile processor would be over $50. The Average ARM processor is $20. For Intel to get to $20 their margins would have to drop to near zero. And that's being very generous to Intel it assumes they can make a line of processors that will compete with the various ARM offerings.

This reasoning literally makes no sense. Do you think that the processors Intel will sell into mobile will have the same cost structures as those sold into PCs/servers?

You can't just look at selling price when you are talking about margins. Cost structure matters.

 

[NTMBK]

It's a new market.

Yay, sales forecasts! I'm just going to roll this out for old time's sakes

I just don't believe that smart watches are that good an idea, frankly. Wearables have a reliable niche market in the fitness tracker stuff, but I don't see it exploding the way phones did. And while the industrial automation side of Internet of Things might be a very big market, I doubt Twitter capable toasters are going to be the Next Big Thing.

But that's just me being an armchair analyst, I may well be wrong :thumbsup:

 

[Erenhardt]

How much $/SoC in contra-revenue? Maybe some company could give intel smart watches for free and still make a top dollar?

 

[AnandThenMan]

It doesn't matter what the margins are:

Really?

Intel is stating very clearly that they're pursuing the market, and they're going to be profitable, probably in 2017.

Of course they are. So far their promises and projections have proved to be highly inaccurate, to put it kindly. As for "smart" watches and the like I predict they will bomb completely including from Apple.

 

[witeken]

Yay, sales forecasts! I'm just going to roll this out for old time's sakes

I just don't believe that smart watches are that good an idea, frankly. Wearables have a reliable niche market in the fitness tracker stuff, but I don't see it exploding the way phones did. And while the industrial automation side of Internet of Things might be a very big market, I doubt Twitter capable toasters are going to be the Next Big Thing.

But that's just me being an armchair analyst, I may well be wrong :thumbsup:

Technically, the slide is correct. A lot of people probably don't even know those things exist, so this really is just the early beginnings of the technology. Obviously many people are optimistic, but it remains to be seen where the 100% penetration will be and what the release cadence will be. I'm optimistic (although I'd argue that because it's realistic), but that's just me. It won't be as big as phones, of course.

 

[witeken]

How much $/SoC in contra-revenue?

It depends on the BOM-delta, because that's what Intel pays. Intel has released updated versions of Bay Trail with less contra-revenue, but at the beginning of that year it was as much as $20.

 

[Phynaz]

Is that why SoFIA is on 28nm TSMC for 2014-2015 and not Intel 22nm or 14 nm ??

Time to market. 14nm wasn't available for Sophia.

Look at your own slide for 2016.

 

[positivedoppler]

So Intel now makes business decision based on the CEO wife's fashion taste. It makes queasy watching the mighty 5 trillion lb gorilla slinging dart at the wall in desperation hoping to hit gold. They still probably will make a lot of money from it so I guess all is fair in business. I can't put an exact finger on what I don't like about Intel doing this, but it's about as stomach turning as watching two 400 lb plumbers kiss. I expect this type of Jerry Springer side show from AMD, but watching Intel...I just wish they didn't.

 

[AtenRa]

They have 22nm FF in mass production for more than 3 years and they needed TSMCs 28nm for a 2015 product for time to market ??

edit: also, in 2016 both Samsung 14nm FF and TSMCs 16nm FF will be way cheaper than 28nm today. 14nm SoFIA on Intels manufacturing will face the same situation they have today.

 

[NTMBK]

They have 22nm FF in mass production for more than 3 years and they needed TSMCs 28nm for a 2015 product for time to market ??

edit: also, in 2016 both Samsung 14nm FF and TSMCs 16nm FF will be way cheaper than 28nm today. 14nm SoFIA on Intels manufacturing will face the same situation they have today.

They acquired a company who was already developing a modem for 28nm TSMC. Easier to keep producing the 28nm part as planned (but with Silvermont cores), and bring the next generation part in house at 14nm.

 

[witeken]

So Intel now makes business decision based on the CEO wife's fashion taste. It makes queasy watching the mighty 5 trillion lb gorilla slinging dart at the wall in desperation hoping to hit gold. They still probably will make a lot of money from it so I guess all is fair in business. I can't put an exact finger on what I don't like about Intel doing this, but it's about as stomach turning as watching two 400 lb plumbers kiss. I expect this type of Jerry Springer side show from AMD, but watching Intel...I just wish they didn't.

They're bringing innovation to the market. The MICA is a collaboration with Opening Ceremony.

 

[AtenRa]

They acquired a company who was already developing a modem for 28nm TSMC. Easier to keep producing the 28nm part as planned (but with Silvermont cores), and bring the next generation part in house at 14nm.

They could port the CPU cores to 28nm TSMC but not the modem to Intels 22nm FF ??

 

[kimmel]

They could port the CPU cores to 28nm TSMC but not the modem to Intels 22nm FF ??

Sounds perfectly reasonable. Analog IP is a giant pain in the rear to port compared to digital logic.

 

[Phynaz]

They could port the CPU cores to 28nm TSMC but not the modem to Intels 22nm FF ??

Yes. You ever try to make an analog chip?

 

[ShintaiDK]

Intel is accustomed to margins around 65%. If they were to go with that for an ARM competitor that would mean each Intel mobile processor would be over $50. The Average ARM processor is $20. For Intel to get to $20 their margins would have to drop to near zero. And that's being very generous to Intel it assumes they can make a line of processors that will compete with the various ARM offerings.

Qualcomm gets between 59 and 72% gross margins.

You are trying to say thats not possible?

 

[witeken]

They could port the CPU cores to 28nm TSMC but not the modem to Intels 22nm FF ??

Pretty much the only thing that Intel has done is taken the low-end, integrated infineon ARM SoC and replace the ARM CPU with IA.

"The first one will be 3G, the next one will be LTE. Step two: we'll really get the advantage, because remember: we're pragmatic, but we're also driven to if it computes, it computes best with Intel. Then the next thing you wanna do? You wanna move that inside. So as we exit 15, you'll see SoFIA move from the outside foundry to the inside, at 14nm. Now you have: an entry level part, fully integrated, 3G, LTE, with IA, on 14nm. Something nobody else on the world will be able to compete against."

 

[AtenRa]

Clearly you havent seen this, its all about cost not time to market.