TSMC 16nm in production, 10nm still on schedule (2017). Intel IDF Shenzhen

[witeken]

TSMC expects 10nm volume production in 2017

“We are working with over 10 customers on their 10nm product design. The qualification schedule remains at the end of 2015. We are working with customers on tapeout, and we expect volume production in 2017.”
(...)
“We had a very advanced metal stack with a set of leading customers who wanted to squeeze every last drop out of planar. So 20nm turned into a great deal of business for us. We didn't have to take on two challenges at the same time—developing this complex metal system and bringing up FinFET.”

There's also: TSMC clarifies 16nm and 10nm schedule:

TSMC’s 16nm FinFET node (16FF) is already online, but the improved 16nm FinFET Plus (16FF+) node should be available soon as well. The company confirmed 16FF+ will enter volume production in mid-2015, roughly three months from now.

Which quotes EETimes: TSMC Outlines 16nm, 10nm Plans

The 10nm process will have 2.1 times the logic density of the 16nm node along with a 20% speed gain and 40% power reduction. The company demonstrated a 256 Mbytes SRAM made in the process. TSMC expects its 10nm to be in production by the end of 2016 and alluded to more than ten partnerships in various stages of design.

Some analysis:

“We think 10nm will be the long-lasting technology node and for TSMC to be accelerating 10nm, I think that is a very good sign for the industry,” Jones [= International Business Solutions CEO] said. “With the acceleration of 10nm — which they might end up going 8nm — TSMC will close the gap with Intel. I think TSMC is on a roll.”

What does he mean with that last part? Is TSMC going to rename 10nm to 8nm ?

Fortunately, he corrects himself at the end of the article:

“If that happens and TSMC has closed the gap with Intel, the issue is then if TSMC’s 10 and Intel’s 10 are the same,” he said.

They're not. Now let's see what Intel's up to.

First, Intel is combining mobile and client in its quarterly report, which is causing some conspiracies of Intel hiding its losses :sneaky:. (Don't ask me how you can hide losses in a quarterly report.) Secondly, the Compute Stick will be launched shortly for $110 or $150.

Lastly, Intel showed a RealSense smartphone at IDF Shenzhen: Intel Marks 30 Years in China with New Products, Investments and Collaborations. More than 45 tablets are in development with Atom x3. They showed x3 LTE for H2. They announced Atom x3 for IoT. Braswell is shipping. They want to reduce cost and TTM for x5 with reference design.

Update: BK also showed Skylake. So no quadcore Broadwell-M, no BDW-K.

Finally, the semiconductor is a $340B business. Intel is 1, Samsung 2, Qualcomm 3: Semiconductor sales up eight percent.

 

[Nothingness]

Finally, the semiconductor is a $340B business. Intel is 1, Samsung 2, Qualcomm 3: Semiconductor sales up eight percent.

One can find the list here for instance. Hope this helps

 

[Idontcare]

witeken, I recommend you break this out into two separate threads. One about TSMC 16nm and 10nm plans/timeline, and the second about the Intel stuff.

Throwing them both into the same thread is basically asking for the thread to become one big mess. Just my 2cents.

 

[NTMBK]

First, Intel is combining mobile and client in its quarterly report, which is causing some conspiracies of Intel hiding its losses :sneaky:. (Don't ask me how you can hide losses in a quarterly report.)

By combining the two, their report will no longer show negative revenue in the mobile division, hiding the fact that they were paying customers to take mobile parts off them.

 

[witeken]

By combining the two, their report will no longer show negative revenue in the mobile division, hiding the fact that they were paying customers to take mobile parts off them.

Intel couldn't have been more clear about contra-revenue. They've disclosed it at IM'13, they've reported on their progress of reducing the BOM, they've disclosed how much the BOM is, they've said what will happen to contra-revenue in 2015 (it will fade away in H2 as SoFIA enters and the Bay Trails with reduced BOM and CT enter the market), and they've disclosed how many contra-revenue tablets there are (46M). Contra-revenue is about $1B, but now they have an appreciable tablet market share. They've also disclosed how much the mobile-only R&D is: a couple hundred million per quarter. They have said that their mobile losses will be reduced by $800M in 2015, and in 2016 they plan to be on the verge of profitability.

People still don't know what contra-revenue is, apparently.

 

[NTMBK]

Intel couldn't have been more clear about contra-revenue. They've disclosed it at IM'13, they've reported on their progress of reducing the BOM, they've disclosed how much the BOM is, they've said what will happen to contra-revenue in 2015 (it will fade away in H2 as SoFIA enters and the Bay Trails with reduced BOM and CT enter the market), and they've disclosed how many contra-revenue tablets there are (46M). Contra-revenue is about $1B, but now they have an appreciable tablet market share. They've also disclosed how much the mobile-only R&D is: a couple hundred million per quarter. They have said that their mobile losses will be reduced by $800M in 2015, and in 2016 they plan to be on the verge of profitability.

People still don't know what contra-revenue is, apparently.

The R&D has nothing to do with contra-revenue, I certainly agree. That is why R&D is not factored into the revenue calculations... the revenue which was negative in the last quarter. They gave away more in contra-revenue than they were paid for their mobile parts.

 

[Idontcare]

The R&D has nothing to do with contra-revenue, I certainly agree. That is why R&D is not factored into the revenue calculations... the revenue which was negative in the last quarter. They gave away more in contra-revenue than they were paid for their mobile parts.

Contra-revenue is literally nothing more than R&D.

The problem, as I see it, is very few people here have ever been in an R&D environment. They read about it. They hear about it. It empowers them to feel confident talking about it, but truthfully they don't know jack about it.

So when it happens in a new way, in a way that is as different to yesterday's R&D as yesterday's R&D was different from yesteryear's R&D, they can't tell six of one from a half dozen of the other.

The only consolation, and this is the truth, is that none of the naysayer's naysaying matters. It is sad, but the reality.

The people who made the very world that we all enjoyed last year will endeavor to make the world that we will all enjoy this year. And to be sure the vast majority of us will be clueless how they actually make any of it happen, last year or this.

And they will all rightly laugh themselves to the bank while we smugly talk about how what they do is wrong and just shouldn't happen the way they made it happen.

Yeah us.

 

[NTMBK]

Contra-revenue is literally nothing more than R&D.

No, it isn't. R&D is a completely different expenditure, which is recorded elsewhere on the report. R&D cost is not a factor in revenue calculation.

Contra-revenue is rebates, market development funds, and so on.

 

[witeken]

Contra-revenue is literally nothing more than R&D.

The problem, as I see it, is very few people here have ever been in an R&D environment. They read about it. They hear about it. It empowers them to feel confident talking about it, but truthfully they don't know jack about it.

So when it happens in a new way, in a way that is as different to yesterday's R&D as yesterday's R&D was different from yesteryear's R&D, they can't tell six of one from a half dozen of the other.

The only consolation, and this is the truth, is that none of the naysayer's naysaying matters. It is sad, but the reality.

The people who made the very world that we all enjoyed last year will endeavor to make the world that we will all enjoy this year. And to be sure the vast majority of us will be clueless how they actually make any of it happen, last year or this.

And they will all rightly laugh themselves to the bank while we smugly talk about how what they do is wrong and just shouldn't happen the way they made it happen.

Yeah us.

Kinda cryptic post. You should elaborate.

How is contra-revenue R&D? What is, then, R&D? What is the purpose of the philosophical explanation that then follows and what does it mean?

 

[AtenRa]

Contra-revenue is literally nothing more than R&D.

Contra-Revenue is literally nothing more than publicly confirming BayTrail was a failure as a product (at least for the mobile market).

ps. im sure mrmt will love that :biggrin:

 

[Abwx]

No, it isn't. R&D is a completely different expenditure, which is recorded elsewhere on the report. R&D cost is not a factor in revenue calculation.

Contra-revenue is rebates, market development funds, and so on.

In principle the RD is factored in the chip price, if it s positive of course.

A Baytrail chip that should normaly be sold at about 30$ cant possibly require more than 10$/chip to amortize the RD on a 46 millions quantity (and soaring currently).

For 2014 this amount to 460 millions losses due to non funding of the RD, an equivalent amount is lost in cost manufacturing wich eat another 10$/chip, this put the losses at 1bn since they give the chips for free, the rest, about 3bn, are subsides paid to the OEMs at a rate of 65$/chip, including the latter this point to 85$ total subside/chip.

 

[witeken]

There's also some GloFo news: yields are double digits and HVM will start this year.

http://fudzilla.com/news/37475-globalfoundries-clarifies-14nm-plans

 

[krumme]

Contra-revenue is literally nothing more than R&D.

The problem, as I see it, is very few people here have ever been in an R&D environment. They read about it. They hear about it. It empowers them to feel confident talking about it, but truthfully they don't know jack about it.

So when it happens in a new way, in a way that is as different to yesterday's R&D as yesterday's R&D was different from yesteryear's R&D, they can't tell six of one from a half dozen of the other.

The only consolation, and this is the truth, is that none of the naysayer's naysaying matters. It is sad, but the reality.

The people who made the very world that we all enjoyed last year will endeavor to make the world that we will all enjoy this year. And to be sure the vast majority of us will be clueless how they actually make any of it happen, last year or this.

And they will all rightly laugh themselves to the bank while we smugly talk about how what they do is wrong and just shouldn't happen the way they made it happen.

Yeah us.

Sure. But even people involved in tech r&d recons 4b is a lot, no matter how you slice and label it.

 

[krumme]

There's also some GloFo news: yields are double digits and HVM will start this year.

http://fudzilla.com/news/37475-globalfoundries-clarifies-14nm-plans

To celebrate 14nm product hitting the streets after 6 years of ppt products, Mubadala in cooperation with fudzilla, plans a camel safari to the construction site of the next fab. Bring guns, water and plenty of money.

 

[Arachnotronic]

IDF is seeing its usefulness diminish for anybody who cares about CPUs. The most interesting things they had to show were some tidbits about Cherry Trail/Airmont, but even still we don't know what microarchitectural changes they made to Airmont from Silvermont.

It's just sad that they spend so much time talking about things like RealSense, wireless charging, and Curie. Intel thinks that they're appealing to technology enthusiasts with this stuff, but as an enthusiast, I'm interested in architectural deep dives and other technical goodness.

Hopefully IDF in San Francisco is more interesting, but I'm not getting my hopes up.

 

[mikk]

IDF in San Francisco is for sure more interesting because of Skylake.

 

[Phynaz]

IDF China is meant to address emerging markets, that's why high performance stuff doesn't get announced there.

 

[Arachnotronic]

IDF China is meant to address emerging markets, that's why high performance stuff doesn't get announced there.

I remember Sandy Bridge-EP got a nice technical deep dive at a China IDF a few years ago...

 

[JDG1980]

There's also some GloFo news: yields are double digits and HVM will start this year.

http://fudzilla.com/news/37475-globalfoundries-clarifies-14nm-plans

Both of the processes being discussed here are low-power processes. How long do we have to wait for the good stuff?

 

[witeken]

I remember Sandy Bridge-EP got a nice technical deep dive at a China IDF a few years ago...

Maybe there is nothing to dive deep on? Which deep dive do you want that isn't already on AnandTech?

 

[Idontcare]

Sure. But even people involved in tech r&d recons 4b is a lot, no matter how you slice and label it.

It goes to show how high the barrier to entry is for x86-enabled tablets and phones over that of the already developed and amortized R&D of arm-based tablets and phones.

That decade of R&D invested by ARM, Nokia, and Texas Instruments made it possible for low-cost (low BOM) products these days based on the same bones. Intel wasn't making the same investments, and it cost them roughly $4B to jumpstart that product development train.

For the people who want to argue that contra-revenue is not about R&D, I'm not saying it is chip R&D, it is product R&D. Intel pays down the BOM penalty that comes with x86 devices not having benefited from as long of an R&D pipeline history as ARM devices. So they are subsidizing the R&D pipeline of the OEMs.

For Intel it comes as a non-R&D reportable expense, but if they were vertically integrated to the tune of Samsung or Apple the contra-revenue itself would have most assuredly been reported as a product development (R&D) expense.

 

[witeken]

About the $4B:

~800M of which is contra-revenue.
~2B of which is shared investment that also had been necessary without the existence of mobile (e.g. Atom core development, graphics, comms,....)
~1.2B of which is specific investments, including non-recurring engineering helping them port designs.

 

[sm625]

So TSMC is "on a roll" yet we've been stuck on 28nm for 4 years?

 

[Snafuh]

I think the discussion about contra-revenue comes up over and over again not because people don't understand it but because people don't think it's fair to subsidize uncompetitive products. Without contra-revenue Intels mobile products could not succeed on a free market with big competition. Intel uses it's market power and money to get into the market. It might be legal and definitely profitable for Intel in the long term but gift-wrapping your product in money is not how a success story should be written.
But it has been discussed over and over again here.

 

[kimmel]

But it has been discussed over and over again here.

And people get it wrong again and again which means to me they don't understand it. You have to spend money to make money. Virtually every success story (and failure) is about people spending money on the hopes that they make money later. I guess people dream of businesses that are continuously profitable without ever needing seed investment or borrowing money from the bank. Can happen, usually doesn't.