Intel Investor Meeting 2014

[witeken]

Intel Investor Meeting 2014 November 20

Presentations and PDFs

After IDF 2014, the next big event for Intel is the Investor Meeting. This is (probably quite literally) the follow-up of the 2013 Investor Meeting. One year ago, Intel announced Cherry Trail and Broadwell on 14nm for 2014. They announced 10nm for 2015 production and Broxton for a mid-2015 launch. They announced SoFIA, a low-end smartphone ship built on TSMC's 28nm, with Silvermont cores. They also announced that they would move it to 14nm EOY 2015.

This was also where they announced their infamous and often misinterpreted contra-revenue program, to accelerate their tablet campaign to 40M units. While they are squarely on track to have 40M+ tablets with IA, 14nm didn't really materialize this year, with only a handful of devices with the now named Core M chip, the first of which has received poor reviews. Yield progress seems to have been worse than anticipated at IM13, but other disclosed information like density, performance etc. is spot-on.

We should expect more information about Broxton and SoFIA, about Intel's smartphone plans, and if we're lucky a roadmap for Skylake and Broadwell and a confirmation or warning of 10nm's health and expected ramp, which should be EOY2015 if there are no delays.

I will update the thread with links and information when it comes available. Which questions do you want to be answered by Intel?

:thumbsup: Schedule: http://www.intc.com/releasedetail.cfm?ReleaseID=883452&ReleasesType=Corporate News
:thumbsup: Webcast: http://intelstudios.edgesuite.net/im/2014/live_im.html
:thumbsup: Brian Krzanich' keynote summary: http://forums.anandtech.com/showpost.php?p=36921511&postcount=8
:thumbsup: William Holt's keynote summary about 14nm: http://forums.anandtech.com/showpost.php?p=36921597&postcount=10
:thumbsup: Stacy Smith's financial keynote summary: http://forums.anandtech.com/showpost.php?p=36921907&postcount=13
:thumbsup: Q&A with above 3 people summary: http://forums.anandtech.com/showpost.php?p=36921967&postcount=14
:thumbsup: Presentations and PDFs
:thumbsup: Diane Bryant's server presentation summary: http://forums.anandtech.com/showpost.php?p=36922169&postcount=18
:thumbsup: Kirk Skaugen's PC presentation summary: http://forums.anandtech.com/showpost.php?p=36922359&postcount=22
:thumbsup: Q&A with above 2 people summary: http://forums.anandtech.com/showpost.php?p=36922438&postcount=26

 

[witeken]

I've added a link to the schedule. It doesn't seem like there's a free webcast, though.

The meeting includes a 30 minutes presentation on advancing Moore's law. Hopefully we'll hear a few new things there on their next node instead of a 3rd part of a 14nm trilogy.

Edit: You can watch the webcast here: http://intelstudios.edgesuite.net/im/2014/live_im.html

 

[witeken]

The IM has just begun, BTW. Brian's presentation will start shortly.

 

[Hyperlite]

Thanks for the links!

 

[NTMBK]

Thanks for the links Witeken Any chance you could post any interesting highlights? Listening to corporate types trying to understand tech is always rather painful!

 

[witeken]

Thanks for the links Witeken Any chance you could post any interesting highlights? Listening to corporate types trying to understand tech is always rather painful!

Sure, already doing, actually.

 

[Hyperlite]

Thanks for the links Witeken Any chance you could post any interesting highlights? Listening to corporate types trying to understand tech is always rather painful!

It's worth noting that Krzanich come up through Intel as an engineer, so he's pretty articulate with the technical bits.

 

[witeken]

Key messages (I won't bother with the lay-out, I think it's reasonably summarizes for a 30-50 minute presentation):

Andy Bryant (introduction):
*Admits missing mobile and (literally) paying a price for it, but now fully committed to mobile.

BK:
*Still focusing on market-driven innovation (e.g. Mica) and fast TTM.
*For the first time, 100M CPUs shipped in 1 quarter (Q3).
*50 years Moore's Law celebration, next year. Important and relentless.
*PC market stabilizing, challenge of growth, financials and innovation (examples: RealSense, wireless charging, multiple OSes).
*Server market: 15% CAGR growth through 2018, Caldera (FPGA).
*Moore's Law: transistor density & performance for cost, TDP and performance products: $11B R&D in 2014; 75% shared and fundamental spending (CPU arch, process tech, graphics IP, validation and testing,...), 25% specific for different segments.
*Cherry Trail out by the end of the year.
*Mobile: above 40M, SoFIA quickTTM solution, shipping in H1, LTE in H2; up to high-end Cherry Trail.
*Phones: Synthesized core (SoFIA), brough to Rockship and Speadtrum.
*Contra-revenue: not proud, but not ashamed either for losses. Choices: wait 1-2 years or move? They're in the business to win.
*Modem: XMM 7260, in tablets and laptops (will grow to >50% in next years).
*New Client Computing Group with mobile group; platform group for modems.
*IoT, wearables: 2B business and growing 20% CAGR, partnerships.
*Open foundry: Panasonic, etc.
*Internal changes: bringing new people and drive change in the company.

William Holt is now up. PDF will be available soon.

 

[III-V]

Cherry trail is still on track for EOY '14? Nice.

Next year is going to be exciting. It'll be a real treat after this year. Graphics industry will be nice too, from all three players.

Intel's looking super "investable" right now. Wish I had the money to get started.

 

[witeken]

William Holt:
*Recap of 2013 statements and slides (e.g. capital spending, wafer costs): Moore's Law still alive, but 6 months later (2.5 years).
*Yield graph extended, in healthy range, but below expectations from 2013 (BTW: 22nm has highest yield ever), catching up. Yields will match in ~H2.
*14nm: true 14nm technology, not all transistors are created equal. 2nd gen FinFET, recap of feature sizes (-> provides good scaling) and 0.54x SRAM scaling. Recap of transistor changes: closer together, taller, fewer and rectangular fins. Image shown. 3.5 year FF lead.
*FinFETs are less flexible: can only choose number of fins. Substantial improvement in number of fins needed (now only 1-2).
*Recap of 2X perf/watt improvement.
*Interconnect: Image shown. Uses much energy. Trying to improve energy, dielectric constant. First use of air gaps in interconnect system in industry: allows for substantial benefits.
*Recap (3rd time since and including IM13) of density scaling. Now lined up in time instead of node!
*20nm planar: higher leakage, more power, etc. Competitors pause at "16nm" to do FinFET.
*Cost reduction recap: although high wafer cost, better scaling allows for lower cost. The "projected" (*) cost from 14nm removed. 7nm: will be below line of trend of 65nm-22nm, just like 10nm and 14nm (means higher density scaling and lower cost than would be expected from new node).
*Density: Broadwell has 2.2X increase in density (compare: 20nm is 1.9x).

No 10nm disclosures.

Edit:

2015 expectations:

*Revenue: mid-single digits growth, up from 55.9B in 2014.
*Gross margins: 62% plus or minus 2 points
*R&D + MG&A: ~20B, but as percent of revenue down.
*$10.5B capex, plus or minus $0.5B

 

[mikk]

Cherry Trail out by the end of the year.

Does anyone believe in this? Cherry Trail this year is nonsense imho. Otherwise we would have much more infos. There is nothing. Braswell has been delayed into Q3 2015. For Cherry Trail Q2 2015 is a realistic target.

 

[witeken]

Does anyone believe in this? Cherry Trail this year is nonsense imho. Otherwise we would have much more infos. There is nothing. Braswell has been delayed into Q3 2015. For Cherry Trail Q2 2015 is a realistic target.

You remind me of some Skylake predictions for Q3'16.

Intel's CFO just started.

 

[witeken]

CFO Stacy Smith:
*Some Haiku and BK fun.
*2014: growth across business, 55.8B (6%), 63% gross margin (up 3 points).
*Very high R&D compared to others.
*Slide on consolidation (e.g. 16 companies in modem in 2016, 7 today).
*Recap of BK's slide on IP reuse (e.g. server borrows heavily from PC, reducing costs).
*DGC: 14B revenue, 7B operating margin.
*MCG: down because of lower 2G and 3G demands.
*PCCG: 9% unit growth in 2014. ASP slightly down.
*Higher costs in 2015 due to 14nm ramp.
*Ivy Bridge ramp compared at equivalent costs: Q2'12, BDW: Q1'15 (almost 3 years).
*Bay Trail now over 50% of Celeron and Pentium.
*Bay Trail platform cost will be reduced by another 30% in 2015; allows high margins.
*BOM delta: from $15 to BT's entry $8. No contra-revenue or BOM delta in SoFIA (will improve margins).
*MCG gross margin up in 2015; positive in H1.
*Relatively small amount of unique spending in R&D in mobile to get into large market.
*10nm start-up costs in 2015.
*IoT: 2.1B, 26% operating margin, growing.
*Software and services: 2.3B, 0.1B operating margin.
*Expects Lollipop IA tablets to be first to market (except Nexus).
*Switching gears to manufacturing: reiterates importance of cost/transistors (cfr 20nm and 16nm cost doubt).
*14nm: 30% higher capex, big increase, just as 22nm. Die slides come down and flat CAGR means fewer wafer starts. Interesting slides, you should check them out when they become available.
*Because BDW-Y is almost 40% smaller, this offsets increased capex.
*Interesting slide on capacity and loadings.
*7.3B of cpaex is for capacity, reasonable portion for 450mm, non-manufacturing and development. Pretty flat over time (although 450mm is new since 2013).
*Talk about depreciation: takeaway is that it won't impact gross margins.
*Gross margin of 62%: down because 10nm spending, cost per unit. This is offset by higher ASPs (not because of raising prices, but servers). In H2: BDW costs come down, as well as 10nm start-up, so higher margins than H1.
*Great slide about leading edge companies: down from 25 (1320nm) to 4.
*Shareholder return, slide over time (personal opinion: could have invested that shareholder money -- almost $100B since 2004 -- in technology...).

Still waiting for some 10nm guidance (also see this post.

Now Q&A (with Stacy, Brian and William).

 

[witeken]

Q&A:

*No breakeven in profit (but big improvement) in mobile in 2016.
*Speadtrum and Rockchip volume will start in ~Q4.
*Q: has ARM a fundamental advantage (e.g. see Apple). A (BK): No. It has simply not been a focus until the last ~2 years. Have some great plans ahead for those markets.
*Question about lack of tick-tock in presentations (10nm). Intel will be more sneaky, careful with 10nm information signalling. Intel's been giving too much information too fast, will give information and has a lot of experience with Moore's Law (50 years). Should watch the webcast for answer in its full glory .
*Intel will develop less own smartphone SoCs, will invest to let others design SoCs and take them to market. Interesting.
*Mobile group includes shared R&D (as talked about earlier in presentations, so it isn't all esclusive R&D for mobile).
*7nm cost/transistor, EUV? That slide (with dotted line) was intentionally ambiguous, but it will be below line (= better). Even if there's no EUV, so that's very impressive.
*Question about tablet. Moore's Law has helped. 2015: 14nm, 2016: SoFIA, then Moore's Law will start to kick in. Forecast for 2015: no follow-up for the 40M number because less aggressive share gaining (growing with market, although not below), but focus on profitability.
*Target: 60% gross margin.

Now Diane Bryant (server).

 

[Hyperlite]

Thanks Witeken!

 

[III-V]

Presentations in PDF form that have been posted so far:

http://intelstudios.edgesuite.net/im/2014/pdf/2014_Intel_IM_Krzanich.pdf

http://intelstudios.edgesuite.net/im/2014/pdf/2014_Intel_IM_Holt.pdf

 

[kimmel]

Presentations in PDF form that have been posted so far:

http://intelstudios.edgesuite.net/im/2014/pdf/2014_Intel_IM_Krzanich.pdf

http://intelstudios.edgesuite.net/im/2014/pdf/2014_Intel_IM_Holt.pdf

I think this is the first time in official context that Intel has talked about Air Gap interconnect stack. Nice picture in the Holt slides.

 

[witeken]

Diane Bryant:

*15% CAGR through 2018: 5% unit CAGR, 6% ASP CAGR. Significant growth to high-end ASP SKU: because of TCO. So it makes sense to pay the premium. Amazon: 14% perf increase of 2-4% TCO.
*Cloud important: 52% CAGR, 94% share.
*More non-CPU revenue (switch, memory, photonics,...), from 12 to 22% in 2018.
*She refers a lot to Jevons' paradox.
*Network/telco: fast move from proprietary to IA. 7.5% share.
*Acquisitions to accelerate roadmap, gain IP etc.
*33% Xeon Phi coprocessor share, just 2 years after launch, which was 5 years after competitor. Investments in Phi. Knights Landing next year, high ASP (>$1000). Transistor count: 7.2B transistors.
*Infiniband 50% growth. Omni path design wins.
*Big Data. TAM will increase to $41B (from ~$13B) in 2018, >50% of that 41B is hardware.
*'Basis' demonstration: She has worn Basis watch. 'Neuron' will help her to improve her sleeping. Gave information and made predictions.
*She likes Mica better than Basis watch.
*Custom and lots of SKUs, Grantly.
*Silicon photonics. Started shipping last month. For data center (fibre optic replacement), and in 2015 HPC and rack scale architecture (dense server) (--> both copper replacement, improves 100Gb/s from 3m to 2km). 30% of HPC is for I/O. Photonics superior. Forecast: 100CAGR through 2018. Will be integrated in CPU through Moore's Law (after fabric).
*Reduce sever capex to increase Intel silicon spending.
*Competition taken seriously: 12 ARM vendors, down from 16 last year. IBM, AMD. Committed to the market to sustain TCO + performance leadership.

I don't have much knowledge about this, but it seems very interesting, certainly the last photonics part.

Now Kirk .

 

[III-V]

I think this is the first time in official context that Intel has talked about Air Gap interconnect stack. Nice picture in the Holt slides.

Officially, yeah. They submitted a late paper to IEDM 2014 that David Kanter picked up on. Looking forward to reading it. Actually, I wonder if it's already available on IEEE Xplore...

 

[Fjodor2001]

Thanks for the summary Witeken! Good work. Some questions that I have:

*Yield graph extended, in healthy range, but below expectations from 2013 (BTW: 22nm has highest yield ever), catching up. Yields will match in ~H2.

Were there any specific numbers mentioned, or a graph presented?

*Recap of 2X perf/watt improvement.

Was that in general, or only for certain segments (e.g. Broadwell Y Core M)? I suppose it will not be true for e.g. Broadwell-K?

*20nm planar: higher leakage, more power, etc. Competitors pause at "16nm" to do FinFET.

Did they forget about Samsung 14 nm?

Also, what does it mean that competitors "pause at 16 nm"? Doesn't e.g. TSMC have 10 nm and beyond in their pipeline/plans?

 

[ViRGE]

Presentations in PDF form that have been posted so far:

http://intelstudios.edgesuite.net/im/2014/pdf/2014_Intel_IM_Krzanich.pdf

http://intelstudios.edgesuite.net/im/2014/pdf/2014_Intel_IM_Holt.pdf

Very interesting. Where exactly are these getting posted, anyhow?

 

[witeken]

Kirk Skaugen (PC Client Group):

*Innovation to transform PC.
*Performance matters.
*New user prices.
*Revenue and operating margin up, compared to flat/decrease in 2013.
*Positive unit growth in 2014 (4 consecutive quarters) and beyond: XP refresh, tablet saturation; productivity on 7" tablet lower time to buy notebook.
*1.8B installed base. 600M 4+ years old.
*They do 250.000 interview per year. Refresh is getting longer, but installed base bigger.
*17CAGR of all-in-one (like iMac), price to $499.
*20 CAGR of mini PCs (like NUCs).
*Compute sticks announcement(?): plug in tv or smart monitor and get Intel compute power.
*Big tablets.
*Notebook innovation: more colors, all OSes (for decades it was Wintel). SteamOS: 100M users.
*Also interesting to note: lines between form factors are blurring.
*2 in 1s (150% up this year and people want them, according to interviews, represent 10% of all 9"-12" market, below $699), touch, Chrome support.
*First new brand introduced in 5 years: Core M. 2X performance of ARM. Demo of 4 year old laptop vs Lenovo Yoga Pro 3.
*Bay Trail: Chrome share from 24% in Q2'13 to 65% and lower price points. Example: Samsung's internal Exynos -> IA. 83% of Celeron and Pentium brand is BT.
*Record vPro (security) shipments in 2014 (e.g. in 2-in-1).
*WiGig, providing silicon for technology.
*14nm valu Braswell.
*Fast mainstream 14nm ramp. Raising system price with Iris Pro graphics.
*Switching gears to performance: (2-3X Core vs ARM; S801-Nvidia K1). Optimizing Core for Chrome, i3 systems.
*RealSense integration: acquired 3D companies, need for performance (demo). Dell Venue 8. Multiple use cases (65+ applications from multiple big companies). For example: getting dimensions of a box instead of measuring by hand or buying expensive tools. 15 designs in spring, more with Skylake.
*Enthusiast: loyal consumer base. Built products for them: Devil's Canyon and 8 core (16 thread) Haswell-E (4k demo).
*China gaming stadium (20k seats).
*5th gen Iris Pro . 100X increase since 2006. Building a brand around Iris. Gaining share, driving ASPs. He showed some notebooks and mini PC with Iris Pro (HSW).
*Roadmap: 5th gen Broadwell: spring. Braswell and Skylake (6th gen Core): H2'15.
*Getting rid of passwords and wires (power brick, HDMI, USB,...), Skylake systems without ports. (demo)


Strange that the Q2 roadmaps that appeared this year seem to be false.

Now Q&A with Kirk and Diane.

 

[Roland00Address]

*For the first time, 100M CPUs shipped in 1 quarter (Q3).

To Give you an idea on how much market share this is.
75 million to 80 million PC sold each quarter (gartner)
315 million smartphones shipped each quarter (google io, note this number is all smartphone shipped including android and ios).

Does anyone believe in this? Cherry Trail this year is nonsense imho. Otherwise we would have much more infos. There is nothing. Braswell has been delayed into Q3 2015. For Cherry Trail Q2 2015 is a realistic target.

They will be shipped to OEMs, but that does not mean OEMS will release designs to the general public by then. It will take times to develop the devices, get the logistics done, and then ship to the end stores. The chance of people getting cherry trail devices under the christmas tree is very slim. Now them being announced at CES and shipping to end stores like amazon and best buy in Q1 2015 is extremely likely. It may not even be January but February due to

*Bay Trail now over 50% of Celeron and Pentium.

This is depressing. I have no problems with Bay Trail, but using Bay Trail for traditional size laptops and desktops suck.

*BOM delta: from $15 to BT's entry $8. No contra-revenue or BOM delta in SoFIA (will improve margins).

So they are still giving the chips away currently. We do not know the real price OEMs are paying for atom chips but the tray price is $17 to $41. Thing is the OEMs do not pay the tray price, they pay much less than the tray price. For example according to the tray price baytrail Celerons (atom architecture) tray price goes up to $107 and the baytrail pentiums tray price go up to $161

*Intel will develop less own smartphone SoCs, will invest to let others design SoCs and take them to market. Interesting.

Very interesting by saying that Intel is admitting they are not sure they can keep their fabs at capacity with the status quo of the last 2 years. Or they are trying to deny their competitors foundry customers to reduce their profits and thus decrease their ability to do R&D to future nodes.

Sounds like they are either licensing out their IP (rockchip), or let other people use their fabs without using their IP (altera)

*Competition taken seriously: 12 ARM vendors, down from 16 last year. IBM, AMD. Committed to the market to sustain TCO + performance leadership.

I am curious if anybody has a list of these? Where they referring to serious ARM licenses, or foundries that produce serious ARM business?

 

[III-V]

Very interesting. Where exactly are these getting posted, anyhow?

On the webcast page. I'd imagine it'd need a refresh periodically, as they're adding new ones.

 

[III-V]

Thanks for the summary Witeken! Good work. Some questions that I have:


Were there any specific numbers mentioned, or a graph presented?

Graph available from the second link I posted (Holt). Check the webcast page. The pdfs there answer all your questions.

Whoops, double posted.