Intel has $55.9B record year, ships 46M tablets

[rootheday3]

in 2014, most of Intel's mobile sales volume was nearly all Baytrail, with a little from Merrifield/Moorefield in the tail of the year, a little from discrete modem sales (like the one in recent Samsung tablet) + a winding down tail of 2G phone SoCs from the former Infeon line.

Let's say cost of goods sold and contra revenue wash out nearly all that revenue. Where is the rest of the loss coming from?

Answer: R&D.
Baytrail got updated with Cost Reduced sku, drivers updated to KitKat, drivers updated to Lollipop - all the engineering for that falls under R&D.

Intel developed the IRDA stack to expedite OEM BOM customization and adoption - also R&D.

Intel's WPRD division is making large investment in wireless modem technology to ramp a number LTE modem products which are just coming to market now - That R&D investment was in 2014 but revenue from the products comes in 2015 and beyond.

Also, Merrifield, CherryTrail and Broxton/Willowtrail and several generations of SofIA line were all in development throughout 2014 - but come to market in 2015 or later.

We are talking about thousands of hw and sw engineers, lots of mask creation, reference motherboards...

In other words, to make relevant products which will win market segment share in the future, you have to invest money now. To catch up in a market where you are behind, you have to invest a lot in a hurry.

In effect, Intel is in startup mode on mobile and is racing to try to get to a position where its products will:
a) have the right features, performance, cost structure and hit the market windows
b) and build the expertise, OEM relationships
c) so that they can sell in high volumes
d) to amortize development costs and keep the fabs full

In the mean time, contra revenue was a way to address some the fact that original Baytrail was a "netbook" chip and thus was designed such that it required OEMs to add other components besides the SoC that drove up platform costs. Baytrail CR corrected some of that gap and didn't require as much contra revenue. Cherrytrail and beyond were designed with total platform BOM in mind. Using contra revenue to avoid having zero unit sales in 2014 seems like a reasonable strategy to ensure that Intel can learn the mobile business, build the sw and hw IP expertise, etc (bullet a and b above).

It also ensured that Intel wasn't marginalized with "zero percent" share in mobile... instead they were able to get ~20% overall tablet SoC share. This helps with motivating ISVs port apps on Android to x86 and to grow the Windows and ChromeOS volume so that there is an ecosystem to support further adoption of Intel products in the future.

 

[tential]

While anandtech "CEOs" cry about Intel's mobile division and spread gloom and doom about it being "unsustainable" blah blah blah, Intel posts RECORD revenue of $55.9B....

You guys keep worrying about Intel's Financial strength... they'll keep hitting record years for making money.

 

[AtenRa]

A few interesting points,

Tablet revenue was measured at the PCCG(PC Client Group) in Q4 2014.

"PC Client Group: Delivering platforms designed for the notebook (including Ultrabook™ devices and 2 in 1 systems), the desktop (including all-in-
ones and high-end enthusiast PCs), and tablets; wireless and wired connectivity products; as well as home gateway and set-top box components"

and

- Notebook platform volumes increased 11% from 2013 to 2014
- Notebook platform average selling prices decreased 7% from 2013 to 2014
- Desktop platform volumes increased 3% from 2013 to 2014
- Desktop platform average selling prices increased 2% from 2013 to 2014

So while Notebook ASP was down, Desktop was up meaning more high end parts were sold in desktop and cheaper parts in Notebook.

Also, best group was the DATA Center.
Volumes were up 8% in 2014 vs 2013 and ASP was 10% higher in 2014 vs 2013. Not only they increased volumes, they increased ASP even more.

 

[dahorns]

And a lousy quarter one forecast, i wonder whats going on

Lousy? Margins are lower, mainly due to 14nm startup, but other than that the Q1 guidance looked fine. Revenue of 13.7 billion would be a 900 million dollar increase from Q1 2014, a 7% increase year-over-year.

 

[witeken]

According to Rueters global PC shipments fell 2.4% Q3-Q4.
According to Rueters Intel's PC group sales fell 3% Q3-Q4.

If Intel's sales declined more than the market then somebody else had to pick up those additional sales Intel lost. In this case there's only one somebody else.

This is plain wrong:

PC CPU division performance: Revenue +3% Y/Y to $8.9B. Volumes +6%, ASPs -2% (mix shift to Atom). Desktop volumes +1%, ASPs flat. Notebook volumes +11%, ASPs -3%. Op. profit +18% $3.98B.

http://seekingalpha.com/symbol/INTC/news

 

[Nothingness]

This is plain wrong:

PC CPU division performance: Revenue +3% Y/Y to $8.9B. Volumes +6%, ASPs -2% (mix shift to Atom). Desktop volumes +1%, ASPs flat. Notebook volumes +11%, ASPs -3%. Op. profit +18% $3.98B.

http://seekingalpha.com/symbol/INTC/news

Phynaz was talking about Q4 vs Q3.

 

[witeken]

Is that an official number? Missed that information :$

For all the people complaining about the contra-revenue, here are some facts:

*BOM delta was $15 or (probably for the low-end even more. Throughout the year, they released new Bay Trails with fewer components = low BOM delta. At the end of the year (probably for wider adoption this year), they released Bay Trails with $5 BOM subsidy. (Source: IM'14 and some slide from AtenRa)

*At Investor Meeting 2013, they announced how much the contra-revenue would impact gross margin, and if you did the math (with the flat revenues they were forecasting back then), you come up with $800M or so.

*There is no other subsidy besides paying the BOM delta (maybe marketing?), so if a company chooses Intel, it's because of the performance and features and competitive pricing.

 

[frozentundra123456]

Yea, kind of ironic. Certain posters keep complaining about intel fans hijacking AMD threads, while it seems every thread about intel profits, no matter how positive, degenerates into how they cheating with contra revenue.

In fact, I do have mixed feelings about it, but it certainly is leading to more choices at a cheaper price in the tablet space. I dont think anybody has to worry about Intel becoming a monopoly or dominating the market in mobile. In fact I just bought a 60.00 Winbook tablet at Microcenter that is far superior in every way to a 200.00 Asus ARM tablet I bought a couple of years ago. I am sure this would not have been possible without contra revenue, and it seems like a win for me. Whether it is ultimately a win for Intel is yet to be seen, depending of whether they can maintain/increase market penetration.

 

[kimmel]

Ohh my bad then, but again, 70M for 2015 without contra revenue seems impossible.

I think there was some rumors saying that they had other goals internally but the one they committed to wall street was 40M.

Q1-Q3 conference calls were pretty annoying as half the call seemed to be how they were progressing against the 40M goal. People were pretty skeptical about even coming close to those numbers. 46M seems like Intel wanted to hit the number solidly but not overachieve to lose even more money. That is assuming they had more upside on the numbers and decided to throttle the subsidy.

Q4 call yesterday indicated that they are pretty happy with the current volume and would "grow at the market". One of the questions from the analyst was that if the tablet market would shrink would Intel tablet shipments also shrink. Quite the mindset change in one year for them to ask that. I think the goal for them to grow at the market is pretty understandable as they have enough volume to make Google and developers play nice with x86 but they don't want to support the contra revenue on much higher volumes. If there's one thing that Intel is familiar with, it is an ecosystem that doesn't want to play nice with others in because there's no customer volume to support.

 

[witeken]

Ohh my bad then, but again, 70M for 2015 without contra revenue seems impossible.

I'm not sure why you still don't understand, but contra-revenue is irrelevant!

Bay Trail was designed as a high-end chip. But then came BK and wanted to penetrate the market, so Bay Trail suddenly became a chip for low-end as well. But the cost structure wasn't there and it would take too long to make it right, so instead Intel decided to pay the BOM delta themselves.

If Intel didn't have a BOM issue, they wouldn't have paid a single dollar of contra-revenue in 2014 and they would still have sold 46M tablets.

The reason why I come up with a number of ~70M is following: Intel only sold ~5M and ~10M tablets in Q1 and Q2, while they sold ~15M and ~16M in H2.

It's quite implausible that tablet sales will drop to 5M again, so if they can maintain this 15M figure and have some uplift in H2 with SoFIA and Cherry Trail, then they can sell 70M or so in '15. It's an educated guess.

 

[witeken]

And a lousy quarter one forecast, i wonder whats going on

The lousy quarter is because of seasonal trends. It was the same in 2014.

 

[NTMBK]

But the cost structure wasn't there and it would take too long to make it right, so instead Intel decided to pay the BOM delta themselves.

No. They did not reduce the cost to match their opponents' BoM. They have been paying customers to take chips off their hands. That is what negative revenue means.

 

[witeken]

Phynaz was talking about Q4 vs Q3.

Sure, but I don't think the PC grew Y/Y, which means AMD had to lose share because Intel had 6% higher volumes.

 

[Arachnotronic]

While anandtech "CEOs" cry about Intel's mobile division and spread gloom and doom about it being "unsustainable" blah blah blah, Intel posts RECORD revenue of $55.9B....

You guys keep worrying about Intel's Financial strength... they'll keep hitting record years for making money.

I think this is an important point. Intel right now has businesses that throw off ungodly amounts of cash and it can still turn in HUGE profit numbers even with the mobile millstone around its neck.

What is a better way to leverage this financial strength: rest on your laurels for as long as you can milk it, or hurt that short-term profitability somewhat to invest in your long-term future?

Intel is doing the right thing from a long-term business perspective by doing all that it can to get into mobile.

 

[witeken]

No. They did not reduce the cost to match their opponents' BoM. They have been paying customers to take chips off their hands. That is what negative revenue means.

Indeed, that's what contra-revenue means. If you have to pay OEMs $15 for the additional BOM costs, and you sell you SoC (that already cost 5-10 bucks to manufacture) for less than that for low-end tablets, then you'll end up with negative revenue.

 

[dealcorn]

Most readers if this forum who are neither CPA's nor CFA's are not trained to understand the difference between GAAP losses and economic losses. As a preliminary observation, note the GAAP stands for "Generally Accepted Accounting Principals" and that definition intentionally does not include words like "economic" or "true". Using page 39 of the Stacy Smith's IDF 2014 pdf it is possible to come up with a "pretend" realistic analysis of the 2014 MCG losses. The intent of this analysis is illustrative rather than analytic.

2014 MCG income statement:

Allocation of "fixed" R&D ($2B)
MCG R&D ($1B)
Contra Revenue, BOM adj, etc ($1.1)
Everything else (net) $0.1
_____
GAAP income (loss) ($4B)

What I call Allocation of "fixed" R&D is comprised of Shared Platform Investments and also Foundational Investments such as process technology. Witeken's point is that even if Intel shut MCG. Intel's total spending on "fixed" R&D would not change; the costs would just be allocated elsewhere. Intel will continue to invest in process technology and make Foundational Investments even without MCG. That is why the economic loss of MCG may be about ($2B) rather than the GAAP loss of ($4B).

From an accounting theory perspective, the ($2B) loss is way overstated. The $1B spent on MCG R&D does not benefit Bay Trail. That money is spent to develop IP necessary for future dominance of mobile markets. Accounting theory suggests these cost be capitalized and amortized over the estimated useful life of the IP developed. This perspective suggests the MCG economic losses are closer to ($1B).

While I understand that the celebration of Intel losses may be emotionally liberating, it is kinda silly. Prudent Intel investors do not cares if Intel's cumulative GAAP MCG losses total $12B before reaching break even. What matters is whether Intel can gain sufficient market share for MCG to generate $2.5B or more in profits per year in due course. Given the size of MCG target markets and the surprisingly strong market acceptance of Intel offerings that goal looks achievable. I have no doubt that with patient investment over time, Intel will figure out how to properly manage it's cost structure so it may economically dominate these markets with the most advanced process technology. It is expensive but they will get there.

Having gone this far, I might as well ask the interesting question. GAAP requires that R&D costs that benefit multiple business units be allocated among the business units using a "reasonable" allocation methodology. Intel clearly is not allocating "fixed" R&D based on revenue. What is the reasonable allocation methodology that results in a $2B allocation to MCG?

 

[Piroko]

Honestly, there's one thing I have a hard time understanding with Intels current direction.

As far as I understand, their goal with Atom is to ...
...decrease the BoM. Integrate modems, integrate parts of the power delivery, reduce the package pinout, reduce the PCB footprint. That is a sound business plan, but they are only catching up to Qualcomm et al here.
...increase revenue through volume. Again, sound business plan, but again only catching up here.
...increase margins to match their remaining product portfolio through... Well, I'm a bit lost. To demand higher prices they will want to offer a superior product. And a superior product will most likely be defined as superior user experience - because, well, see current iphones.

And that is where I'm lost. Two years ago a new SOC would significantly improve pretty much anything you do with your Smartphone. Surfing, taking photos/videos, navigation, gaming, even calling (noise cancellation). But 2015? 2016? 2017?
I feel like the SOCs itself are quickly shifting to being commodities, differentiation will come from other hardware aspects and software. And commodities, supplied from more than two market participants, usually don't do 50% margins. Or 20%, for that matter.

 

[mrmt]

How much growth do you think Mediatek and QC would have had without Intel's dumping tactics?

Doesn't matter. Given Intel's minuscule market share, no regulator will punish them for giving chips away for free.

 

[NTMBK]

Doesn't matter. Given Intel's minuscule market share, no regulator will punish them for giving chips away for free.

Minuscule? We're looking at estimates of ~230mn tablets sold globally in 2014, meaning that Intel was in a fifth of them. That's not minuscule.

 

[mrmt]

Minuscule? We're looking at estimates of ~230mn tablets sold globally in 2014, meaning that Intel was in a fifth of them. That's not minuscule.

What about the entire SoC market for phones and tablets?

 

[witeken]

Minuscule? We're looking at estimates of ~230mn tablets sold globally in 2014, meaning that Intel was in a fifth of them. That's not minuscule.

The contra-revenue will fade away in 2015. This is all much ado about nothing.

It amazes me that almost no one notes how this is in fact Intel's best year ever for revenue -- and 20% or so of that money is net income.

 

[kimmel]

...increase margins to match their remaining product portfolio through... Well, I'm a bit lost.

You are lost because they never said that. They said they would stop losing money in mobile. That's all they said.

Frankly the real play here is to drive volume to Intel fabs vs TSMC/GloFo. Leading edge fabs are getting more expensive and if Intel can drive enough profit out of their competitors fabs they will slow the node progression. That will eventually allow Intel to sell chips with better margins / lower prices simply by the fact that they will be much smaller than the competition.

 

[R0H1T]

You are lost because they never said that. They said they would stop losing money in mobile. That's all they said.

Frankly the real play here is to drive volume to Intel fabs vs TSMC/GloFo. Leading edge fabs are getting more expensive and if Intel can drive enough profit out of their competitors fabs they will slow the node progression. That will eventually allow Intel to sell chips with better margins / lower prices simply by the fact that they will be much smaller than the competition.

If anything their lead is shrinking ~
http://forums.anandtech.com/showpost.php?p=37084147&postcount=273

When was the last time you saw a competitor launch products on the same node within a year? Their 10nm products will also be delayed unless they do some magic wand trick.

 

[kimmel]

If anything their lead is shrinking ~
http://forums.anandtech.com/showpost.php?p=37084147&postcount=273

When was the last time you saw a competitor launch products on the same node within a year? Their 10nm products will also be delayed unless they do some magic wand trick.

Keep up the good FUD fight.

 

[witeken]

If anything their lead is shrinking ~
http://forums.anandtech.com/showpost.php?p=37084147&postcount=273

When was the last time you saw a competitor launch products on the same node within a year? Their 10nm products will also be delayed unless they do some magic wand trick.

Some facts:

1) Objectively speaking, Intel's 14nm transistor is about 1.5x smaller.
2) TSMC's and Samsung's 16/14 nodes have only a first generation FinFET transistor, while Intel's been in volume production of their second generation since mid-'14. Transistor power and perf will be similar to Intel's 22nm, which entered HVM in... 2011: that's 4 years later (with iso yields).
3) Mark Bohr, Intel fellow, said (at IEDM last month) 10nm won't have problems because they learned from 14nm. You wouldn't say that if you weren't so confident (because Intel refuses to say much about 10nm except that it will have better cost per transistor reduction than historically, which is something no other company can claim.

5) TSMC and Samsung 10nm will go into meaningful production 1-2 years later and why they call is actually a glorified Intel 14nm node. So they will be much later and have much worse transistor.