Intel lost $1bn on mobile in Q3 '14

[NTMBK]

Hey, at least it was marginally less than last quarter.

 

[USER8000]

It just shows how strong they are in other areas,ie,servers and the like to swallow such losses.

 

[swilli89]

Looks kinda odd to me because they are basically reporting zero revenue from mobile to begin with

 

[NTMBK]

Looks kinda odd to me because they are basically reporting zero revenue from mobile to begin with

Yeah, does make me wonder what the heck falls under "mobile". I guess their modems, Android tablets and phones- Windows 2-in-1s fall under "PC Client Group".

 

[Nothingness]

Yeah, does make me wonder what the heck falls under "mobile". I guess their modems, Android tablets and phones- Windows 2-in-1s fall under "PC Client Group".

Their modems are in the Mobile group:

Mobile and Communications Group: Delivering platforms designed for the tablet and smartphone market segments; and mobile communications
components such as baseband processors, radio frequency transceivers, Wi-Fi, Bluetooth*, global navigation satellite systems, and power
management chips.

And 2-in-1 are indeed in PCCG:

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

 

[AtenRa]

Looks kinda odd to me because they are basically reporting zero revenue from mobile to begin with

They are selling at almost ZERO price, that is they are selling you one SOC at $20 but give you back $19 as Contra Revenue. So revenue at the end is $1.

Add the SoCs sold(cost) + R&D + MG&A and they loose 1B each quarter.

 

[Abwx]

They are selling at almost ZERO price, that is they are selling you one SOC at $20 but give you back $19 as Contra Revenue. So revenue at the end is $1.

Not true, with 15 millions chips delivered if they got 1$/chip that would amount to 15 millions $, to compare with the 1 million they got from thoses so called sales.

Do the calculations with 1 bn losses distributed in 15 millions chips.

 

[AtenRa]

Not true, with 15 millions chips delivered if they got 1$/chip that would amount to 15 millions $, to compare with the 1 million they got from thoses so called sales.

Do the calculations with 1 bn losses distributed in 15 millions chips.

Yea, i only gave an example to understand how Contra Revenue works

Obviously selling 15M SoCs with 1M of revenue means you sold each one of them at close to $0.066.

 

[III-V]

Hey, at least it was marginally less than last quarter.

Yeah, and it'll continue to be a huge loss until they can get Cherry Trail out.

 

[AtenRa]

Yeah, and it'll continue to be a huge loss until they can get Cherry Trail out.

And what will happen with Cherry Trail to make any difference ?? Do you believe they can Sell Cherry Trail at $2 (no Contra Revenue) and make a profit ?? Because they will have to sell at those low price in order to keep selling millions of SOCs, otherwise sales will plummet.

 

[III-V]

And what will happen with Cherry Trail to make any difference ?? Do you believe they can Sell Cherry Trail at $2 (no Contra Revenue) and make a profit ?? Because they will have to sell at those low price in order to keep selling millions of SOCs, otherwise sales will plummet.

Much, much better cost optimizations. Don't play dumb.

 

[AtenRa]

Much, much better cost optimizations. Don't play dumb.

Again, they are currently selling ATOM SOCs at $2 or less, how they will manage to sell more Cherry Trails if it will cost more to the OEMs than what they pay today ???

 

[Khato]

One thing to keep in mind with respect to MCG losses - part of the reason why they're quite so high is that they get charged for PEG (platform engineering group) costs when they're the lead customer for a design. So despite the fact that PCCG is going to be using Cherry Trail under the Braswell name MCG gets hit with basically all the development costs while PCCG gets it for 'free'. Note that Intel could always reverse this in order to make MCG look better...

Anyway, it'll be interesting to see how MCG sales change during the coming year. The possibility of starting to get back some modem sales in addition to the fade-away of contra revenue following the 40M tablet goal + more cost-competitive SoCs should improve revenue at least.

 

[Khato]

Again, they are currently selling ATOM SOCs at $2 or less, how they will manage to sell more Cherry Trails if it will cost more to the OEMs than what they pay today ???

$2 SoC + $14 cost adder over ARM SoC > $10 SoC + $5 cost adder over ARM SoC. No idea on the exact numbers, but that's how.

 

[lefty2]

Yea, i only gave an example to understand how Contra Revenue works

Obviously selling 15M SoCs with 1M of revenue means you sold each one of them at close to $0.066.

It's more probable that some SoC/modem sales are making profit, while others are less than $0 (i.e OEM is effectively being paid to use Intel SoC)

 

[dahorns]

Again, they are currently selling ATOM SOCs at $2 or less, how they will manage to sell more Cherry Trails if it will cost more to the OEMs than what they pay today ???

I thought this was well covered ground. Right now, ignoring the SOC costs, it costs the OEMs more to produce a device using Bay Trail. This is the higher bill of material cost that results from the fact that Bay Trail wasn't designed from the ground up to go into the tablet market. This is the reason contra-revenue exists. To compensate the OEMs for those other costs.

As Intel improves their SOC design, the OEMs will no longer have to spend the additional money on the bill of materials. That money will then return to Intel as contra-revenue is reduced.

Basically the cost to OEMs looks like this (numbers for example only):

$20 Bay Trail + $30 BoM - $20 Contra-Revenue = $30 to use Bay Trail; or
$20 ARM SOC + 10 BoM = $30 to use ARM SOC

As Intel improves on the BoM problem it will look like this:

$20 Intel SOC + $10 BoM = $30 to use Intel; or
$20 ARM SOC + $10 BoM = $30 to use ARM SOC.

Your argument that the OEMs will in effect tell Intel that they won't use their products unless they continue to get the SOCs for free is shallow. The OEMs will use whatever product makes them the most money overall.

 

[el etro]

A Billion for intel is nothing...

 

[III-V]

Again, they are currently selling ATOM SOCs at $2 or less, how they will manage to sell more Cherry Trails if it will cost more to the OEMs than what they pay today ???

Whoever said that they need to sell more? Their market share is fairly respectable right now.

 

[witeken]

Quite a boring earnings call. Not much interesting technology question, and the ones asked are left unanswered because they don't want to talk about 2015.

BTW, 80% of their tablets are Android.

 

[NTMBK]

BTW, 80% of their tablets are Android.

They seem to distinguish between "tablets" and "2-in-1s"- most Windows "tablets" are in fact 2-in-1s, like the Transformer T100. Or is it only a 2-in-1 if the keyboard is permanently attached, like the Yoga?

 

[cbn]

Yeah, and it'll continue to be a huge loss until they can get Cherry Trail out.

While Cherry Trail will be an improvement over Bay Trail, I don't believe four small cores will be the answer. Sure, the processor will be good....but I don't think it will be as good as it could have been.

Instead of four small cores, I would have rather seen Intel consolidate Cherry Trail/Braswell and Core based Celeron lines into one SOC SKU with a single fully enabled big core with HT enabled.

Give me that with Core M style power managemnt and AVX/AVX II over four little cores without AVX any day of the week for tablets. Likewise for the 17 watt and 53 watt SKUs, I would much rather have a much higher clocked single big core with HT (and AVX/AVXII) over the current heavily disabled and downclocked 2C/2T arrangement.

 

[cbn]

Another reason I would like to see Intel move to single big core SOC is that I believe it could help them dedicate more resources to high drive current/low leakage type transistors. This rather than having a good amount of the product line-up (low end/high volume tablets) on low drive current/low leakage.

With that mentioned, I certainly don't believe in atom leaving the Intel line-up. I would just rather see atom confined to phones, wearables, micro-servers (Avoton, etc) and IoT rather than being used in tablets.

 

[AtenRa]

I thought this was well covered ground. Right now, ignoring the SOC costs, it costs the OEMs more to produce a device using Bay Trail. This is the higher bill of material cost that results from the fact that Bay Trail wasn't designed from the ground up to go into the tablet market. This is the reason contra-revenue exists. To compensate the OEMs for those other costs.

As Intel improves their SOC design, the OEMs will no longer have to spend the additional money on the bill of materials. That money will then return to Intel as contra-revenue is reduced.

Basically the cost to OEMs looks like this (numbers for example only):

$20 Bay Trail + $30 BoM - $20 Contra-Revenue = $30 to use Bay Trail; or
$20 ARM SOC + 10 BoM = $30 to use ARM SOC

As Intel improves on the BoM problem it will look like this:

$20 Intel SOC + $10 BoM = $30 to use Intel; or
$20 ARM SOC + $10 BoM = $30 to use ARM SOC.

Your argument that the OEMs will in effect tell Intel that they won't use their products unless they continue to get the SOCs for free is shallow. The OEMs will use whatever product makes them the most money overall.

OEMs and companies worldwide started using ATOM SOCs not because they are faster or for anything else but ONLY because both SOC + BOM was cheaper than ARM SOCs. In 2013 Intel only sold 1M of ATOM SoCs for Tablets. In 2014 they will sell more than 40M because OEMs and everyone else gets the SOC + BOM + technical advice almost for free.
And next year Intels plans are to reach 80M Tablets and that without Contra Revenue, i dont see how they will manage this with Cherry Trail on a very expensive 14nm.
But there is the rebate strategy as well, OEM will get a rebate next year if it will buy certain number of SOCs. So even without Contra Revenue for 2015, they can lure OEMs in to using Cherry Trail and aim for rebates for 2016 and so on.

 

[AtenRa]

Whoever said that they need to sell more? Their market share is fairly respectable right now.

Intels plans are to reach 80M tablets per year.

 

[dahorns]

OEMs and companies worldwide started using ATOM SOCs not because they are faster or for anything else but ONLY because both SOC + BOM was cheaper than ARM SOCs.

Is this based on anything besides conjecture? And even if it is cheaper, the question is how much cheaper. Do you really think Intel is doing all this to set itself up in a line business that it can't make money at?

I'm going to assume they've targeted a price point that is competitive with the ARM community and will still make them some money (after reduction of contra-revenue).