TI pulling out of smartphones and tablets

[dagamer34]

The Verge:

Chipmaker Texas Instruments (TI) has announced a major shift in its business strategy, moving away from consumer electronics devices such as smartphones and tablets and towards embedded applications such as computer systems in cars. Speaking at an investor meeting on Tuesday, VP for embedded processing Greg Delagi described the consumer electronics market as "less attractive as we go forward," according to a report from Reuters.

With no major smartphone products coming out in the US because of LTE requirements from pretty much all new phones on carriers this year, it seems the market has squeezed TI out. I guess they thought that between dual A9 in OMAP4 and dual A15 in OMAP 5, they would be able to hold in 2012 so that they could come out strong in 2013, but that seems not to be happening.

This also explains why Toshiba wasn't going to be making any Windows RT tablets, there's no reason to start production of a tablet when your chip supplier has decided to leave the game. Makes me wonder if/when Microsoft is going to approach Samsung to put Exynos 4/5 support in Windows RT...

 

[dguy6789]

I was very surprised to read this. Their smartphone processors have almost always been pretty great. OMAP5 is going to smoke the S4 and the A6. Less competition is bad.

 

[dagamer34]

I was very surprised to read this. Their smartphone processors have almost always been pretty great. OMAP5 is going to smoke the S4 and the A6. Less competition is bad.

Based on what? We never saw an OMAP5 with final clocks. We never heard of any OMAP5 design wins for next year. Cortex A15 isn't finished but S4 and A6 are shipping now. We can't just say things we hope or think to be true, but prove to be true. So far, we've only seen an OMAP5 @ 800Mhz beat a "competitor's tablet" (likely Tegra 3) at loading webpages, but that was a few months ago. No doubt nVidia is working on Tegra 4, so beating a year old chip is NOT an accomplishment.

But realistically, this isn't about SoC performance, but the fact that LTE gave Qualcomm a HUGE leg up since nearly every LTE phone uses a Qualcomm baseband, and up until the iPhone 5, if you wanted great LTE battery life, you had to use a Qualcomm SoC that only came out just 6 months ago. That's a LONG time in the smartphone game as we can see today.

 

[lothar]

Very sad news.
Why can't they use their own SoC and use a separate Qualcomm LTE chip like Apple?
Qualcomm wouldn't sell it to them?

 

[Bateluer]

Seems like they're moving away from a massive and still growing market to a relatively small, more niche market. Doesn't seem the wisest choice, and seems to send the message that they're giving up trying to compete with Qualcomm.

 

[dagamer34]

Very sad news.
Why can't they use their own SoC and use a separate Qualcomm LTE chip like Apple?
Qualcomm wouldn't sell it to them?

The iPhone 5 I believe is the first product shipping en masse with the MDM9615 and Apple is probably taking a significant chunk of that.The problem is there is no TI OMAP SoC that can compete with the Tegra 3 or S4 right now, and their competitor won't arrive for another 6 months.

With embedded products, cost is the biggest fact and performance only needs to be "good enough".

 

[Suspicious-Teach8788]

What's wrong with this?

OMAP was used by who? Motorola? I'm not even sure why Google picked OMAP for their GNex. It was probably because of some under the table business. Exynos would've been the right choice given its popularity in the SGS phones. But nope.

I think this brings good news in that Google will be picking between Samsung and Qualcomm this round.

At least with the Nexus S, all development efforts helped the SGS1 phones. It's the reason why SGS1 phones have CM10 running well while SGS2 is still struggling to get it to even work. The SGS3 got a break because of a leak and now JB is dropping anyway.

 

[Bateluer]

What's wrong with this?

OMAP was used by who? Motorola?

Amazon, B&N both use OMAP.

 

[dguy6789]

Based on what?

OMAP 5 is Cortex A15 based. It's not exactly complicated.

 

[Suspicious-Teach8788]

Amazon, B&N both use OMAP.

and those happen to use some modified OS or whatever crap.

The most popular android phones don't, which to me means that google should be aligning their Nexus device with the industry standard. I don't mind OMAP being in the game, and if they succeeded, that would've been great too. I do have beef with Google picking less popular chipsets like Tegra 2 and OMAP as launch partners for their "Nexus" devices when Samsung's pushing out volumes more with Exynos and HTC with Qualcomm.

Edit: I guess from a developer's perspective, OMAP > Snapdragon > Tegra > Exynos but if no one adopts OMAP, how does this matter? I like Google pushing initiatives like to go to soft keys only and no microSD, but no one's adopting.

 

[lothar]

What's wrong with this?

OMAP was used by who? Motorola? I'm not even sure why Google picked OMAP for their GNex. It was probably because of some under the table business. Exynos would've been the right choice given its popularity in the SGS phones. But nope.

I think this brings good news in that Google will be picking between Samsung and Qualcomm this round.

At least with the Nexus S, all development efforts helped the SGS1 phones. It's the reason why SGS1 phones have CM10 running well while SGS2 is still struggling to get it to even work. The SGS3 got a break because of a leak and now JB is dropping anyway.

Price?
If Google didn't, I doubt a Galaxy Nexus would be selling for $350 today like it has been doing since about March or so.

 

[cl-scott]

Seems like they're moving away from a massive and still growing market to a relatively small, more niche market. Doesn't seem the wisest choice, and seems to send the message that they're giving up trying to compete with Qualcomm.

Chip fabrication and design is VERY expensive, so it may not really be the dumbest move in the world. Might be better to dominate a niche market than spend huge amounts of money on the phone and tablet market where there would probably be little in the way of profit.

 

[Suspicious-Teach8788]

Price?
If Google didn't, I doubt a Galaxy Nexus would be selling for $350 today like it has been doing since about March or so.

Does choosing Exynos mean an extra $100 for cost? I don't think so. These SoCs are just razor thin margins for chip suppliers. The overall effect of cost wouldn't be that big of a difference. And come on, Google selected Tegra 3 for the Nexus 7. OMAP isn't the only choice.

 

[lothar]

Does choosing Exynos mean an extra $100 for cost? I don't think so. These SoCs are just razor thin margins for chip suppliers. The overall effect of cost wouldn't be that big of a difference. And come on, Google selected Tegra 3 for the Nexus 7. OMAP isn't the only choice.

Was the GSM Nexus S selling for $350 out of pocket three-four months after it's initial release?
If not, then there is your answer.

It is not the only choice, but it is most open from what I hear.

 

[wirednuts]

um.. there is a zillion cars being made every year and i would imagine they get much better margins on those deals. and, those contracts are much more reliable i would imagine. sucks to see TI go but i dont question why. theyre moving from a highly competitive volatile market to a much more reliable one.

 

[Suspicious-Teach8788]

Was the GSM Nexus S selling for $350 out of pocket three-four months after it's initial release?
If not, then there is your answer.

It is not the only choice, but it is most open from what I hear.

What are you talking about? The Galaxy Nexus sold for $350 for who knows WHAT reasons. Google's clearly trying to lure people in. After all the Nexus phones aren't that popular to begin with.

We all know phones cost ~$200 to make (see iPhone 5 BOM), but in general they cost anywhere from $500-$600 for a brand new one. SGS2 prices were at $550 before the SGS3 release.

I'm talking about THIS quote:

Price?
If Google didn't, I doubt a Galaxy Nexus would be selling for $350 today like it has been doing since about March or so.

How does picking OMAP result in a $350 device? I don't see it. The A6 processor is $17.50 in the iPhone 5 BOM. I don't see how picking Exynos or Tegra or Snapdragon results in a jump in price. And even if it's a higher price, it's not a $250 difference to make that $350 => $600. The under the table business I was referring to was probably something like some TI luring Google in to use OMAP. Why would they force their hardware partner, who's already used two generations worth of Exynos devices to suddenly develop an OMAP based device? Someone like Motorola probably has better experience with OMAP. Anyway I was just a little upset at picking OMAP knowing how good the Exynos chipset was in terms of performance and graphics power. Perhaps OMAP is easier for developers, but at the same time a Nexus device should also represent the hardware that vendors are going with--that just happens to be Snapdragon and Exynos.

I don't see how comparing the Nexus S and Galaxy Nexus pricing even means anything?

 

[WelshBloke]

and those happen to use some modified OS or whatever crap.

The most popular android phones don't, which to me means that google should be aligning their Nexus device with the industry standard. I don't mind OMAP being in the game, and if they succeeded, that would've been great too. I do have beef with Google picking less popular chipsets like Tegra 2 and OMAP as launch partners for their "Nexus" devices when Samsung's pushing out volumes more with Exynos and HTC with Qualcomm.

Edit: I guess from a developer's perspective, OMAP > Snapdragon > Tegra > Exynos but if no one adopts OMAP, how does this matter? I like Google pushing initiatives like to go to soft keys only and no microSD, but no one's adopting.

Why the hell would you like the bolded?

 

[dagamer34]

OMAP 5 is Cortex A15 based. It's not exactly complicated.

Please show me a benchmark with Cortex A15 at final clocks beating the A6 or S4. I don't care what you believe to be true, just what you can prove with objective tests. Otherwise, we might as well start reviewing phones subjectively by how pretty they look.

 

[vshah]

What's wrong with this?

OMAP was used by who? Motorola? I'm not even sure why Google picked OMAP for their GNex. It was probably because of some under the table business. Exynos would've been the right choice given its popularity in the SGS phones. But nope.

TI has historically been the best about publishing specs and developer documentation for it's SOCs. maybe that had something to do with it.

 

[antef]

Was the GSM Nexus S selling for $350 out of pocket three-four months after it's initial release?
If not, then there is your answer.

It is not the only choice, but it is most open from what I hear.

The Nexus didn't drop to $350 until Google I/O at end of June 2012, 7 months after initial release. When it debuted on the Play Store it was $400.

 

[dagamer34]

The Nexus didn't drop to $350 until Google I/O at end of June 2012, 7 months after initial release. When it debuted on the Play Store it was $400.

And that debut was only in April, 6 months later. When it launched in the UK, it was £549, which if you adjust for the UK VAT and rough dollar conversions, would have cost you about $700 (of course, since we are talking about phones which only are $200 in parts, price is really arbitrary and whatever the market will bear).

It probably would have cost $649 off contract if the GSM version were available in the US at launch.

 

[Suspicious-Teach8788]

Why the hell would you like the bolded?

Well the specific feature itself is meh, but what I'm trying to say is I like Google taking the initiative to direct hardware changes. It might not be everyone's cup of tea, but it's one step in clamping down on fragmentation.

Unfortunately it's like Google's directives go through one ear and out the other. HTC and Samsung still have hardware keys and the 3 button arrangement on the One X is different than the 3 buttons on the SGS3, which is different from the 3 soft keys on the Galaxy Nexus.

TI has historically been the best about publishing specs and developer documentation for it's SOCs. maybe that had something to do with it.

Fair enough. I did note that, but while this is good as a developer phone, the flagship Android phones don't use OMAP. So then what? This goes along the lines of Google pushing no microSD, soft keys, etc, but manufacturers are still doing their own thing. I think the buttons is the biggest kind of fragmentation we have. Not only did every phone have a different order for the 4 Android hard keys, now we are seeing different 3 button keys on different phones. It's ridiculous.

 

[Red Storm]

Seems like a pretty stupid move. Giving up a rapidly expanding market for one that does not seem like it will take off is just dumb.

 

[dagamer34]

Seems like a pretty stupid move. Giving up a rapidly expanding market for one that does not seem like it will take off is just dumb.

It's not as simple as that. Pretty much every SoC developer is dependent on Qualcomm for LTE baseband chips, and they build in their latest modems into their SoCs giving them a 6 month lead. For example, MDM9615, a 28nm 2nd gen baseband is just now available for phone OEMs to use, but that IP has been in MDM8960 SoC for 6 months. If you're building a phone, I don't see how you can even think about considering a nVidia or TI SoC in the United States when the carriers demand good LTE phones. And next year, we're going to have LTE-Advanced rolling out with carrier aggregation and using last years baseband puts your solution at a distinct disadvantage.

This doesn't phase Samsung which builds their own SoC for their own very popular phones, but LG, Motorola, Sony, etc... aren't going to bother with nVidia and TI when Qualcomm clearly offers a superior solution. It's why you saw nVidia buy Icera a year ago with hopes of integrating an LTE modem on die, but that's not coming until 2H 2013. TI showed no ambition in creating integration on that scale.

While the smartphone market has amazing growth now, it's important to keep in mind that most people only own one cell phone, but can have several different "smart" devices. When you start thinking about the "Internet of things" that need good SoCs at low cost, selling those at volume could be a very profitable business. I imagine their chips being in cars, TVs, coffee makers, refridgerators, washing and drying machines, heck, why not even blenders, ovens, microwaves, etc... And the best thing about that market is that no end consumer cares about what SoC is running in those products, just how good is the experience?

 

[jhu]

What's wrong with this?

OMAP was used by who? Motorola? I'm not even sure why Google picked OMAP for their GNex. It was probably because of some under the table business. Exynos would've been the right choice given its popularity in the SGS phones. But nope.

Samsung also used OMAP on some GS IIs (eg international )

I think this brings good news in that Google will be picking between Samsung and Qualcomm this round.

At least with the Nexus S, all development efforts helped the SGS1 phones. It's the reason why SGS1 phones have CM10 running well while SGS2 is still struggling to get it to even work. The SGS3 got a break because of a leak and now JB is dropping anyway.

I have CM10 running on my SG II just fine.