Texas Instruments will no longer make ARM for phones/tablets (OMAP)

[Roland00Address]

Texas Instruments withdraws from smartphones, goodbye OMAP
26 September, 2012 | Comments (17) | Post your comment

Tags: Misc

Texas Instruments is dropping from the system-on-chip for smartphones and tablets manufacturing and will give up on its OMAP lineup.

The company’s OMAP boards are less and less popular among mobile manufacturers – most of them bet on Qualcomm, while Samsung and Apple are developing their own solutions (Exynos, A6). The major disadvantage of the OMAP chipset is the lack of on-board 3G/4G modem.

That forces manufacturers who rely on OMAP chipsets to use additional radio chips, which increases battery consumption and production costs. Now you understand why smartphone manufacturers prefer Qualcomm’s complete solutions, rather than this expensive process.

TI says its focus will shift on “to a broader market including industrial clients like carmakers”, though it did not announce specifics and the investors were left wondering.

Anyway, TI will continue to support its current clients, but will significantly reduce efforts on developing new OMAP chipsets.

The news might come shocking for some, as the TI OMAP 5 was expected to be the first chipset with dual Cortex-A15 CPU, and now it's fate is uncertain. Nonetheless, TI OMAP's presence was barely felt on the market, so the company's exit won't create too much of a disturbance

http://www.gsmarena.com/texas_instruments_backtracks_from_smartphones_goodbye_omap-news-4861.php

Boo

So besides Nvidia and Samsung who is still using the reference arm cpu design and not doing a custom design like Qualcomm or Apple?

 

[Idontcare]

This was foreseen a long time ago, it was why TI got out of the advanced CMOS process node development in the first place beginning with 45nm. They couldn't compete with Qualcomm, contracts are signed 4yrs in advance and we knew all the way back then that we lost pretty much all the business that was going to be had 4 yrs later. We went from ~80% marketshare to 10% marketshare in that time.

If you look at what TI has done in the past 10yrs or so you will notice that they don't really innovate or create anything anymore. Once Tom Engibous took the helm, and Rich Templeton followed in his footsteps, the business of TI is basically about acquisition. They acquire companies that have already innovated and then TI scales them up volume-wise because it has the fabs to do that.

It isn't a bad way to do business, TI's financials are solid, but the innovation engines of TI have long gone cold and the reality of their once crown-gems like OMAP becoming outdated and outmoded is the outcome of that.

 

[exar333]

This was foreseen a long time ago, it was why TI got out of the advanced CMOS process node development in the first place beginning with 45nm. They couldn't compete with Qualcomm, contracts are signed 4yrs in advance and we knew all the way back then that we lost pretty much all the business that was going to be had 4 yrs later. We went from ~80% marketshare to 10% marketshare in that time.

If you look at what TI has done in the past 10yrs or so you will notice that they don't really innovate or create anything anymore. Once Tom Engibous took the helm, and Rich Templeton followed in his footsteps, the business of TI is basically about acquisition. They acquire companies that have already innovated and then TI scales them up volume-wise because it has the fabs to do that.

It isn't a bad way to do business, TI's financials are solid, but the innovation engines of TI have long gone cold and the reality of their once crown-gems like OMAP becoming outdated and outmoded is the outcome of that.

Kind of sad.

 

[meloz]

No big loss...

Qualcomm are doing a great job. nvidia + Samsung can provide the competition to keep prices in check. Samsung also have foundries which will need a lot of new work to fill the gap when Apple decide to move to TSMC.

There are plenty of players in the ARM-sphere to keep the momentum going. I won't be surprised if we eventually end up with just two major ARM players: Qualcomm and Samsung. nvidia (or more likely, their IP) might one day be acquired by some other players.

 

[Idontcare]

Kind of sad.

It was personally quite devastating at the time. 4yrs later though I see how unavoidable it was and why management made the tough decisions it made back then.

TI management did the right thing, rather than live in denial and hope that somehow they'd win back some of the business they instead opted to throw in the towel and redefine themselves as leaders in Analog CMOS.

The following graph is obviously quite dated, but it shows just how quickly TI started losing marketshare to Samsung and Qualcomm. The line continued to fall to now where TI only has something like 4-5% marketshare in 2012.

Qualcomm now has >50% share, ST-Ericsson has <1%.

 

[Phynaz]

IDC, now that companies are making progress into taking what have traditionally been analog chips and recreating those circuits in the digital realm, where does that leave TI long term?

I'm an old school guy, and I remember fondly the days when TI was an engineering powerhouse. Now, not so much it seems.

 

[Idontcare]

IDC, now that companies are making progress into taking what have traditionally been analog chips and recreating those circuits in the digital realm, where does that leave TI long term?

Just a guess but it probably leaves them exactly where the LP (vinyl record) manufacturers were left when music for the masses went digital (CD's).

There will always be a niche for analog because the performance can't be absolutely replicated. LP's didn't die out, but their TAM shrunk to a crazy small value. (same with horse-buggy manufacturing )

But I'd venture to estimate that 80% of the analog market could be served with digital devices that provided 80% of the quality as the analog parts they would replace...and they'd replace them because being digital they will shrink in size and in cost much faster than the analog parts. (the whole Moore's law thinking that Intel went with as the basis for creating their digital radio IC)

So a company like TI which is highly dependent on analog sales will either have to contract as the TAM for analog contracts or they will have to dive back into the nitty gritty of design and get some elbow grease flowing.

The problem with this of course is that elbow grease is costly and not all that appealing to shareholders.

Look at the following chart that tabulates R&D expenses by company. Don't fixate on the sale/R&D ratio (that is merely a measure of R&D ROI), rather fixate on the absolute dollar number for R&D as this is telling you how much actual R&D activity is going on:

source

 

[Eug]

I wonder how much wireless IP that Texas Instruments would have to buy to make mobile ARM products actually viable for them.

 

[podspi]

I was really looking forward to OMAP5... Oh well... Didn't realize they were unable to integrate the model onto the SoC. Without that they were borderline doomed in this market.

 

[WhoBeDaPlaya]

TI Dallas is actually axing their Wireless and ASIC business units. Man, they're getting out of digital big time

 

[soccerballtux]

This was foreseen a long time ago, it was why TI got out of the advanced CMOS process node development in the first place beginning with 45nm. They couldn't compete with Qualcomm, contracts are signed 4yrs in advance and we knew all the way back then that we lost pretty much all the business that was going to be had 4 yrs later. We went from ~80% marketshare to 10% marketshare in that time.

If you look at what TI has done in the past 10yrs or so you will notice that they don't really innovate or create anything anymore. Once Tom Engibous took the helm, and Rich Templeton followed in his footsteps, the business of TI is basically about acquisition. They acquire companies that have already innovated and then TI scales them up volume-wise because it has the fabs to do that.

It isn't a bad way to do business, TI's financials are solid, but the innovation engines of TI have long gone cold and the reality of their once crown-gems like OMAP becoming outdated and outmoded is the outcome of that.

where do the master engineers that worked in such companies gone stale go? Who do they go to work for?

 

[Phynaz]

Look at the following chart that tabulates R&D expenses by company. Don't fixate on the sale/R&D ratio (that is merely a measure of R&D ROI), rather fixate on the absolute dollar number for R&D as this is telling you how much actual R&D activity is going on:

source

Holy smokes, Intel invests more into R&D than the next three companies combined. Kinda scary when you think about it.

 

[WhoBeDaPlaya]

where do the master engineers that worked in such companies gone stale go? Who do they go to work for?

We have a LOT of ex-TI guys at Maxim :whiste:

 

[OBLAMA2009]

now phone makers wont be able to manufacture fones that get half an hour of 4g browsing time, what a loss...

 

[Idontcare]

where do the master engineers that worked in such companies gone stale go? Who do they go to work for?

We have a LOT of ex-TI guys at Maxim :whiste:

Yep, Maxim, Marvell, Qualcomm, Samsung, etc. The winners generally have to expand their headcounts to build on their winnings.

 

[WhoBeDaPlaya]

It's pretty much Hobson's choice in Dallas because a lot of the semicon action is now in Austin (not like the good 'ol days when Nortel was big here).

 

[ShintaiDK]

Consolidation is the keyword for the future.

 

[Blitzvogel]

Crazy thought, could TI buy AMD? My god.....

 

[wlee15]

Crazy thought, could TI buy AMD? My god.....

If they wanted to they probably wouldn't be dropping their OMAP line.

 

[Blitzvogel]

If they wanted to they probably wouldn't be dropping their OMAP line.

I thought about this too, but OMAP is one thing, dropping ARM altogether is another and of course AMD is now in semi-cahoots with ARM (sort-of). The combo of TI and AMD would bring a wealth of knowledge and experience to both sides, not to mention perhaps stabilize the AMD side of things while providing valuable research funds.

Not likely to happen, but it would be interesting never the less. AMD would be fabless no more too, though that could make things an issue when it comes to them and their relationships with the console companies

 

[Phynaz]

Crazy thought, could TI buy AMD? My god.....

What good is AMD without the x86 license? It's not transferable. And who wants to take on Intel?

 

[NostaSeronx]

AMD is the bro of ATIC.

Or, in the broader picture:
AMD is owned by
http://www.mubadala.ae/

Why do you think the 16-core Piledriver's product code name is "Abu Dhabi." Now, if you are going to ask me about "Seoul" that means a contract win for Samsung. "Delhi" is the successful contract to make a new Fab at or around Delhi, India.
ATIC/Mubadala:
Abu Dhabi contract(ATIC and Abu Dhabi fab, GloFo)
Samsung contract(28-nm APU)
Delhi contract(Delhi fab, GloFo)

If you want the big picture I gave it to you.

 

[blastingcap]

What good is AMD without the x86 license? It's not transferable. And who wants to take on Intel?

I don't know the details of the x86 licensing agreement, but presumably if AMD could keep its x86 license despite eating ATI, then AMD can also merge with other entities as well and keep its x86 license.

How possible is it to have a technical merger but with AMD being the "acquiring" entity even if it's the smaller entity?

 

[thilanliyan]

Why do you think the 16-core Piledriver's product code name is "Abu Dhabi."
If you want the big picture I gave it to you.

Not sure if my sarcasmeter is broken but amd uses codenames from Formula 1 venues... Barcelona, interlagos, abu dhabi, etc...at least for the server parts.

 

[nismotigerwvu]

It's compelling in a way to think of the X86 license ramifications the acquisition of TI might have on AMD considering that TI historically produced X86 chips themselves under a license from Cyrix. I mean, all of these chips are public domain now so there's no real legal ramifications, it still is an interesting twist. I really wish the barriers of entry into the market weren't so high, all of the competition on the desktop is essentially gone (the same holds true for the graphics sector as well).