But not Will Strauss, principal analyst at Forward Concepts Inc. Back in November 2010, Strauss identified the trend that would lead to TI's strategic shift, and wrote about it in his monthly newsletter—the future of applications processors involved integrated baseband, and TI had a decision to make.
[Get a 10% discount on ARM TechCon 2012 conference passes by using promo code EDIT. Click here to learn about the show and register.]
Back then, Strauss reported that OMAP was the leader in 3G cell phone apps processor shipments, but that Qualcomm was already nipping at its heels with the Snapdragon communications processor, which offered both an applications processor and cellular modem on the same die. Strauss predicted that combo chips that integrated both apps processor and modem would make up nearly three-quarters of applications processor shipments by 2014.
TI had already all but exited the cellular modem business, pledging only to continue shipments to Nokia. Now TI had to make a choice—jump back into cellular modems with both feet in order to remain competitive down the road with OMAP for smartphones and tablets, or prepare to change strategies.
"Both companies [TI and Nvidia] are big enough to buy one of the few remaining modem houses, should they decide to take that route," Strauss wrote at the time.
Strauss singled out Icera Semiconductor as the most attractive modem maker left. Nvidia bought Icera for $367 million a few months later. TI, having tired of the baseband business after being undercut on price by MediaTek and others, had no intention of reversing its field.
And the rest is history. While Intel and Nvidia were willing to ante up to add baseband capability as table stakes to get into the game for future smartphone and tablet sockets, TI choose the low road. Nvidia is now offering both Tegra and the Icera baseband chips, and plans to start shipping a version with both integrated on the same die next year. Strauss believes Intel will ultimately put either Atom or a successor on the same die with the baseband, but he suspects it won't happen until at least 2014.
About Apple, they are one of the biggest price-bargainers. That is the main reason, they dont get along with any big semiconductor companies, and that is the main reason for them to start their own Apps Processor Integration. I dont think TI would have ever won that business from Apple, without changing much of its personality (and be on its knees). I am sure Apple would gradually will do their own modems, and even do their own non-ARM-processors, because they wont get along with ARM for too long, since ARM is developing its personality now. And Apple is not getting along with Qualcomm, but sticks to it since no other choice.
One biggest missing point is the "margin". It looks like the profit margin on these chips became extremely low over the decade. For a company like TI, which is more familiar with "huge" margins in Analog and other components business, smartphone/tablet was becoming more of a "NoGO" year-by-year.
Remember TI came to this business, when it was a completely different story. Nokias/Samsungs/Motorolas were literally "stuck" with each chip vendor, and transition from one chip vendor to another was not easy, which kept the semiconductor profit margin high. In the days of Android, that is not the case.
Now combine that low-profit margin with reducing "space" of market with Apples/Samsungs going with their own chips.
@ravan- I can't speak specifically to what happened with MXC. As you mention, Freescale decided to get out of baseband around the same time TI did, and I know that Rich Beyer and others were absolutely 100 percent sure it was the right move. Freescale has the i.MX apps processor, which I believe has been pretty successful in some applications like e-readers. But Beyer told me Freescale was happy aiming i.MX at these less glamorous sockets and was not interested in competing with Nvidia and Qualcomm with a cutting edge apps processor, simply because of the massive R&D costs involved.
Dylan, why Freescale (Motorola) didn't make it with their MXC processor? They integrated the modem and the app processor on the same SoC. If I am not mistake, RIM was one of the biggest consumers for MXC chips. And that was 5-6 years ago. But somehow, that technology didn't make it to have a big success and Freescale dropped wireless communications division (cellular).Somehow, Qualcomm benefits by all this key player that made an exit from the bussiness.
Glad you asked, Frank. I posted the original TI-OMAP story, but it's just as relevant here. I asked Will Strauss that very question. Here, in a nutshell, is what he said:
Apple, obviously, has no baseband. They will have to buy somebody who has one to integrate the two. But that won't happen until some company comes up with a baseband that can match Qualcomm's capabilities. Qualcomm has a virtual lock on CDMA modems (little Via Telecom can field a CDMA-only modem), a technology necessary to serve Verizon, Sprint, and some smaller operators. So Apple is stuck with Qualcomm for that big U.S.-centric CDMA market. CDMA will be with us through the rest of the decade (for over half of the U.S. market), so Apple will remain the exception to the integration trend.
Samsung has an LTE baseband and is coming up with one that is LTE/HSPA/3G/2G and will integrate it with their own application processor. However, to serve the U.S. CDMA market, they will probably tack on a CDMA modem from Via Telecom (as they did for Thunderbolt and other handsets for that market).
For the record, TI declined a request for an interview on this topic. A spokeswoman said via email that, as communicated during the investor meeting this week, the smartphone market has become a less attractive long-term opportunity for OMAP, primarily because of market consolidation and vertical integration. "We are reprofiling our investments accordingly, but have no additional details to share at this time."
The spokeswoman went on to say that TI remains committed to the OMAP platform and its customers. "We are accelerating the expansion of OMAP processors into a broader set of embedded applications such as automotive, industrial, enterprise communication, vision and robotics, to grow the OMAP footprint beyond mobile. We will share further advancements with you as we move forward in this process."
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.