Rumor: Apple plans to bring iPhone baseband chip design in-house
A rumor from the Far East on Tuesday claims Apple is mulling the creation of a research and development team that will be responsible for designing future iPhone baseband processors to be used as soon as 2015.
Citing industry sources, DigiTimes reports Apple is planning to develop an in-house baseband processor to be used in 2015 iPhone models. The component will supposedly to be produced by Samsung and Globalfoundries.
One of the most important pieces of silicon in a cellphone, the baseband chip, or baseband processor, manages all on-board radio functions. Currently, Qualcomm supplies Apple with BPs used in both the iPhone and iPad product lineups.
The publication mentions a separate rumor that claims Apple is looking to incorporate a baseband processor into its A-series application processor, though sources say the company will stick with two discrete chips for now.
Tuesday's rumor lines up with Apple's ongoing efforts to exert tighter control over the design and manufacturing of chips used in its product lineup. Most recently, the company was reportedly in talks to take a $479 million stake in Japanese chipmaker Renesas SP Drivers, which is said to be the sole supplier of LCD chips for the iPhone.
Citing industry sources, DigiTimes reports Apple is planning to develop an in-house baseband processor to be used in 2015 iPhone models. The component will supposedly to be produced by Samsung and Globalfoundries.
One of the most important pieces of silicon in a cellphone, the baseband chip, or baseband processor, manages all on-board radio functions. Currently, Qualcomm supplies Apple with BPs used in both the iPhone and iPad product lineups.
The publication mentions a separate rumor that claims Apple is looking to incorporate a baseband processor into its A-series application processor, though sources say the company will stick with two discrete chips for now.
Tuesday's rumor lines up with Apple's ongoing efforts to exert tighter control over the design and manufacturing of chips used in its product lineup. Most recently, the company was reportedly in talks to take a $479 million stake in Japanese chipmaker Renesas SP Drivers, which is said to be the sole supplier of LCD chips for the iPhone.
Comments
Digitimes! my fav.
The power-saving potential here is tremendous. This is where Apple could set itself apart from the rest yet again.
hmm... how about Qualcomm?
Could you elaborate.
hmm... how about Qualcomm?
Could you elaborate.
Qualcomm has all the pieces to do the same.
So does Intel.
They also have all the parts to make their own ARM SoCs but since they build for others and try to cast the widest possible net they aren't designing a bespoke chip that will work specifically with Apple's HW and OS. That's the difference Apple can bring by building a baseband chip designed to work with their HW and only their HW, something that perhaps Samsung could also do because they do sell their higher-end devices in the millions but I don't see how anyone else could do it. We're talking size and power and reductions already, but if Apple can make it a PoP as part of their A-series chips they could really push this to the next level.
If you still don't think so then tell me Qualcomm's Snapdragon isn't wiping the floor with Apple's A-series chips and why Apple has spent hundreds of millions if not billions investing in their own chips.
I hate to say this but I think its best to keep Qualcomm as a vendor, then to piss them off and make them a competitor. They are a royality revenue generating monster when it comes to do with anything related to wireless. Apple would be best to avoid that fight.
case in point, look at why Samsung uses Snapdragon and Qualcomm baseband only in US Galaxy S models and their own SoC elsewhere.
Apple did buy those Nortel patents for reason but because of Qualcomm's patent holdings it's hard for me to see how any baseband HW will not include components or IP from Qualcomm. I also don't think it's unreasonable for Qualcomm to allow Apple to pay for access to certain parts of the baseband that it can integrate into their own design. Last year Qualcomm stopped requiring their MDM9x15 to be paired with a Qualcomm SoC for voice capabilities. According to AnandTech, this "opens it up for use in platforms like the iPhone where an OEM has a specific non-Qualcomm SoC it wants to use. This same IP block is again shared with MSM8960 and a few other SoCs." Who know what else they could allow or why they opened this up at all. Was that change last year because of Apple?
I hope this is true, I have been hoping for this for a long long time. And even 2015 is late.
Quote:
The power-saving potential here is tremendous. This is where Apple could set itself apart from the rest yet again.
There is possible power saving. But definitely not tremendous. And i would bet Apple's design won be as good as Qualcomm as well.
Qualcomm's baseband, transceiver, receiver, RF Filter are state of the art. The only possible technical improvement Apple could bring is by potentially ditching some flavor of technology that dont want to support, or potentially design for LTE only. But even those benefits will be small.
The major potential here is cost reduction. Apart from the Screen and Touch Display, the next most expensive piece of technology is the Wireless Related parts. For Apple to add anymore value, such as Sapphire Screen without increasing the selling price of iPhone there has to be a reduction somewhere.
The estimate cost of BaseBand parts currently stands somewhere from $28 - $35. With the majority of estimate going at the higher end range. Apple could potentially make one themselves in TSMC for less then $15. R&D cost excluded.
There are already a few IP company offering 4G solution, although as far as i am concern none of them are anywhere as good as Qualcomm or even Intel's offering.
So are their Snapdragon SoCs but Apple built their own and it's clearly proving Apple's bespoke designs are better suited for Apple's own needs so why assume it's not reasonable for Apple to use their talents to streamline other generalized chips in their system?
These arguments sound the same as the ones that said Apple is foolish for making a smartphone and Apple is foolish for buying a fabless semiconductor company.
What if Apple has plans for some kind of communication device that is unique and they want to keep it under wraps until the last moment? They could keep it disguised as a project to knock off the Qualcomm chip....
I'm not an engineer so I'm just guessing, but these radio processors must be one of the main power hogs after the LCD and CPU. As someone alluded to above, it's likely one of the more expensive parts as well. In other words, there's a lot of motivation for Apple to take this in-house. Reductions in power consumption and cost would offset new innovations such as the sapphire glass.
I can't see Apple doing this without first buying a baseband design house. It's a very specialist industry.
Oh, wait, the rumour is from DigiTimes. Nevermind!
I'm not an engineer so I'm just guessing, but these radio processors must be one of the main power hogs after the LCD and CPU. As someone alluded to above, it's likely one of the more expensive parts as well. In other words, there's a lot of motivation for Apple to take this in-house. Reductions in power consumption and cost would offset new innovations such as the sapphire glass.
The modem is one of the more expensive components but the main cost is licensing. That's not a cost that Apple can realistically drive down.
There's certainly other advantages for bringing it in house though. Simply being able to define and control the interface between the AP and modem would probably make Apple's engineers very happy.
What Apple realizes is that if they buy a chip from Infineon or Qualcomm or Broadcom, their competitors will have the exact same chip and can make close to the same phone. That's the Android market. Everybody has nearly the same phone, just with different shovelware. Or the PC market.
Apple is not the first company to realize this. Cisco does the exact same thing, they roll their own chips. Their weakest competitors, a bunch of minor players like 3Com, all lost when the Chinese Huawei bought off the shelf router chips and made a cheap product. Then companies like Google and Facebook realized they could roll their own router from the reference design and cut out the middlemen all together.
As for Apple not having the expertise I'm not sure I buy that either. Over the past few years they have acquired a bunch of IP and more than a few engineers from failed companies. What they don't have engineering wise they can easily hire.
As for reasoning here, one only has to look at the micro photography's of some of Qualcomms processors to see why this is a reasonable story in and of itself. The space savings alone are huge and would allow for far more design flexibility for Apple. Think doubling flash storage or making access wider (faster). This is nothing new for the likes of Qualcomm and a process shrink frees up a tremendous amount of space for new transistor functionality. The only real question here is what is taking Apple so long.
As for Apple not having the expertise I'm not sure I buy that either. Over the past few years they have acquired a bunch of IP and more than a few engineers from failed companies. What they don't have engineering wise they can easily hire.
It's really not that easy. (Good) baseband engineers are a rare breed. There's a few companies that Apple could hire from locally but raising a whole department of engineers is going to be a slow process... unless Apple buys a company with existing expertise.
So are their Snapdragon SoCs but Apple built their own and it's clearly proving Apple's bespoke designs are better suited for Apple's own needs so why assume it's not reasonable for Apple to use their talents to streamline other generalized chips in their system?
These arguments sound the same as the ones that said Apple is foolish for making a smartphone and Apple is foolish for buying a fabless semiconductor company.
You are comparing Apple to Orange. Did any one ever said Snapdragon were ever state of the art? Surely it isn't to me. And i could foresee huge amount of headroom and potential for improvement in those.
And there is nothing generalized in those baseband chips. Unless you count GSM and CDMA as uesless. Which i would doubt Apple dropping it.
The technical improvement coming from the hardware baseband in house are minimal at best. The benefits ( both cost and power consumption ) are likely coming from the Integration inside the SoC rather then housing it as an seperate unit.
Well LG and Nvidia ( Icera ) have both managed to make a Software defined solution support full stack LTE without paying. And even a licensed hardware solution doesn't cost that much, at least MediaTek will have a SoC with LTE support selling for less then $25.