Apple acquires PA Semi, furthers thrust on convergence

Wednesday, April 23, 2008 reported late Tuesday that Apple (AAPL) has agreed to buy PA Semi, a 150-person microprocessor design company. Coming on the eve of its quarterly earnings announcement, the deal rumored to be worth $278 million has caught many chipset and handset vendors by surprise.

PA Semi was started in 2003 to design and develop high-performance low-power PWRficientTM processors targeting high-performance embedded computing and control markets. The team is headed by the visionary designer Dan Dobberpuhl and is backed by Bessemer Venture Partners, Highland Capital Partners, and Venrock Associates. The 64-bit multi-core processors, the company claims, is “three to four times lower power than other similar processor platforms available today.”

I will leave further details of this deal to the frenzy of articles you will find in the web tomorrow. To me there are two very interesting questions to be addressed. Why should Apple buy PA Semi? What are the industry ramifications? I will pen my thoughts on the first question here and leave the second to a sequel.

The Apple spokesperson declined to comment on the plans and purposes of the acquisition. But there are definitely some obvious angles that I can think of.

Firstly, Apple knows very well that the future is about mobility, future is about technology convergence. Mobility, in turn, is limited by power and battery life. Whether it is the iPhone, the iPod or the Air, Apple has pushed its vendors towards efficient power management. Moving into the future, the company is perhaps looking to use this as a further differentiator from the innumerous clones springing up in the market. It wants to be the only one to leverage the design superiority of these PWRficientTM processors for most, if not all its mobile platforms.

Secondly, PA Semi’s homepage proclaims the words ‘green computing’ in bold letters. Combine this with the timing of the announcement – Earth Day – and you get the message that Apple is trying to drive. It is committed to low-power, it is committed to going green.

Thirdly, an in-house application processor development unit, especially for Apple, will enable tighter and more efficient integration. This will also allow for further form-factor and power optimization. Customized software applications that enhance the user experience can be written. Furthermore, the internal development and optimization strategy will make it difficult for the copy-cats to reverse engineer. This certainly creates more value for any mobile product that comes from the Apple stable.

In summary, Apple has made this acquisition as a very well-thought out strategic initiative. While it is unlikely that we will see PA Semi-enabled products in the next year or two, you will likely find it as the application processor not just for the iPhone, but for most of Apple’s future convergence devices – mobile phones, music players and laptop replacement devices. In a sequel, I will look at what this means to the other chipset vendors sharing the luscious mobile pie.

(For the interested reader, I have covered iPhone and its impending 3G successor in great detail in my site. You can access it here, here and here.)

Posted by Vijay Nagarajan at 12:01 AM  


Sorry Vijay, but I'll have to disagree completely here, see my news piece on the subject:

No offense intended with my title, you obviously just took Forbes' information to its logical conclusion without realizing they didn't do proper research. This processor is indeed more power efficient than the alternatives in the x86 and PPC space (or at least as efficient), but...

It's not designed at all for the handheld market and couldn't touch anything designed by ARM in terms of performance/watt. And raw wattage would be substantially higher than Silverthorne even on 45nm with a single core.

BTW, I agree with your opinion on SiRF; I don't think they can really prospere on their own going forward. Marvell or Intel are indeed viable acquirers, of course whether that will happen or not (and how much they'd be willing to pay) is another question entirely...

Arun Demeure said...
April 23, 2008 at 8:15 AM  


Thank you much for your comments. Read your post. Interesting insights.

I think I might not have come across as effectively as I wanted to but if you sieve through this piece again, I have made the point that this is not for just the iPhone (Om Malik also speculates incorrectly that it is for the iPhone) but much beyond it - for any future mobile platform. And by mobile platform, I don't imply mobile phones but convergence devices pretty much along the lines of your "ultraportables and mobile devices between the iPhone and the Macbook." So, I am more in sync with your thoughts that off it, I think:).

Of course, the one difference is that I also think that down the lane, we may see this in the iPhone as well. This is not because the chip architecture is suited for the iPhone as is today but more because the iPhone will also morph towards further convergence in the future. (I have to confess though that taking it to iPods may be a stretch but as you may be aware, I am a big fan of the device convergence. I think there is a good possibility that in 10 years, the market for iPhone and iPod may merge.)

Also note that I have given importance to the design team than the existing design itself. I think that is also the key. The ability to have in-house experts further optimize the power at the hardware and software levels. In this context, I do think that designing a mobile application processor in-house now to compete with TI or Marvell is not beyond Steve Jobs' out-of-the-box thinking, especially if it can make sure that the Apple signature is unique to these devices.

And certainly, I think the PA Semi product will be a direct competition to the Atom putting Intel's business plans in the lurch. I agree with you there.

As always, keep your comments coming. More debate on this topic is very welcome. Your articles give me a good perspective and educate me further on the technical aspects.


Vijay Nagarajan said...
April 23, 2008 at 8:45 AM  

Yup, certainly their technology is very attractive for UMPC/MID, Ultraportables or Tablet PC-like devices. So on that front I certainly agree with you.

However, the way I look at it is that if Apple is serious about using P.A. Semi's technology for that market, and maybe the server market too (see my UPDATE at the bottom of the news post), then they need to keep working on high-performance processors that aren't suitable for handhelds.

The P.A. Semi design team isn't small, but it's also far from huge and I very much doubt they could support two independent architectures at the same time. Right now, ARM's Cortex-A9 on TSMC's 40nm process is likely to deliver better power efficiency than anything P.A. Semi can come up with in the next few years.

Now, if we're talking about, say, 2012... Who knows! That's at least *FIVE* iPhone generations away. You'll have to excuse me for not trying to speculate about such long-term things; Apple's decisions are already quite unpredictable in the short-term as this acquisition points out, so in the long-term...

And honestly I don't see why Apple would feel they could deliver a better solution than what is offered by Samsung, NVIDIA, TI, or [insert company name here]. They care a lot about pwoer efficiency, but the best way to achieve power efficiency from a hardware perspective may just be to be able to pick and choose between available designs and take the best one. That's exactly what they did with the original iPhone at least.

I understand where you're coming from when it comes to the software aspects of power consumption, and I agree that matters - however it's worth remembering that every single application processor company would *kill* to have Apple's business. So they'd be more than willing to allocate every possible resource to make sure the software side of things is finely tuned - which is why I don't worry too much about that in Apple's specific case.


Arun Demeure said...
April 23, 2008 at 9:16 AM  

Post a Comment