iPad A4 Chipset Cost Apple $1 Billion?

Apple A4 chip

We typically don't run those stories about iPhone or iPad component costs anymore because they're just silly -- a couple of hundred dollars in metal parts per unit never takes into account R&D and marketing costs, and things like paying $1 billion for the new Apple A4 chipset in the iPad. Or so hints the New York Times:

At the same time, Apple, Nvidia and Qualcomm are designing their own takes on ARM-based mobile chips that will be made by the contract foundries. Even without the direct investment of a factory, it can cost these companies about $1 billion to create a smartphone chip from scratch.

And this would be for an ARM + PowerVR chipset assembly, how much will it cost when Apple starts spinning their own PA Semi designed chipsets whole? And what's the competitive advantage that they're willing to spend so much?

Have something to say about this story? Leave a comment! Need help with something else? Ask in our forums!

Rene Ritchie

EiC of iMore, EP of Mobile Nations, Apple analyst, co-host of Debug, Iterate, Vector, Review, and MacBreak Weekly podcasts. Cook, grappler, photon wrangler. Follow him on Twitter and Google+.

More Posts



← Previously

Should Apple TV Switch to the iPhone OS?

Next up →

Monday Fun Video: Apple Center Seen from the Sky

Reader comments

iPad A4 Chipset Cost Apple $1 Billion?


I think the $1billion figure is for a completely custom chip. Slapping together a standard ARM CPU, GPU, and a bridge/controller or two on one die is not a completely custom chip. So I highly doubt the A4 cost Apple that much.
Also think about ROI. The A4 is likely to be obsolete in a year simply because of the progress of technology. I don't see iPad sales being anywhere near iPhone sales so they would never recoup costs at that rate.

The competitive advantage is the complete marriage between the hardware and software elements of their products. I think I read something similar to what I just wrote in that interview with Steve Jobs in Rolling Stone or somewhere. Something as simple as the presence or absence of a hardware decoder can alter the battery life by maybe a factor of five to ten -- right in the vicinity of where Jobs was putting the differences in battery life with Flash versus h.264. An mp3 player typically has hardware supported mp3 decoding; playing ogg or FLAC files on the same player cuts the battery life substantially.
Having custom hardware also means that the software and hardware cannot be easily separated from one another. That forms protection against issues like Psystar -- a company that was financially supported by people motivated against Apple to test the law that enables Apple to control the sales of its OS. If the OS and the hardware are functionally and practically inseparable, then issues relating to the possibility of their being separated cannot come up at all.
Finally, there is an issue of company secrets -- things that are not patented but are maintained in secrecy, because they are considered to be crucial to the existence of the company and its products. Sometimes they are not patented because they aren't actually patentable; sometimes they aren't patented to keep certain details out of public documents.
All of these things -- performance, protection, and secrecy -- are aspects of the same thing. That's why it makes sense for Apple to spend $1B on a project like this. They also have an incredible amount of money, more than they've ever had -- what else are they going to do with it?

maybe you should include this link in the next idevice component cost article so people will stop complaining about companies not selling them hardware for $1 over cost..

Why is it Rene is so hung up on PA Semi?
What have they ever done or designed that can compete with an ARM, or Atom?
My guess is Apple bought them for patents, not for anything they could actually produce, design, or integrate.

I think Rene is just wondering what PA Semi is really doing. Again, I don't think PA Semi can or even should try to provide custom chipset components for Apple, as IMO they can do just fine using available components to design chipsets to certain specifications. I think there's a lot that can be done in the design of the chipset that discourages or prevents hacking, for instance. (One example I've seen recently was from a forum discussion somewhere, of what to do with an old Tivo; apparently due to the hardware design, it's basically impossible to hack a Tivo into a Linux box, for reasons that I couldn't understand.)
Regarding the $1B number, there are all kinds of things that could have been put into that calculation, including the cost of acquiring PA Semi, all the expenses involved in assuming control over all of its operations, infrastructure costs, bonuses and stock options that changed hands (if any), the cost of the big "You're a Part of Apple Now!" party that they certainly threw, the salaries of all the employees who spent time working on the acquisition of PA Semi as well as the chipset design, the cost of prototyping and testing, etc. The particularities of all that matters more to stockholders and for the requirement of publicly reporting expenses and earnings, and for accounting and tax issues.
The main thing is that this number was released in response to the recent stories about the cost of producing the iPad versus its retail price. Apple has been very active in monitoring its reputation and image, possibly because they were taken a little off guard by the intensity of some of the criticism after the iPad announcement.

The $1B figure is a ridiculous exaggeration. Sure maybe Intel spends that or even more developing a new CPU, but this isn't anything even remotely like that. You just take an existing licensed ARM core and PowerVR silicon from ARM and wrap it in some logic to connect it to the PINs, add RAM, add some buses, maybe some custom logic. I worked on one that involved maybe 8 hardware engineers ($100k each per year) for maybe 9 months, plus NRE to the fab company, plus licensing costs. Probably in the small single digit millions.