Apple's iPad Chip Development Cost Estimated in $1 Billion Range

[cmaier]

Makes me wonder what the did spend it on...

Is it because it doesn't have an Apple logo on it?

The article is very vague. But seriously, they need to pay:

1) for a license. This is typically per-chip.
2) for the fab lots. This is typically per wafer start.
3) for engineers. A team of 50 is more than enough to design, verify, test, etc. these little chips. Let's call it 100. Average salary is, say $125K.
4) for EDA licenses. Even a huge company like AMD spent less than $20M on that.
5) for workstations, electricity, cubicles, etc.

Do the math. Like I said, it's never cost anyone $1B unless they also bought a fab (MicroUnity comes to mind).

I PM'd Fred Weber on facebook to ask him what he thinks, since the NYT article doesn't imply that he had anything to do with the $1B estimate.

 

[OnePotato]

Fortunately, they're NOT starting from scratch, so the cost is probably at least an order of magnitude less.

Now, if the clueless posting here would just buy a clue, they'd realize that Apple just tweaked an existing design, they did not design it from scratch, so it cost a whole lot less.

 

[cmaier]

I am sure it was easier and cheaper to build a Transmeta chip. Where is that chip now? Billion may be overexaggerated, but I don't think they are off by as much as you may think they are.

It is WAY harder to design an x86 chip than a chip which only has to handle RISC instructions - the verification costs are astounding. (Even when you cheat and use transmeta's software-based translation methods).

I worked in that industry for over a decade, and designed Opteron and Athlon 64, UltraSparc V, etc. I worked on RISC processors more complicated than ARM, and on much more cutting edge technology, and it cost way less than $50M. I'm telling ya, it is way closer to $100M than to $1B. Not even CLOSE to $1B.

 

[lincolntran]

R&D is still one of the better tax deductions.

Apple changes minds. That's what it does.

Rocketman

With so many idiots and dogma out there. I just know it costs a lotSSSSSS to accomplish this alone.

 

[whooleytoo]

That price sounds freakishly high - but it might include the purchase of PA Semiconductor.

Incidentally, was reading some back stories about this, and found this snippet. News to me..

When Apple bought PA Semi, it had no intention of morphing into a military supplier. The Defense Department, however, received push-back from suppliers who had already committed to long-term designs based on the microprocessor. These equipment makers urged the Pentagon to pressure Apple into continuing to supply PA Semi’s chips for several years, according to people familiar with the negotiations, who declined to comment on the record because of Pentagon restrictions. Apple eventually caved.

Source: NYT

 

[vt220]

how about "real" computers?

I like apple's products, they are working and always worth their money.
For this reason i will buy an ipad and net an upcomings whatever tablet pc or convertible.
I also like the dual hardware way -> normal x86 systems and arm based iphone os systems.
But apple also has to focus an their old core business; it's really time for new mac pros und macbooks. it couldn't be to hard to get this done :-(
I can't be that you pay 2500$ for hardware (mac pro) that you can build today for 700$ (except the good lookings). So please update the macbook pro and macbook series.

 

[CFreymarc]

This article is more for stock boost than real numbers. First, companies like PA Semi goes through round and rounds of VC to get where they are. Second, not all the money to get to the A4 running on Apple products came out of one pocket. My take is that at most $100 million was brought about have the A4 happen. Of that, Apple probably dropped half of that and the rest was written off as a loss by dozens of investors over several years.

Either way, it is nice to see someone break out of the Win-tel mold. Android is totally set up to be the smart-phone Win-tel like device. That is, a cheap manufacturer license, customized targets that will lead to a variety of binary compatibility fragmentation and such.

However, this time around, unlike Microsoft DOS / Windows marketing, the purchase decision is more individual other than corporate. Thus, the "it is cheaper and I can't tell the different due to my lack of knowledge" logic is not going to float in this space. Android is more clunky in UI and programming environment than the iPhone and people are picking that up quick.

IMO, I see Google not making any money of Android licenses for a long, long time. If anything, they are doing it as a strategic write off for their search engine market.

 

[gnasher729]

In an article from The New York Times, the cost for a company like Apple to develop an ARM-based mobile chip such as the A4 used in its forthcoming iPad tablet is estimated at approximately $1 billion, even without the need to invest in manufacturing facilities for the chips due to agreements with existing chip foundries for production.

I'd say they are totally talking out of their arse. The cost is basically the purchase price for PA Semi and the cost of running the company. The first was about reportedly $278 millions, the latter is probably less than 40 million per year.

<random clueless idiot>
...but, but.... iSuppli said it's only 17 dollars per chip?! They would have to sell 60 million iPads and still make zero profit?!
</random clueless idiot>

That should make it clear that iSuppli numbers are in no way a reliable way to find out how much money Apple is making with their hardware products...

Can we turn this into a proper rumor? iSuppli estimates more than 60 million iPads sold!

3) for engineers. A team of 50 is more than enough to design, verify, test, etc. these little chips. Let's call it 100. Average salary is, say $125K.

My guesstimate was based on 200 engineers at a total cost (salary, taxes, offices, HR, parking space, equipment) of $200,000 per year. Which would be an awful, awful lot of engineers.

And you can correct me, but I think building a chip that is leading edge in speed is harder than building a chip that is leading edge in power consumption.

 

[John.B]

My take is that the 1b is much, much closer to reality than anything iSuppli has ever put out. It is not very hard to make an educated guess about R&D costs when a lot of companies design ARM chips.

Maybe... If they started the A4 from scratch. Which they didn't...

 

[German]

Maybe... If they started the A4 from scratch. Which they didn't...

This is not sure

A former Apple engineer writes, “A4 was stated to contain an ARM Cortex-A8, not an A9. It was an internal source that told me. I trust them. No confirmation on which SGX, but it is one.”

http://venturebeat.com/2010/02/05/apple-a4-ipad/

A very trusted source tells me: PA Semi didn’t do the A4. It was the existing VLSI team. Apple has made custom chips for years like the Northbridges for G4 and G5.

http://venturebeat.com/2010/02/06/apple-a4-cooler-battery/

 

[anthonylambert]

How can this be true???

Apple bought PA-SEMI for 3xx million dollars. They were in production with their version of the Power PC that they had developed. They had to do everything that Apple did to get their chip to market. If the value of their company was only 3xx million and I'm guessing they had a lot less VC investment to get to this point.

Apple bought the reference designs for the chip so had a lot less to do.

How can this 1 Billion value be true????

 

[je_wallace]

if they really did

If Apply really did spend $1 Billion and/or the market believed that they did spend that much, then the Apple stock would likely drop or dip today....cause no investor would want to be holding a company with such run away costs.

But knowing that Jobs et al. squeeze costs tighter than tight, there's no way this cost them 10 digits. Even 9 large seems unbelievable.

 

[Strobe]

I can't be that you pay 2500$ for hardware (mac pro) that you can build today for 700$ (except the good lookings). So please update the macbook pro and macbook series.

Put me a cart together on like Newegg or something that is $700 that matches a $2500 dollar mac pro. Don't forget to use a Xeon with a server board and have 4 PCI-E slots and a case with 4 drive sleds and etc... Excluding looks, but maintain all specs. Now I'm not saying the xeons and server board and all this stuff is worth it, but to say 700 bux buys the same thing as a mac pro is wrong.

 

[cmaier]

I'd say they are totally talking out of their arse. The cost is basically the purchase price for PA Semi and the cost of running the company. The first was about reportedly $278 millions, the latter is probably less than 40 million per year.



Can we turn this into a proper rumor? iSuppli estimates more than 60 million iPads sold!



My guesstimate was based on 200 engineers at a total cost (salary, taxes, offices, HR, parking space, equipment) of $200,000 per year. Which would be an awful, awful lot of engineers.

And you can correct me, but I think building a chip that is leading edge in speed is harder than building a chip that is leading edge in power consumption.

Speed or power makes little difference - what matters is the number of transistors, because someone has to put them all in the appropriate place and figure out how to wire them together. Opteron was done with around 100 engineers. Intel usually uses about 400 per design. Still ain't getting to $1B.

Note the article doesn't even say Apple spent a billion. It says companies like Apple could spend a billion. It also doesn't provide a source for its assertion.

It's 100% nonsense.

 

[jouster]

Well for that much money Apple either think the iPad is going to be a phenomenal success or they plan to use the technologies they have developed in other Apple products (iPhone)

My guess would be both.

 

[firewood]

This wild guess is way more than one order of magnitude off, even given 100's of Silicon Valley annual salaries and several sky-high sub-micron mask set costs.

The foundry costs are also no where near that ballpark. I helped a poorly funded start-up tape out an SOC chip a couple years ago.

What's Apple entire annual R&D percentage?

 

[twoodcc]

well i guess this means we'll see this chip in the next iphone and ipod touch then

 

[tdream]

well i guess this means we'll see this chip in the next iphone and ipod touch then

you are smarter than the average bear!

$1 billion shouldn't hurt them too much since last quarter's revenue was $15 billion!!

 

[cmaier]

you are smarter than the average bear!

$1 billion shouldn't hurt them too much since last quarter's revenue was $15 billion!!

THrowing money away always hurts, and spending $1b for something that everyone else pays $50M for is pretty dumb business. Which is why it's untrue.

 

[BlueQ]

too bad they are only going to sell about 20 of these things.

 

[PBG4 Dude]

I'm guessing that the next incarnation of the iPhone and iPod Touch are going to use a smaller/slower/cooler version of the A4 processor. And I'm even willing to bet they're going to call it the A2.

I was thinking Apple would call it the #10 envelope.

 

[Sgt Pepper]

In an article from The New York Times, the cost for a company like Apple to develop an ARM-based mobile chip such as the A4 used in its forthcoming iPad tablet is estimated at approximately $1 billion, even without the need to invest in manufacturing facilities for the chips due to agreements with existing chip foundries for production.Chip industry expert Fred Weber notes in the report that Apple's iPhone was the first "really aspirational device" not based on Intel chips, demonstrating the power and versatility of ARM-based chip designs. The iPhone's success has consequently driven a surge of interest in the platform from other mobile vendors and even more traditional notebook vendors like HP and Lenovo looking to incorporate the power-saving yet capable chips into their products.Apple acquired chip design firm P.A. Semi in early 2008, reportedly enabling Apple to pursue in-house ARM-based designs for system-on-chip platforms for the iPad and iPhone.

Article Link: Apple's iPad Chip Development Cost Estimated in $1 Billion Range

ARM Chip - if my memory serves me right was this originally developed, or something to do with Acorn Computers, Cambridge, UK, makers of the BBC Micro in the 1980s. Anyone got a more thorough knowledge of history? (I might be wrong?)

 

[cmaier]

ARM Chip - if my memory serves me right was this originally developed, or something to do with Acorn Computers, Cambridge, UK, makers of the BBC Micro in the 1980s. Anyone got a more thorough knowledge of history? (I might be wrong?)

You are correct.

 

[MarkHarrisonUK]

Fred Weber notes in the report that Apple's iPhone was the first "really aspirational device" not based on Intel chips

So, his opinion is that the original Mac's weren't "really aspirational" then?

Huh?

I remember when they used 68000-series chips, before the move to PowerPC. It's only VERY recently in Apple's history that they've started using Intel.

Though, to be fair to Fred Weber, this may be a massive edit by the reporter of what he actually said - it's hard to believe that a "chip expert" wasn't aware of the Mac until the move to Intel

 

[cvaldes]

ARM Chip - if my memory serves me right was this originally developed, or something to do with Acorn Computers, Cambridge, UK, makers of the BBC Micro in the 1980s. Anyone got a more thorough knowledge of history? (I might be wrong?)

As cmaier mentioned, you are correct.

Wikipedia article: http://en.wikipedia.org/wiki/Acorn_Computers_Ltd