“Thanks to the folks at Chipworks, we now know how large the A9X chip inside of Apple’s iPad Pro is,” Ashraf Eassa writes for The Motley Fool. “Armed with this knowledge, I’d like to offer up an estimate of how much it costs to build that particular chip.”
“There are three major variables that go into determining the manufacturing cost of a chip: Finished wafer cost; chip die size; [and] manufacturing yields,” Eassa writes. “The impact of silicon wafer cost should be obvious. The chip die size impacts both the total number of chips that can be physically built on a given wafer as well as the manufacturing yields of the chip (there is more that can ‘go wrong’ on a larger chip than on a smaller chip).”
Eassa writes, “I arrive at an estimate of $37.30 to manufacture the A9X.”
See how Ashraf got to $37.30 in the full article here.
MacDailyNews Take: Seems like it’s in the realm of possibility. If so, the A9X is markedly more expensive than the A9, which Eassa estimates costs Apple between $22-$24 per chip (which is in line with an IHS iSuppli estimate).
I estimate $157, or maybe $19.95. Pulling numbers out of your ass can be fun.
i like pulling the number 000000000000 out of my ass, it is much more fun
How does the cost of the A9 and A9X compare to Intel’s various CPU products? That would be interesting…
It all comes down to the number of square mm, and yield. Both are probably good designs with reasonable yield. So the price for comparable performance parts is probably identical.
Not really comparable since A9 and A9X are SoCs, not CPUs. The CPU cores on them only take up a fraction of the space.
It is still a relevant question, since it would play into the total product cost. I understand your distinction between the two.
Think about it. The chip could only be made by this company after it had made previous versions and only thanks to lots of r&d. To say that the cost is “x,” is to ignore almost all of the important thought, decisions, and years of experience which were involved. I find the premise to be unworthy of print, let alone reference.
You should re-read the article:
“There are three major variables that go into determining the manufacturing cost of a chip.”
Nowhere is he estimating R&D, just cost to manufacture.
Apple’s R&D is only about 3% of revenue, so would only have a small impact when that is distributed across all their products and designed parts.
As Steve used to say, innovation doesn’t come from spending lots of money.
“You should re-read the article:” (ouch!)
So the cost of manufacturing does not include anything done previously? Nothing that was learned at earlier stages is part of the cost of producing an item? I guess this means that diamonds should be nearly free because we ignore the previous digging that created the mine?
The point that I am trying to make is the word “manufacture” is not equal to the word “raw materials,” the word “manufacture” is equivalent to “cost to make” which involves quite a few (expensive) steps, especially in the tech world.
Ignoring the semantics, the “cost to manufacture” (in this context) means the total of the wholesale price Apple is paying for all components that go into device, plus the associated labour for putting them together. This is an important reference point for determining profit margins, among other things. R&D, marketing and other costs are irrelevant here.
I don’t mean to ignore semantics, I mean to point out the difference between current or future cost of production of an item (in this case a chip) and the cost to manufacture an item. That is, production cost is not manufacture cost, and I agree that knowing the cost to (continue) to produce an item is important.
People often wonder why Apple doesn’t sell an iPhone 4s for $99 because they confuse what something is worth with what it costs to produce. Likewise, we should make sure we understand that the cost of manufacturing an item includes a great deal of history and is much higher than production cost to continue to make each item.
What you say is completely correct. The reason I responded is that it is also not relevant to the discussion here, which is focusing on the manufacturing cost of a specific chip, as it compares to the competing chip makers. In other words, is current chip maker offering the best deal for Apple, or should they look at Intel, or other makers. It also looks at it from the perspective of profit margins on the final device (iPhone 6s), as it compares to the prior generations. Bill of parts (as well as manufacturing cost) are numbers frequently discussed and used to measure profitability.
Figuring out the rest of the expenses is practically impossible. For example, Apple has a large team of developers who work on OSX; sometimes, they are on Mac OS, sometimes, they are on iOS. When on iOS, their work is for all iOS devices, including AppleTV. How does one break down this work in order to determine exactly how much was spent on development of the iPhone?
The point is, we simply don’t even try to do this. Instead, we estimate what percentage of Apple’s gross revenue is spent on R&D (for the entire line of products and services) and we leave it at that.
There is a reason why there are companies like Isupply, whose business is to estimate component cost. When we know exactly how much does it cost to make an iPhone, we can estimate how long would it take for Apple to recover the money invested in R&D, advertising, etc.
They could cost a penny if someone was willing to pay millions or even billions for the first one.
If you exclude R&D, equipment, security, clean room costs, human resource managers, and on and on, the cost is . . . free. Sand doesn’t cost any money, unless the government gets involved. 🙂
I wonder what Apple’s unit price for Intel chips is vs the $37.30?
What about the license cost to ARM – the British company that actually designs the chip and then licenses that design to 3rd parties like Apple to tailor to their needs. That, presumably, is also a significant part of the cost.
ARM does supply off-the-shelf chip designs, and that is what most Android OEMs use as their chips, including Samsung (until recently when they customized the design of one or two of their chips).
But Apple doesn’t license chip designs, it licenses the chip instruction sets. Apple designs most of its GPUs (though maybe not GPUs so much). They bought PA SemiConductor after all. It’s not a case of Jony scribbling possible designs on a napkin, wadding them up, chucking them towards the bin, and saying, “darn, this is harder than I thought, let’s just go with ARMs design”.