The contracts were long (10-15yrs), used the same node and were high margin because of it. Once the node depreciated in 4yrs your margins went sky-high because the fab and process tech tools were already written off. You basically pay for labor and consumables.
(...)
Where this model of high margins falls apart is when you go fabless. You don't own the asset that is depreciating, so you don't scoop up high margins once the first four years have passed. The foundry does.
IDC, this is one of your best posts ever, I didn't reply to it before because I was travelling and wouldn't have the time to properly answer it.
I don't think the contracts should have good *gross* margins, because you don't go for the bleeding edge. If I were to choose a node I would chose a node that just intercepted the price curve of the previous node, because this way I would ensure that there wouldn't be too much cost reduction left to the foundry. On top of that, you don't need a TSMC for that, you have plenty of options on the market that can deliver 65 or 55nm for example, this should add more price pressure on the contract. No supplier management worth its salt will leave too much for the foundry here, and this is just the basic drill.
As you cleverly pointed out, what is interesting on these contracts is the *operational* margins, because you will have just a small SG&A, but no R&D and no depreciation/amortization. I don't see this yielding high returns because of the pressures I pointed in the previous paragraph. But it's a constant stream of money, constant ROI, it does make sense for TI to keep those old fabs working.
But what I would like to point out is that this embedded business model is another confirmation that AMD is leaving the bleeding edge.
When you are on the bleeding edge, you need to rake the biggest amount of money as soon as possible in order to invest on the next big thing, then rake money... you got the picture. Life of a bleeding edge chips is something like 3, 4 years. With embedded, you are aiming for, let's say, the same amount of cash flows but spread in a 10-15y period. This is very important, as
adjusted for present value, the same amount of cash flows of a bleeding edge operation is far bigger than the value of a embedded operation.
The cash flows spread around such a large spam also hinders your ability to invest. You cannot sustain a bleeding edge operation with cash flows of an embedded operation, and embedded operations don't get many benefits from bleeding edge investments.
In AMD's case, the fact that they have no control on the factories just ensures that whatever margins they have, it will be small. The lion' share will be to GLF's. All in all, I see AMD embedded business more of a consequence of the shifts of their business model than a recently discovered business opportunity. It's just AMD trying to scrap a little more of cash and sell WSA chips.