INTC is behind TSMC in wafer fab technology i.e. TSMC already in mass production 5nm node and working on its 3nm node.
Ok. But does that mean Intel won't make money?
I contend that you want to look at who is providing a good value to their customer, what is their volume of sales and what is their profit margin.
Often being on the bleeding edge is a great way to lose money.
Would you invest in an aircraft company based on who makes the fastest airplane?
Especially when the company making that fastest airplane has to have their engines fabricated by a third party? Especially if that third party is the only one who can fabricate those engines and knows it?
For me, Intel having their own fab is important. I believe it puts them in a significtanly different market position than many other companies.
I think AMD is going to lose pricing power over time because there will be more and more CPU designs out there and AMD does not have exclusive access to TSMC's fab. I think more and more of the total profit on a CPU will be going to the fab and less to the designer. Intel will not be hurt as badly because they have a fab.
IMO, the specific design of processors is going to become less and less of a differentiator over time. The game these days seems to be plopping down as many cores are possible, clocked as high as possible.
We're not really seeing change like that which happened from 8086 ->286 -> 386 ->486 -> Pentium
The last big change was going 64 bit and that was around 1999.
Now it's a matter of transistors per area, clock rate, and cost to fabricate that area.
(You can't discount the importance of that third parameter, because you have to recoup the cost of a REALLY expensive fab.)
I'd rather Intel not spend money like crazy to guarantee that no one ever beats it on those first two parameters, because doing so would damage the third.
If I were a pro, I would try to build a computer model looking at fab costs, expected speeds of processors, dollars per unit of processing power, etc. I bet you could find a spot in the model where Intel could be a few years behind TSMC in technology essentially forever and still make money. The fabs Intel would build, would be cheaper by not being as bleeding edge. Thus they could sell the product from that fab cheaper. Most people don't care what nanometer size the transistors in their CPU are, they want the processing horsepower. Intel might be able to sell more square area of silicon (more cores rather than higher clock rate) for less money because it's made on a cheaper process.
Now if the management can't get out of their own way, none of this matters, so it's certainly not a sure thing...
