2003: Intel deliberately went for a lousy microarch for desktops
Intel can't compete on price, the other competitors are happy to live with lower margins and turn those savings into OEM profits or a lower street price, either way Intel is not the first choice.
Apple or Qualcomm don't need to inwest billions and billions dollars into technology process and manufacturing plants like Intel.
That's the main reason why Intel don't like low margins unlike it's ARM competitors.
Intel has no high end SoC. They have only BayTrail. And you can't go into the high end with 1/2 half of the GPU performance of nearly every other high-end ARM SoC.
It's not correct: Pentium 4 was released late Autumn 2000 and first models were slower than Pentium 3.
Apple, Qualcomm or smaller players don't need to inwest billions and billions dollars into technology process and manufacturing plants like Intel.
That's the main reason why Intel don't like low margins unlike it's ARM competitors.
Why? Because you can use legacy software which run with 1 frames?
AMD also had similar press release wars with intel, back when they had a clear performance and efficiency lead. What is AMD doing these days?
Heck I think that is exactly what TSMC and ARM players wanted, let Intel foot the bulk of the R&D bill for shrinking nodes because they know Intel *must* have the best process while they pay much less shrinking half a node behind.
You don't really think that though, right?
TSMC would love to have Intel's process tech and node advantage. And its customers would love for TSMC to have it as well.
What TSMC doesn't have is Intel's R&D budget and R&D pipeline know-how. But they do all right for themselves.
You don't really think that though, right?
TSMC would love to have Intel's process tech and node advantage. And its customers would love for TSMC to have it as well.
What TSMC doesn't have is Intel's R&D budget and R&D pipeline know-how. But they do all right for themselves.
You don't really think that though, right?
TSMC would love to have Intel's process tech and node advantage. And its customers would love for TSMC to have it as well.
What TSMC doesn't have is Intel's R&D budget and R&D pipeline know-how. But they do all right for themselves.
No, they just pay TSMC's margin
Fabs are great if you can keep them fully loaded...
Heck I think that is exactly what TSMC and ARM players wanted, let Intel foot the bulk of the R&D bill for shrinking nodes because they know Intel *must* have the best process while they pay much less shrinking half a node behind.
Ofcourse firstmovers pay. Intel pay for the suppliers learning to. Imagine intel using the same node as everybody else? Lol. There is no such thing. Intel need to reinvent their business.
Excactly. Tsmc can because everyone and his brother uses their capacity from expensive phones to microowens. Giving the perfect depreciation while Intel is shut down here. Even usb is run on 32nm. Utterly waste of ressources.
As for all the doubters, I just find it funny. We'll just wait and see what happens I suppose, it will be quite funny to see ARMH's hubris catch up to them. AMD also had similar press release wars with intel, back when they had a clear performance and efficiency lead. What is AMD doing these days?
AMD and ARM are/were in pretty different places. AMD tried to poach Intel's natural habitat during a prolonged period of suckage in their bread-and-butter parts, and Intel returned to form and retook its traditional home. ARM on the other hand is entrenched in a market where Intel has next to no experience or traction, and Intel are desperately trying to break into it.
It's going to be interesting, and I think Intel has a very good chance. But I wouldn't be writing ARM's obituary just yet...
This is the main problem I see for Intel. I am old school, and strongly value x86 and TBH, really hate android, especially in anything larger than a phone. But I work in a large university with a lot of undergrads, and android is to them what x86 is to me... the OS they have grown up with. IoS and android have such a large user base and so many apps that even if Intel (or AMD) develop superior hardware, which Intel at least has the resources to do, I think they will have a hard time breaking into the tablet/smartphone market.
Personally, I see this as a trivialization or dumbing down of the computing experience, but it is undoubtedly happening.
Mobile market means lower margins, more competition and less impact of marketing (as it's the device brand that sells, not the cpu brand). A consequence of declining PC sales and entering the highly fought mobile market is less profit. Add in skyrocketing costs for fab upgrades that can only be maintained with large enough unit sales, both to redeem fix costs on each node for process R&D and mask costs (which also increase together with wafer and lithography costs), the overall costs of staying one nose ahead are about to increase while margins and most likely profits are about to decrease. So Intel's first mover advantage is largely reduced with the next few nodes. With multiple exposure/patterning techniques, their increasing costs but diminishing returns on performance and transistor costs, it will be very hard to maintain the process advantage beyond the next few nodes (let's say 10nm), absent some drastic change in the industry (or consumer behavior).Ofcourse firstmovers pay. Intel pay for the suppliers learning to. Imagine intel using the same node as everybody else? Lol. There is no such thing. They are stuck. It shows Intels need to reinvent their business more fundamentally.
Excactly. Tsmc can because everyone and his brother uses their capacity from expensive phones to microowens. Giving the perfect depreciation while Intel is shut down here. Even usb is run on 32nm. Utterly waste of ressources.
You think tsmc profit is bad. You have to realise this is not the former soviet union, and the profit matches the business effectiveness. And that includes the customer at least as well.
Their chipsets relative power draw is pretty large when the CPU is idling, they have to at most fab it a node behind the CPU to keep it overall power consumption down.
As for why they would rather shutdown their old plants I bet it's unprofitable to fab for 3rd parties or/and unwillingness from potential clients. Most likely because TSMC does it cheaper and have a TON more experience with custom designs.
Excactly. Tsmc can because everyone and his brother uses their capacity from expensive phones to microowens. Giving the perfect depreciation while Intel is shut down here. Even usb is run on 32nm. Utterly waste of ressources.
You think tsmc profit is bad. You have to realise this is not the former soviet union, and the profit matches the business effectiveness. And that includes the customer at least as well.
I'm just looking forward to a real price war. Intel's old way of doing business simply cannot work here, so they'll be forced to bite some hard bullets and in the end they'll need to accept less revenues and profits, and drag Qualcomm down with them. The real winner will be the consumer.