AMD Ryzen 8000U and Ryzen 7000G series spotted in shipping manifests AMD is preparing at least several new parts right now, all based on the Zen4 architecture. There has been a discussion regarding the complex nomenclature employed for Ryzen 7000 mobile CPUs. With the unveiling of the series, AMD mixed multiple architectures into a single […]
Is that due to a shortage of the chips, or that Intel has a firm grip on the partners?
AMD has offered better mobile CPU:s than Intel for a long time now, yet the flagship models från Lenovo, Dell, etc continue to use Intel CPU:s. Doesn’t make sense unless Intel has exclusivity/offers a lot money to their partners to keep exclusivity.
Witht that said, it is my impression that the number of models with AMD CPU:s have slowly increased.
Intel only got fined last year after a decade of their anti-competitive practice regarding that, I’d be surprised if they still aren’t doing a small scale of that.
Its due to limited capacity from TSMC. AMD can only make so many chips. Server and AI are super lucrative right now, so AMD makes mostly those chips. Very little incentive to make these low margin chips when they can inly fab so many chips at a time.
TSMC is working to build more plants to solve the issue, but its years away.
Intel has its own fabs, and tons of capacity because no one is buying their chips. So, OEMs stick with Intel because they can produce as many chips as they want. This is why Intel still dominates prebuilt and laptop markets.
As others have said before, TSMC has capacity, they’ve had capacity since both PC and datacenter demand fell early this year. Shipments are currently lower this year than last year. Even before when AMD was supply constrained, it wasn’t wafers, but substrate supply and AMD resolved that early in the year.
TSMC is building more capacity because, cyclically, chip demand increases and advanced packaging supply is the priority right now.
AI server is low volume as far as wafers go, but it’s straining on advanced packaging. I believe server CPUs outship accelerators 10 to 1. AMD doesn’t use advanced packaging for their mainstream Phoenix Point APUs.
AI demand has also cannibalized the traditional server market, so many companies have shifted their budgets for this year towards GPUs.
It’s 8 GPUs to 1 CPU per server and each GPU roughly costs as much as a server without GPUs.
My guess for AMD’s notebook supply issues are due to their lack of demand visibility and poor marketing and sales relative to Intel.
I believe they’ve got a classic chicken egg dilemma. AMD has little supply because they have little demand and they have little demand because they have little supply. And even though they are gaining more design wins, the time-to-market is quite long as OEMs don’t seem to have them widely available until two whole quarters after Phoenix launched. That’s almost like the lead time of both the chip and system combined.
That’s just my guess. I heard there were platform teething issues, but I don’t actually know.
On Intel’s side, it’s often said that they spend a lot on their channel and I think this is kind of evident outside looking in. OEMs stick with Intel because Intel not only because of their volume (as a result of IDM), but because they support their scale by helping partners downstream whether it’s marketing or even design.
And that goes full cycle with Intel’s customers basically becoming captive to Intel’s ecosystem. Many enterprises simply buy Intel on a cycle. AMD is competitive, but they’re not truly disruptive in mobile, except maybe for integrated graphics, but that’s not that big of a market, yet.
There’s no reason to swap to AMD unless Intel screws up both on supply and technology.
I do wonder if Intel can be as effective as they outsource for Meteor Lake and actually begin to charge their business units for using the internal fabs.
Some of it is shortage of chips. GPD accused AMD of not supplying enough chips, and I’m guessing that’s a result of high demand, as quite a few mobile gaming devices are using Phoenix.