Disclaimer that I know very little about the hardware industry so a lot of this will probably sound stupid as fuck. I'm basing this on a few tenuous at best assumptions: 1. ARM is the future and x86 will die a slow death. 2. China might invade or blockade TW, making chip supply from TSMC unstable or completely impossible. Intel has been struggling for years now but they still have an advantage in that they are really the only major player with fabs in the US and manufacture their own chips. Afaik AMD, nvidia and Qualcomm all outsource their manufacturing to TSMC. If TSMC is cut off by an invasion of China, what will these companies do? Would it be possible for intel to start taking manufacturing contracts from other firms and operate as a TSMC of the west? The main problem with this is that their fabs are probably nowhere near as advanced as the TSMC ones, but would the mere fact that they already have these factories set up be a big enough advantage? Their competitors are mostly fabless with no manufacturing capabilities.
TSMC is a sweatshop propped by Equipment suppliers from the West. As soon as Intel starts to really engage it's supplier base they can do much more than TSMC can. Have you ever heard of any innovation coming out of TSMC except shrink made possible by Equipment vendors?
Like i said i dont know much here. But ive just heard from various sources that TSMC has their new 2nm process figured out and intel is way behind, and i heard it's pretty difficult to catch up. If intel has the means to do this then why dont they already? I feel like they could do much better as a company this way than struggling to innovate on the dying x86 ISA.
I think once they admit that vendors are not equipment suppliers but innovation partners, they should be competitive again. But they don't have that culture! Someone needs to shake them a bit more.
No chance
I think your guess is pretty good. The only catch is that can Intel really pull it off though? They do have an opportunity in-front of them no doubt. But I highly doubt Intel can move fast and execute. Or at least haven’t seen them do such a thing in the past. But who knows…
Intels already working on this pivot for a couple of years now. https://www.theregister.com/2022/10/12/intel_sets_up_internal_foundry/
O wow had no idea, thanks for the article
OTOH this is not the first time they tried this, and they've had a hard time in the past trying to sell foundry space then: https://www.elektroniknet.de/international/intel-apparently-closes-down-foundry-business.161048.html
Yeah intel will do it. And also crash the life style company in cpu/gpu.
Technically it can, but the management are completely incompetent IBM-style farts and ho can’t prioritize properly or innovative.
I am more and more of the opinion that TSMC and fabless companies are over hyped at this point in time.
If China invades Taiwan, and it's a big if, then the world needs to be worried about a lot more than a shortage of chips. Anyways, TSMC is expanding in the US with an advanced process node but not their top of line process node which is still reserved for Hsinchu. This is simply to keep the US happy and hedge its bets while still not losing the strategic advantage of having it's best fabs in Taiwan. Intel needs to learn how to support customers in the development of fab. Fabs are not a one way street where some foundry engineers decide "Oh well these are the electrical characteristics of the transistors. Let's ask Apple to design with it." The tier 1 players, whose orders justify spending billions on a new fab, dictate the lyrics and the tune of what the fabs make. Small players don't get a say unfortunately. Intel is new to this game. Let's see how they fare. It's not only technology, it's the customer engagement that makes or breaks a fab line. Also you don't need 3nm for every application. You don't need 3nm for the computer responsible for the infotainment system in your car, 10nm will be more than enough. Intel can always fill this role. The latest and greatest development in HW is no longer achieved by going to the latest process node, but with new architectures. It's suddenly a lot more hard work than yester years. Huawei already announced a new phone completely developed without any sanctioned American technology. It maybe terrible but they will get better. And better.
Regarding the lower end 10nm chips, i think there are existing players in that slice of the market like global foundry, although i dont know how dominant they are. Also regarding the taiwan invasion, if it does happen, obviously we will have a humanitarian crisis first and foremost. But economically speaking, the microchip industry is probably the most important thing coming out of taiwan so i think its pretty high on the list of things to worry about.
Intel has all the tools and resources to be a giant chip manufacturer. What are the problems then? 1- Intel is producing its own chips (cpu, bluetooth, graphic) and it requires a huge amount of human resource and very strong partnerships. Intel lost a significant market sofar due to not having a constructive relationship with hyperscaler like Apple and etc. They need to make the partnerships again. 2- Intel management is pretty old fashioned and after introducing OKR by Andrew Grove, not too much innovation happened in the way they run the company. Being said, they made the company less attractive for top talents. As you may know the whole innovation in each company happens by less than 30% of the top talent and rest are helping to build it. Intel is not that attractive any more. 3- The trend of education system went more smoothly towards software and we saw less students interested in chip development. As a result, less sharp talent has been accessible for Intel and other companies with bigger market capitalization absorbed not only fresh graduates , but also talent from Intel. 4- Marketing, Sale and Smart Investment are isolated islands in Intel. They lost the market then had no income. Sale’s didn’t help on creating new market or foreseen future activities and they worked very old fashioned. Intel investment arm is not actively supporting enterprises which may be acquired or become part of system later time. If you listen to their talks, they insist they don’t give a single thought about Intel and just looking for making money by investments. They may want to gain trust in their customers interactions, but it became a culture in the firm and hard to change the story. Intel need to gain trust from their current customers, creat new customer base and act very wisely at the top level.
How do you feel about pat gelsinger? It seems that theres a positive buzz about him ever since he took over and his overall plan and strategy for intel seems sound to me.
It’s not about my feeling. But he has the credit, experience and power to make change. Luckily, government is supporting him too. But he needs more moonshots than just managing people to return Intel to its original position.
That's pretty much what the CHIP act aims to do. If the Taiwan straight really heats up, this just might happen sooner. But even so, as long as TSMC comes to the US, I don't think anyone would object. The goal now is to get TSMC Arizona up to capacity before anything stupid happens in the Taiwan straight afaik.