https://www.techspot.com/news/71255-tsmc-building-world-first-3nm-semiconductor-facility-taiwan.html
TSMC is building the world's first 3nm semiconductor facility in Taiwan
10 replies (most recent on top)
Thank you to those who are engaging on this thread. It's refreshing to finally read an intelligent exchange of ideas in this forum.
100% confirmed.
@1iqv The rush in pushing 10nm is to make idiots like Ashraf happy. Idiots who have no clue about chip technology at all, but who are easily impressed by marketing talk.
@1tzw: Uh, no, genius. PCs and servers are high transistor power products. Point to products that TSMC producing in that range right now. No, GPUs don't count.
They don't.
Premise stands that nobody cares what is inside a PC or phone. Intel can stay on 14nm a long time and no hurry to move to 10 or 7 and thus slower cadence and needs far less people and money.
Dang on that thought just have TSMC build it for Intel and dump TMG and all that wasted money and let cheap Chinese labor build it for you like Apple, Nvidia, and Qualcomm do with far faster and fatter margins.
Currently intel for has not reaped any benefit for being a IDM as Canon and Ice have languished, held hostage by delays on intel 10
Uh huh. Massive layoffs every time TSMC builds another fab. I don't think so.
Layoffs nowadays are for boosting flagging profits. Massive layoffs are saved for repositioning. Intel had just repositioned- The layoffs from now on are going to be yearly stealth ones that are smaller.
The only way there'd be another round of massive layoffs is that somehow PC and servers are massively replaced by ARM-based products.
Talk about 10nm all you want... Do you see 10nm products having higher transistor performance than 14 right now? Nope. Early 10 has lower performance than late 14 variants, so what's the rush to push 10 into everything? "Oh lookie! A shiny new process"? The bulk of what Intel is making money off of right now isn't going to benefit from a move to 10, performance-wise or economic-wise.
Only people who read too many of those junk "enthusiast websites" want everything shunk as fast as humanly possible. "10" sounds so much better to the average Joe than "14++"
@aux. I disagree. There are only two emojis in the main text. No distraction.
@ncp - With all due respect, you make a good argument. But I have to say your excessive use of emojis undercuts your message, whether it is of reference to BK or just the long run at the end of your message. In English vernacular, that emoji use doesn't add to your point - it actually detracts. Just something to consider.
TSMC will announce -1nm fabs next year. They’re good competition, but known to exaggerate.
Scale of huge size is required in semiconductors to feed critical R&D cost and building of ten billion dollar plus fabs to play in silicon is a must.
Two decades ago intel scale and profits on the back of x86 allowed them to ascend to become the worlds largest and and most advanced silicon manufacture and they dominated for two decades.
Yet their arrogance and dogmatic attachment to x86 set them on the inevitable path they are on now.
Intel problem is they make huge investment in leading edge and then need to retool it again for the next generation. Samsung and TSMC after getting their cellphone profits from the leading edge fab keep it running and making money for all sorts of other designs. The leading edge pays the R&D and depreciation while making a profit and the next 15 years they capture 60% margins for other things in th fab. Intel on the other hand closes fab and has to retool having failed completely in Foundry and can’t capture and continued revenue stream. No their yields on 10 are terrible and x86 doesn’t command margins they need on leading edge they are in a huge pickle they can’t escape!
The failure of CRB, PSO, and the 🤡 to win mobile and bungled delays on 22, 14 and 10 while Samsung and TSMC ride the volume of Mobile to become the new leaders for the next decade has pretty much assured where Intel will be in a decade.
Of course those McFubar and other intelers are still in denial the slip and public positioning of 10 and 7 on longer cadence is implied admission that x86 can’t lead any more, nor does Intel have the stomach to invest to pull in ththe schedule. Look at the public slip of 10nm as well as the empty shell F42 in the desert as where one company is going and the revenue and profits as well as stock market cap of Samsung and TSMC as three companies, with one going in different direction.
Tipping point has been pasted and intels future is assured right there with IBM, Motorola, Kodak. The only question can they make the right hand turn like IBM or end up like Motorola and Kodak
Amazing the BoD aren’t asking the 🤡 these hard questions and let him spend the billions they do t have
😂🤣😂🤣😂🤣😂🤣😂🤣🤣😂🤣😂🤣😉