It was just two days ago when we got word that Intel’s 7nm node was 12 months behind schedule. Though we were hoping to leak this as an exclusive soon, it seems like Intel wants to get ahead of the press for once and outed it themselves in their quarterly earnings report. Intel has said that their yields are 12 months behind internal targets (this is the snippet we caught) and that this would result in 7nm launching 6 months behind schedule, resulting in not only a late launch but also a premature launch in terms of yields.
However, we still have something to report on regarding this recent news. Our sources reported to us that not only was there a delay, but also that Intel would not be relying on TSMC in any capacity to make up for this delay. This is in respect to CPUs, however; the Xe project seems to have some leeway with TSMC, since DG2 will apparently be fabbed there according to our sources in January and parts of Ponte Vecchio could be as well (Samsung is also a possibility). Though former Intel executive Jim Keller pushed for Intel’s processors to be fabbed at TSMC, leading to conflict with colleague Murphy Renduchintala, it seems that Intel has finally decided to keep their CPUs on their own fabs.
What does this mean for Intel? Well, while they retain their pride and independence, their roadmap has been further uprooted (if it even still exists at this point) and in the long term, further delays will hurt just as badly. Intel could have pivoted to TSMC at some point to mitigate the damage these delays cause, damage which is already manifesting in Intel’s stock price, which has fallen around 10% in after hours trading. 7nm is now 5 years behind schedule, and while they are far from bankruptcy, Intel has contenders on all sides wanting a piece of the silicon empire that they have grown in the past decade.
Our opinion is that Intel is making a huge mistake not going to TSMC, and they should have done so a long time ago. It’s not that a 6 month delay is deadly for Intel, it’s that a 6 month delay on top of a 4+ year delay further erodes Intel’s ability to compete with its rivals. Having an inferior node is not a death sentence for Intel (AMD has served a worthy competitor at times without a node advantage, for example), but the node disparity when 7nm finally launches will be hard for Intel to negate.