Intel 10nm Delay Raises Speculation of Foundry Business Scale-Down

DooKey

DooKey

[H]F Junkie
Joined
Apr 25, 2001
Messages
12,909
By now everyone knows the 10nm Cannon Lake processors aren't going to hit the market until late next year. Further, due to these delays at the fab Intel has had trouble grabbing orders and there's speculation they will be scaling down their fabs. Also, in addition to this bad news for Intel, SemiAccurate is reporting that Intel's 10nm process isn't what they promised for 10nm and is in fact more of a 12nm process. Putting two and two together spells big time problems for Intel and I believe they'll get Cannon Lake out the doors and abandon 10nm for 7nm ASAP. There is one bright spot for Intel and their 14nm process because some sources believe that process is very competitive with rivals' 10nm and 7nm processes. Next year should be very interesting.

Intel continues to break and probably has abandoned the "Tick-Tock" cycle, with its long-delayed 10nm Cannon Lake processors, the observers noted. Intel was supposed to volume producing its 10nm Cannon Lake processors in mid-2016 but the company released a 14nm Kaby Lake instead. With Intel confirming another 10nm chip delay, the company's ninth-generation family slated for 2019 would be built on its 14nm+++ process technology, the observers said.
 
Apparently Intel's 10nm is expected to be, at first mind you, worse than 14nm++ on everything. Yields, performance (clocks), and power.

Now to be fair Intel's 14nm had some issues that made it not so good at first too. It wasn't until 14nm+ that it really got decent and that took over a year to come out after their 14nm was first introduced in late 2014 (though announced in 2013). So I'd expect their 10nm to improve over time but if SA's article is anything to go by it'll never be as good as originally advertised.

So TSMC's and GF's 7nm might actually turn out to be a fair bit better overall, and not just at launch either, but for a fair amount of time afterwards too.

The other interesting rumor is that Intel is trying to advance their 7nm process launch date and is renaming the updates to it as 5/3nm as well. Right now it seems to still be delayed to 2021/2022 time frame though.

Either which way it'll be damn interesting to see how this all plays out. I think its safe to say though that Intel can no longer count on having its once insurmountable process lead anymore from here on out.
 
I can't help but wonder if knowledge retention is an issue; Intel was ahead for a LONG while, and I can't help but wonder if Intel has lost personnel to retirement to the point where they've lost their primary knowledge base.
 
gamerk2 said:
I can't help but wonder if knowledge retention is an issue; Intel was ahead for a LONG while, and I can't help but wonder if Intel has lost personnel to retirement to the point where they've lost their primary knowledge base.
Click to expand...
According to Charlie they basically fired a huge amount of critical talent for a one time cost reduction. If you look around on twitter there are a bunch of those guys essentially saying SA is spot on here with their story.
 
  • Like
Reactions: c3k
like this
gamerk2 said:
I can't help but wonder if knowledge retention is an issue; Intel was ahead for a LONG while, and I can't help but wonder if Intel has lost personnel to retirement to the point where they've lost their primary knowledge base.
Click to expand...
What they lost was the corporate culture that made Intel great. "Constructive confrontation" was killed in the name of political correctness by the burgeoning middle-management bureaucracy in the late 1990's -- and that gave us the Pentium IV and Intel's commitment to RAMBUS, mistakes every engineer I talked to knew were going to be a hot mess. The company's been coasting on past glories (the entire processor line now is just a tweak of the old Pentium Pro architecture, which survived in laptops during the P4 era/error) and it's huge cash pile ever since.

That's why I don't work there anymore.
 
Last edited:
Eventually intel won't be ahead anymore but even with everyone else.
 
Montu said:
SemiAccurate is reporting that Intel's 10nm process isn't what they promised for 10nm and is in fact more of a 12nm process. Putting two and two together spells big time problems for Intel and I believe they'll get Cannon Lake out the doors and abandon 10nm for 7nm ASAP. There is one bright spot for Intel and their 14nm process because some sources believe that process is very competitive with rivals' 10nm and 7nm processes. Next year should be very interesting.
Click to expand...

Then it is doomed or not?
 
mesyn191 said:
Apparently Intel's 10nm is expected to be, at first mind you, worse than 14nm++ on everything. Yields, performance (clocks), and power.

Now to be fair Intel's 14nm had some issues that made it not so good at first too. It wasn't until 14nm+ that it really got decent and that took over a year to come out after their 14nm was first introduced in late 2014 (though announced in 2013). So I'd expect their 10nm to improve over time
Click to expand...

?url=https%3A%2F%2Fg.foolcdn.com%2Feditorial%2Fimages%2F438820%2Fintel-10nm.png
 
Montu said:
I don't think so. They just need to learn from their mistakes and press on. However, they are going to suffer some pain until they right the ship.
Click to expand...

I mean first part claims that 10nm will fail to hit targets, it must be more a kind of 12nm, and it would be just abandoned. Then in the second part it says that current 14nm technology is "very competitive" with TSMC/GF 7nm.
 
Another Charlie article. Just slide it into the ignore pile until it gets corroborated by other sources.
 
akaliel said:
Another Charlie article. Just slide it into the ignore pile until it gets corroborated by other sources.
Click to expand...

There is a lot of smoke with that fire. I think there is quite a bit of truth in his article but scaling down the foundry seems unlikely to me. I mean 10nm was such a hot mess they dumped what little they could make on to the Chinese market just to say they made them. But I am sure there is quite a bit of speculation in there still but I think things are getting a bit grim in Intel land.
 
You must log in or register to reply here.
Back
Top