Intel’s 10nm Cannon Lake chips are delayed again

Intel is once again delaying the release of its next-generation 10nm Cannon Lake processors, with mass production on the chips now expected for 2019 instead of the end of 2018 as originally planned, via Engadget. For those who haven’t been keeping up with Intel’s internal code names, Cannon Lake is Intel’s upcoming 10nm chip architecture, and the next step from the company toward creating smaller and faster processors.

Last year, when Intel first launched its eighth-generation of Core chips, the company broke from previous policy by announcing that the product lineup would consist of a mix of multiple architectures. The line launched with a revised version of the seventh-generation 14nm+ node (Kaby Lake R), and earlier this month saw the release of chips built on the 14nm++ node (Coffee Lake), Intel also promised that we’d see eighth-generation chips from the next-generation 10nm Cannon Lake node too, which Intel was originally planning to release back in 2016.

Part of the delay comes from the increasing breakdown in Moore’s Law — as we approach smaller and smaller transistor sizes, it becomes increasingly harder for companies like Intel to keep up with the two year doubling that Moore’s Law demands. It’s also why we’re seeing things like the Kaby Lake R and Coffee Lake releases, or AMD’s updated Ryzen 2 chips, which look to improve on the existing technologies at the current node rather than try to force an early jump ahead to the smaller size.

Intel isn’t coming off the best few weeks in the news: Apple is reportedly planning on dropping Intel processors for its Mac computers as early as 2020, and last week Intel abruptly abandoned its Vaunt smart glasses project (to say nothing of the disastrous Meltdown and Spectre debacle earlier this year.) That said, Intel did beat estimates for its quarterly earnings this week, posting a record $16.07 billion in revenue largely on the strength of its PC processor and data center businesses, so don’t exactly count the famed chipmaker out yet.


Maybe the hardware-side protections against Spectre, Meltdown, and the like are proving to be more difficult to implement than Intel first thought.

Maybe.. But those have little to do with the 10nm process node.

I know; my comment was half tongue-in-cheek (based on the fact that it is obvious that isn’t the real reason for the delay, as Intel is happily shipping this year another batch of 14-nm processors, Whiskey Lake and Cascade Lake, a detail that was not mentioned on this article).

I bet Whiskey Lake will still have 0% IPC improvements over Sky Lake (Well, maybe 5% tops) with a bit more boost and more power efficiency. Yet, will be called 9th gen.

Too bad for Intel. They really need new chips in their fight against AMD.

Not really, intel has all the chips they need to fight amd, they just need to cut their prices.

Indeed. Intel has clearly looked at their books and decided that the number of users switching to AMD represents a much smaller revenue loss than simply cutting the prices of their chips to be more competitive.

High-end desktop users, where Ryzen / Threadripper are competitive, represents a very small part of the total addressable market. And an even smaller portion of that market niche will be willing to throw away their brand trust in Intel products that they’ve grown used to over the past decade (at least) since the introduction of their Core line.

Well Jim Keller just joined Intel. So expect something new in the future from intel.

I think Intel’s major revenue is from their enterprise/datacenter side of business (Xeon processors). So from a consumer standpoint, YES, it looks huge to lose on things like Apple choosing to ditch Intel chips for their new Mac’s or the delay in launching the 10nm chips. But it would be not make a major dent on their revenues and hence not a big loss from an investor or revenue standpoint (I guess).

Just my theory ..

Plus, they manufacture plenty of chips for other makers (including ARM designs). Delaying their own consumer chips probably won’t hurt their manufacturing arm at all.

Too bad AMD doesn’t have contracts with all the OEMs that would put their CPUs front and center when you order a new laptop, let’s say.

Not to mention AMD is nowhere to be found on Apple products.

It’s also why we’re seeing things like the Kaby Lake R and Coffee Lake releases, or AMD’s updated Ryzen 2 chips, which look to improve on the existing technologies at the current node rather than try to force an early jump ahead to the smaller size.

Quick FYI, the Zen+ architecture that’s used in Ryzen 2 is actually on a slightly smaller node then Zen (12nm vs. 14nm, respectively).

Intel deserves all of this because of the way they handled spectre and meltdown. Hope the stock tanks

It was up 8% after earnings this week.

So not just 7nm, we may get 7nm EUV before it? Other fab nodes usually name one ahead of where they actually are vs Intel, but I think 7nm EUV is truly competitive with Intel 10nm.

Point of order: can we please stop talking referring to ’Moore’s ’’Law’’’ without multiple quotations and a general air of derision? It was a prediction by a self-interested engineer. Not some inexorable ‘law’ of physics.

It’s also dead (atm at least).

I want to know what is behind all these damn delays in Intel’s process. Global Foundries, TSCMC, and Samsung have all surpassed Intel at this point. Why?

Have they? Are they churning out 10nm or 7nm chips with ease?

Yup. Samsung is on 10nm+ this year. TSCMC is on 12nm. Both are on schedule to have a 7nm process ready next year. Global Foundries went to 12nm this year (just started mass production), is skipping 10nm and going directly to 7nm. 7nm will be a 2019 product. AMD already has 7nm test Ryzen chips and Vega GPUs. Global Foundries and TSCMC will be mass producing 7nm GPUs by Q2 2019 at the latest.

They’re not comparable though. It’s true that TSMC and Samsung are starting to get ahead of intel, but the process technologies aren’t apples for apples.
As I stated over, Intel 10nm is closer to TSMC 7nm.

The thing to be aware of is the actual node size is different from the advertised node size. Specifically, a chip manufactured at an advertised size of 14nm may not actually have any part that scales to that size. Intel’s are closest to their advertised size. So if you normalise the node sizes across manufacturers, Intel is actually still ahead in fabrication process. The gap is closing fast and in 2019 Intel will probably have lost their fabrication process lead altogether – but they certainly aren’t behind!

Basically, don’t bother comparing chips on their advertised node size, it’s just marketing and mostly meaningless.

Not true, TSMC 7nm actually has smaller half pitch than Intel 10nm, not even half node smaller, but it is indeed smaller. There is no way to downplay this as Intel will have lost their process lead that has been going for two decades.

Intel is concerned about AMD, Jim Keller has just joined them.

View All Comments
Back to top ↑