It appears that 2020 and 2021 are going to be long years for Intel. CFO George Davis presented at the Morgan Stanley conference yesterday covering a wide range of topics, but noted that despite being "undoubtedly in the 10nm era," the company felt that it would not reach process parity with competitors until it produces the 7nm node at the tail end of 2021. Davis also said that Intel wouldn't regain process leadership until it produces the 5nm node at an unspecified date.
Davis commented that the company was "definitely in the 10nm era" with Ice Lake client chips and networking ASICs already shipping, along with the pending release of discrete GPUs and Ice Lake Xeons. Intel is also moving well along the path of inter-node development, which consists of "+" revisions to existing processes. Davis said the 10nm inter-node step provides a "step-function move" with the Tiger Lake chips based on the 10nm+ process as the company awaits its 7nm process.
However, Davis noted that in spite of the shipping products and pending "+" revisions to the 10nm process, its process node still lags behind competitors, stating:
"So we bring a lot of capability to the table for our customers, in addition to the CPU, and we feel like we're starting to see the acceleration on the process side that we have been talking about to get back to parity in the 7nm generation and regain leadership in the 5nm generation."
Intel's 10nm process provides comparable density to TSMC's competing 7nm process, so it's hard to tell if Davis's reference is to either the performance of the 10nm node, or how economically the company can produce it. In either case, Davis predicts Intel will regain parity with the industry at the end of 2021 with its own 7nm node (presumably with TSMC's 5nm being the interception point). That will obviously have an impact on Intel's competitive positioning and financial performance in the interim, particularly as the company is now fully embroiled in a price war with AMD.
Davis says the company will offset those challenges by offering differentiated platform-level solutions, which includes tight hardware integration on the AI and software fronts, but he also noted that the company's 10nm node won't be as successful as Intel's prior nodes:
"As we said back at our analyst day in May of 19: Look, this isn't just going to be the best node that Intel has ever had. It's going to be less productive than 14nm, less productive than 22nm, but we're excited about the improvements that we're seeing and we expect to start the 7nm period with a much better profile of performance over that starting at the end of 2021."
Davis noted that the company is trying to be clear with investors about the impacts of 10nm on the company's gross margins: "...but the fact is that I wanted to be clear what was happening during the 10nm generation. The fact is, it isn't going to be as strong a node as people would expect from 14nm or what they'll see in 7nm."
"Also, we were at a time when in order to regain process leadership we had to accelerate the overlap between 10nm, 7nm, and then 7nm and 5nm, so the cost that you're absorbing, starting in particular in 2021, you’ve got this intersection of the performance of 10nm, the investment in 7nm, and were also well into starting the investment in 5nm: All of those elements just combine to impact gross margin."
Davis also noted that "....the effect of 10nm in 2021. It's sort of built today, you've got to get through that product cycle and the node. We're excited about the products, but the node isn't going to be quite the performer that historically we've had."
Intel plans to roll out the 7nm process in late 2021, and the company hasn't released an official projection for the launch of its 5nm node. TSMC is moving along briskly with its new processes and should be on the 3nm node in late 2022, so it's unclear if Intel's projection to regain leadership with its 5nm node are based on TSMC's current schedule for 3nm.
In the interim, Intel will have to fend off the resurgent AMD, which could be painful, particularly in the data center. When asked if the company was forecasting a share loss in servers, Davis responded, "We expect to see stronger competitive dynamics in the second half of this year. We thought we would see that sooner, but what we've seen is very strong demand for our products. …we do expect there to be strong competition through this period…as we look at our product roadmap we expect to offer a more competitive and compelling position as we go from 7nm to 5nm."
Intel is facing challenges, but the company has signaled a new focus on aligning its vast IP portfolio under new leadership and a new six-pillar ethos. As such, we can expect to see Intel continue to invest in its latest tech that isn't entirely contingent upon process leadership, like EMIB and Foveros, and adopt new chiplet-based architectures across the breadth of its portfolio to both leverage its packaging advantage and sidestep some of the issues with shrinking some functions to smaller nodes. We also expect Intel will continue to work on making its architectures portable across nodes to blunt the blow from potential future missteps.
AMD apparently has not reached process parity with Intel 14nm in terms of clock rates, as you can see from the 5.3GHz boost rates.
On the cost ... Swan has repeated that their 14nm chips are very profitable since much of the 14nm equipment is fully depreciated. Intel continues to report higher margins, so ...
Let's wait and see how these 5.3 leaks / rumours perform ... I'm thinking the 5.3 is their new level three boost only that will work as well as Ryzen's 4.7 boost ... you get a 0.5 second boost, because it would be thermal throttle city if it lasted much longer.
Keep in mind Intel would never have hit 5.0+ on 14nm if, 1) 10nm worked initially as intended, and 2) AMD didn't suddenly became an aggressive competitor. Had 10nm worked out 3-4 years ago, the increase in IPC would have kept it well ahead of Ryzen one launch, even if clocks stayed around the 4ghz mark (as Intel's 14nm was at the time) ... then Intel would have moved to 10nm. But that didn't happen ... 10nm didn't work and Intel needed to do whatever they could to help stave off zen and zen plus, so focused on refining, and focused on refining, and focused on refining the same architecture ... this wasn't a plan, this isn't "normal" - it was a result of being forced into it due to a chaotic situation. I think everyone would have laughed (including Intel) if in 2016, one said that Intel's desktop CPUs would be pulling up to 300w at the processor (rumour alert) - ... the company that was making some incredibly efficient CPUs at the time. But here we are ... it certainly wasn't what they planned, but it had the nice side effect of keeping a few gamers happy.
So I see the 5.0ghz+ frequencies they are hitting as an exception ... not any rule. 10nm will never hit that clock, Zen3 will never hit that clock and I doubt any smaller nodes from this point on ever will as well - from either camp.
At the end of the day though 7nm, 10nm, 14nm, 5.3 ghz are all just labels ... the only thing that is really going to count is performance for a given task, performance per dollar, and the amount of extra money you need to spend on cooling to get that performance (contributing to performance per dollar) - that's what drives sales for the most part. I, mean big numbers on a box do as well, but who wants to admit that they make their purchasing decisions based on marketing tactics as opposed to actual product value based on performance for a given need?
Given their rapid, on-time perfect 10 mm desktop CPU deployment, I personally and officially believe everything Intel says, without reservation!!! :/ (I mean, who would EVER doubt them going to 7 nm in a year, and 5 nm a few months later?)
More important is where are the manufacturers of photo-lithography equipment in their current development cycle.
I think that is why Intel may be the better long-term bet.
I didn't see AMD make much of a dent in server sales last quarter, so apparently the purchasing decisions were made on something other than what AMD was offering. Maybe avx512, optane support, dlboost and higher boost clocks were the difference. Or, maybe Intel's large 14nm capacity just lets them supply the chip orders that tsmc had no room for.
one thing. Intel 7nm is about the same as TSMC 5nm, so They would be even in that. But in the meanwhile amd seems to have edge with TSMC.
Well intel's stupid architecture allows them to use less wafer space for their whole CPU plus the giant (in dimensions at least) iGPU than AMD needs only for the cores plus I/O.
Complete 9900k= ~178nm2
Ryzen 3rd Gen Eight Core Zen 2 chiplet 10.53 x 7.67 mm = 80.80 mm2, IO die is 13.16mm x 9.32 mm = 122.63 mm2 ,together more than 200nm2.