Intel 18A Panther Lake and Clearwater Forest CPUs are booting — steady progress toward the next-gen lithography node
The first third-party product will tape out in 1H 2025.
On Tuesday, Intel provided an update on the progress of its 18A (1.8nm-class) fabrication process, a crucial technology for its foundry initiative. By now, the company has a process design kit (PDK) version 1.0 ready, so its third-party customers can start (or even finalize) the development of chips in this manufacturing process. Furthermore, two essential Intel products using this production node have powered on, which is a good sign.
"The Panther Lake client processor is powered on and booting Windows, yielding well, in use inside Intel and ahead of schedule on product qualification milestones," said Kevin O'Buckley, Intel senior vice president and general manager of Foundry Services. "Clearwater Forest [CPU] for datacenter is powered on, booting operating systems, in use inside Intel and performing well."
Intel's 18A is the company's second fabrication technology to use gate-all-around RibbonFET transistors and backside power delivery called PowerVia (particularly crucial for data center-class products that need a lot of power) after 20A. Compared to the 2nm-class manufacturing process, 18A promises optimized RibbonFET design and some other enhancements, which leads to a 10% improvement in performance per watt.
Perhaps more important is that Intel 18A is a process that Intel Foundry's potential customers are very interested in, as it is believed to be more competitive than TSMC's 3nm and 2nm-class offerings available in 2024 – 2025. As a result, it is crucially vital for Intel and its ecosystem partners, such as Ansys, Cadence, Synopsys, and Siemens EDA, to adjust their tools for PDK 1.0 so as to enable Intel's lead customers to finalize their 18A designs that are currently in development and other clients to begin developing their 18A products.
"Ecosystem partners are updating EDA and IP process flows and tools to the Process Design Kit (PDK) 1.0, which will enable customers to begin their final production designs," said O'Buckley. "We are seeing continued interest from external foundry customers who are actively designing on Intel 18A. These positive outcomes are a signal for fabless customers and the industry at large that IDM 2.0 and our systems foundry strategy is working."
Intel expects its first external customer to tape out its first 18A design in the first half of 2025, so expect that design (assuming it is free of bugs and defects) to enter high-volume production in the first half of 2026. It means that Intel's 18A will be a little behind TSMC's N2 (a 2nm-class technology), which is set to be used for high-volume manufacturing in the second half of 2025.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Anton Shilov is a contributing writer at Tom’s Hardware. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.
-
Reality_checker Looks good.Reply
If it is on track, Intel is set to regain leadership in the Data Center segment by the time Granite Rapids comes out.
Competition is good, right? -
JRStern from your Panther Lake link:Reply
>It's close to impossible to tell how these CPUs will stack up against
>Meteor Lake, Arrow Lake, and Lunar Lake in terms of performance.
Also from the link, Panther Lake 15w-25w TDP meant for laptops ... maybe I'll give up the workstation and go for laptop plus docking station, I dunno. What about staging RAM inside the processor package?
Intel sure has a lot of irons in the fire. -
Roland Of Gilead "The Panther Lake client processor is powered on and booting Windows, yielding well, in use inside Intel and ahead of schedule on product qualification milestones," said Kevin O'Buckley, Intel senior vice president and general manager of Foundry Services. "Clearwater Forest for datacenter is powered on, booting operating systems, in use inside Intel and performing well."Reply
Well, of course they're gonna trump this kinda crap out. After the debacle of the last few weeks with the continuing issues for 13th/14th Gen CPU's they have to try put some kinda positive spin out. I'd say it's squeaky bum time at Intel right now! -
vanadiel007 It always amazes me how they are able to develop so many transistor circuits. That schematic must be crazy big at normal viewing size.Reply -
gg83
Either this is to ward off further stock drops or this is great news. I hope Intel gets their act together.Reality_checker said:Looks good.
If it is on track, Intel is set to regain leadership in the Data Center segment by the time Granite Rapids comes out.
Competition is good, right? -
TheSecondPower
(Why not both?)gg83 said:Either this is to ward off further stock drops or this is great news. -
thestryker I'm really curious what the followup to LNL will be as there haven't really been any words on it. I'd say maybe they're waiting to see how LNL goes, but realistically they'd have had to already decide to have any parts ready to go. It seems like Intel should be able to do the entire CPU with IFS if they wanted to do a PTL based update.Reply
GNR is Intel catching up with AMD with regards to high performance density using Intel 3 which is a long term node. I think it's fair to say it'll be competitive, but I doubt it will put Intel back in the driver's seat.Reality_checker said:If it is on track, Intel is set to regain leadership in the Data Center segment by the time Granite Rapids comes out.
SRF is Intel's first foray into truly high core count and seems to be their most promising enterprise part. This seems to be the reason for the focus on CWF and I'm not sure there's even a codename in the wild for the next P-core Xeon.
It has been weird to see Intel pushing Xeons later than the desktop parts which has been consistent since Ice Lake. It'll be interesting to see what the next P-core Xeons look like. -
watzupken Booting up is a good sign, but we won’t be able to tell how Intel’s node will fare against TSMC. The problem for Intel is, when they use TSMC almost entirely for their Lunar Lake CPU, they have a high bar to meet when moving back to their own foundry in the succeeding generation. Any weaknesses in Intel foundry will quickly become apparent.Reply