AMD Achieves First TSMC N2 Product Silicon Milestone

[Daniel Nenni]

— Next-generation AMD EPYC CPU, codenamed “Venice,” is the first HPC product to be brought up on TSMC’s next-generation N2 node —​

SANTA CLARA, Calif., April 14, 2025 (GLOBE NEWSWIRE) -- AMD (NASDAQ: AMD) today announced its next-generation AMD EPYC™ processor, codenamed “Venice,” is the first HPC product in the industry to be taped out and brought up on the TSMC advanced 2nm (N2) process technology. This highlights the strength of AMD and TSMC semiconductor manufacturing partnership to co-optimize new design architectures with leading-edge process technology. It also marks a major step forward in the execution of the AMD data center CPU roadmap, with “Venice” on track to launch next year. AMD also announced the successful bring up and validation of its 5th Gen AMD EPYC™ CPU products at TSMC’s new fabrication facility in Arizona, underscoring its commitment to U.S. manufacturing.

“TSMC has been a key partner for many years and our deep collaboration with their R&D and manufacturing teams has enabled AMD to consistently deliver leadership products that push the limits of high-performance computing,” said Dr. Lisa Su, chair and CEO, AMD. “Being a lead HPC customer for TSMC’s N2 process and for TSMC Arizona Fab 21 are great examples of how we are working closely together to drive innovation and deliver the advanced technologies that will power the future of computing.”

“We are proud to have AMD be a lead HPC customer for our advanced 2nm (N2) process technology and TSMC Arizona fab,” said TSMC Chairman and CEO Dr. C.C. Wei. “By working together, we are driving significant technology scaling resulting in better performance, power efficiency and yields for high-performance silicon. We look forward to continuing to work closely with AMD to enable the next era of computing.”

Supporting Resources

• Learn more about AMD EPYC Processors

About AMD
For more than 50 years AMD has driven innovation in high-performance computing, graphics and visualization technologies. Billions of people, leading Fortune 500 businesses and cutting-edge scientific research institutions around the world rely on AMD technology daily to improve how they live, work and play. AMD employees are focused on building leadership high-performance and adaptive products that push the boundaries of what is possible. For more information about how AMD is enabling today and inspiring tomorrow, visit the AMD (NASDAQ: AMD) website, blog, LinkedIn and X pages.

Contact:
Aaron Grabein
AMD Communications
(737) 256-9518
aaron.grabein@amd.com

Liz Stine
AMD Investor Relations
+1 720-652-3965
liz.stine@amd.com

https://www.amd.com/en/newsroom/press-releases/2025-4-14-amd-achieves-first-tsmc-n2-product-silicon-milesto.html

 

[Daniel Nenni]

Expect more TSMC N2 tape-out news in the TSMC Tech Symposium run up. Exciting times!

 

[fansink]

Expect more TSMC N2 tape-out news in the TSMC Tech Symposium run up. Exciting times!

This adds a Very Large bargaining chip to CC WEI's US tariffs toolbox.

 

[FIvr]

I am having a hard time trying to make out any semblance to chiplets in that circle.

 

[nghanayem]

I'm surprised. AMD was always slower moving to new TSMC nodes. Launching N2 products in 2026 alongside first Apple products is a big shift. And a welcome one, because I feel like their old strategy of not using TSMC's best was a disadvantaged strategy.

 

[Artificer60]

I'm surprised. AMD was always slower moving to new TSMC nodes. Launching N2 products in 2026 alongside first Apple products is a big shift. And a welcome one, because I feel like their old strategy of not using TSMC's best was a disadvantaged strategy.

Yeah, Intel's design group is going to have to bring their A game if they intend to beat AMD without a process advantage.

 

[siliconbruh999]

Yeah, Intel's design group is going to have to bring their A game if they intend to beat AMD without a process advantage.

Agreed it's now 18A Diamond Rapids vs N2 Zen6 it will mostly come to how good your design is now

 

[Markwrob]

I'm surprised. AMD was always slower moving to new TSMC nodes. Launching N2 products in 2026 alongside first Apple products is a big shift. And a welcome one, because I feel like their old strategy of not using TSMC's best was a disadvantaged strategy.

I seem to recall reading some articles that suggested or outright stated that for the prior several years AMD was basically priced out of early adoption of the smallest nodes at TSMC. Apple and/or Nvidia paid more to be first than AMD could or would offer. And this was impacted some by Intel's purchases at TSMC. Does anyone else recall this?

 

[osnium]

I seem to recall reading some articles that suggested or outright stated that for the prior several years AMD was basically priced out of early adoption of the smallest nodes at TSMC. Apple and/or Nvidia paid more to be first than AMD could or would offer. And this was impacted some by Intel's purchases at TSMC. Does anyone else recall this?

If I were TSMC, I would prioritize giving leading edge access to AMD in order to help bleed Intel dry for daring to compete with them on the foundry side.

 

[nghanayem]

If I were TSMC, I would prioritize giving leading edge access to AMD in order to help bleed Intel dry for daring to compete with them on the foundry side.

Bad business call. If I was TSMC I would be doing everything possible to be fighting tooth and nail to win every Intel design I could. Even something as simple as getting the U series PCH is more wafers than all of AMD. Think about it like this, Intel products even after having their sales collapse can fill at least 2 18A fabs single handily. That is Apple levels of volume right there. And if Intel products ever do bounce back and fill 4-5 modern fabs at peak ramp for a given node TSMC's TAM on a given node would literally double (since 4-6 fabs is what recent TSMC peak ramp seem to look like). Think about a hypothetical world where all the intel 7 stuff was on N5 and Panther lake 18A stuff and meteor lake i4/3 compute was on N3. TSMC with DOUBLE the N5 and maybe 50% more N3 revenue. Transplanting that to last quarter, TSMC's corporate wide net earnings would be 47% higher! That is money that CANNOT be ignored just because Intel is entering the foundry game. No matter what discount TSMC has to give, winning as many Intel accounts as possible should be top priority. The only problem is how do you fill so much extra capacity after Intel products finishes the last of their production runs on a given node.

I seem to recall reading some articles that suggested or outright stated that for the prior several years AMD was basically priced out of early adoption of the smallest nodes at TSMC. Apple and/or Nvidia paid more to be first than AMD could or would offer. And this was impacted some by Intel's purchases at TSMC. Does anyone else recall this?

NVIDIA has never paid more to be first because they have never been first. They wait for a while because they have BIG die sizes and need to wait for DD to mature to the point for NVIDIA to be comfortable with the yields they would get. First NVIDIA N7 chip was 2020, and the first AMD N7 was 2019. First NVIDIA N5 chip was in 2022, first AMD N5 chip was also 2022. First NVIDIA N3 chip is in 2026, AMD's first N3 chip was 2024. As for getting "priced out by Apple". That isn't how it works. If TSMC got prepays to build out x wafers/month by this date they would do it. There isn't this fixed wafer capacity that TSMC rents out to the highest bidder that people have in their minds for whatever reason. Or I guess I should say not for someone with the pocketbooks of AMD. Now if they want to go above their wafer agreement with TSMC, then yeah they have to bid against others for what little surpluses capacity TSMC builds out. But be it waiting for higher yields, better performance, waiting for the wafer prices to come down a bit, or some other reason; AMD wasn't telling TSMC to build out their reserved capacity early. With N7 and N5 they at least had the excuse that the HPC versions weren't ready for years after first Apple ramp. But I wouldn't be shocked if N7/5 HPC came "later" just because there was no demand for it to be any sooner. After all N3 HPC and N3E HPC were ready with the LP version of the processes out of the gate. And seemingly N2 HPC is also ready out of the gate.

Yeah, Intel's design group is going to have to bring their A game if they intend to beat AMD without a process advantage.

Agreed it's now 18A Diamond Rapids vs N2 Zen6 it will mostly come to how good your design is now

Considering how Intel has been losing in Xeon even when they have had a process lead (i14nm vs GF12LP, i7 vs N7P, i3 vs N4P, and how even client lost on N3B vs AMD N4P) I can't say I have tons of reasons to be super optimistic for Intel product's competitive ability when competing from a slight process disadvantage. If I had to say something nice. I guess we will have to see how the Raja reforms to chip designs have been going (some positive signs with LNL) and how the various Pat/Navid/Schlomet era reforms turn out over the next 5 years. Just a shame that Intel products can take Intel foundry down with them until Intel foundry can diversify to the point that they don't need Intel product's business just to keep the lights on. Today, that ~$15B in external yearly revenue by 2030 seems like a good place to be for Intel foundry to be able to ride out whatever trials and tribulations intel products might run into. But that is assuming there isn't some step cost increase to leading edge fab or R&D costs like there was in the 2010s with the rapid proliferation of multipatterning.

 

[lefty]

I seem to recall reading some articles that suggested or outright stated that for the prior several years AMD was basically priced out of early adoption of the smallest nodes at TSMC. Apple and/or Nvidia paid more to be first than AMD could or would offer. And this was impacted some by Intel's purchases at TSMC. Does anyone else recall this?

That's nothing to do with it. AMD is first customer for the node this time because of timing. N2 volume production will start in 2H 2025, that's too late for the iPhone 17, and a bit early for the iPhone 18 - so there was a gap where a non-Apple customer could sneak in (otherwise it would have been Apple as usually).

 

[nghanayem]

That's nothing to do with it. AMD is first customer for the node this time because of timing. N2 volume production will start in 2H 2025, that's too late for the iPhone 17, and a bit early for the iPhone 18 - so there was a gap where a non-Apple customer could sneak in (otherwise it would have been Apple as usually).

Not buying it. Smart money is on iPhone 18 being the first product. With cycle time being what it is N2 HVM start is perfectly lined up for IPhone 18 ramp. N3 HVM started in December 22 and N3 chips showed up for the 15 pro in September 23. And CC has reiterated time and time again that N2 is drastically more complex than N3E. This is shown in how TSMC says first wafers will roll off the lines in late Q1'26 or early Q2'26 depending on when in Q4'25 they start HVM. Now in typical AMD fashion I'm sure they will announce their N2 chip early summer next year. But actual qualified chips powering public cloud instances or retail availability of servers won't happen until late 26 or early 27 (just like the past 2.5 generations of Epyc processors).

 

[Daniel Nenni]

I'm surprised. AMD was always slower moving to new TSMC nodes. Launching N2 products in 2026 alongside first Apple products is a big shift. And a welcome one, because I feel like their old strategy of not using TSMC's best was a disadvantaged strategy.

AMD and TSMC really feel the heat of Intel. Nobody is in nobody's rear view mirror. The semiconductor industry needs competition to innovate and disrupt. I do think that AMD and TSMC will continue to lead but I am rooting for intel, absolutely.

Apple's volumes are very high and they ship systems not just chips so comparing Apple to AMD or Intel is difficult. The Apple SoCs are also incremental, they are not completely new designs. AMD and Nvidia do more new design work than Apple thus it takes more time. Multi die may help that but multi die brings new challenges. Apple is a fierce competitor, they are a systems company with the best SoC, absolutely. Apple is also responsible for 25% of TSMC's revenue, ain't nobody gonna beat that.

 

[Artificer60]

Considering how Intel has been losing in Xeon even when they have had a process lead (i14nm vs GF12LP, i7 vs N7P, i3 vs N4P, and how even client lost on N3B vs AMD N4P) I can't say I have tons of reasons to be super optimistic for Intel product's competitive ability when competing from a slight process disadvantage. If I had to say something nice. I guess we will have to see how the Raja reforms to chip designs have been going (some positive signs with LNL) and how the various Pat/Navid/Schlomet era reforms turn out over the next 5 years. Just a shame that Intel products can take Intel foundry down with them until Intel foundry can diversify to the point that they don't need Intel product's business just to keep the lights on. Today, that ~$15B in external yearly revenue by 2030 seems like a good place to be for Intel foundry to be able to ride out whatever trials and tribulations intel products might run into. But that is assuming there isn't some step cost increase to leading edge fab or R&D costs like there was in the 2010s with the rapid proliferation of multipatterning.

I really think that Intel's best hope on the design side is to pull another Netburst --> Core transition out of their hat. Your point about being tied to what is already in the pipeline is a good one, but if Lip-bu Tan is to be believed then he is looking for ways to disrupt the design cycle. Core has been a good architecture, but it is getting a bit long in the tooth and it is time to move on.

I also have to believe that listening to your customers and designing what they need would go a long way towards addressing some of Intel's gaps as well. This is another area Lip-bu Tan promises to address.

But I'm not a design guy, so maybe I'm just being completely delusional here.

 

[Daniel Nenni]

AMD advances US chip strategy with TSMC, joining Nvidia in massive domestic shift​

In a visit to Taiwan on April 15, 2025, AMD CEO Lisa Su disclosed that the company is partnering with TSMC to fabricate chips in Arizona, a development that came after a similar plan by Nvidia to produce AI infrastructure in the US with manufacturing...

AMD advances US chip strategy with TSMC, joining Nvidia in massive domestic shift

In a visit to Taiwan on April 15, 2025, AMD CEO Lisa Su disclosed that the company is partnering with TSMC to fabricate chips in Arizona, a development that came after a similar plan by Nvidia to produce AI infrastructure in the US with manufacturing partners, including TSMC and Foxconn.

www.digitimes.com

 

[Daniel Nenni]

I seem to recall reading some articles that suggested or outright stated that for the prior several years AMD was basically priced out of early adoption of the smallest nodes at TSMC. Apple and/or Nvidia paid more to be first than AMD could or would offer. And this was impacted some by Intel's purchases at TSMC. Does anyone else recall this?

Apple has a most favored nation contract with TSMC. Nobody gets silicon for cheaper or faster than Apple. That's what you get for being TSMC largest customer.

Does anybody really think TSMC would be in the position they are today without Apple? Apple signing on with TSMC was a huge disruption that brought TSMC to the leading edge. Look at the graph. Apple came to TSMC at 20nm. More process improvements in less time, right? And notice they do not list 6nm and 4nm. 2nm will be listed of course.

 

[fansink]

AMD and TSMC really feel the heat of Intel. Nobody is in nobody's rear view mirror. The semiconductor industry needs competition to innovate and disrupt. I do think that AMD and TSMC will continue to lead but I am rooting for intel, absolutely.

Apple's volumes are very high and they ship systems not just chips so comparing Apple to AMD or Intel is difficult. The Apple SoCs are also incremental, they are not completely new designs. AMD and Nvidia do more new design work than Apple thus it takes more time. Multi die may help that but multi die brings new challenges. Apple is a fierce competitor, they are a systems company with the best SoC, absolutely. Apple is also responsible for 25% of TSMC's revenue, ain't nobody gonna beat that.

In FY25, TSMC will very likely double Intel's revenue, after surpassing their revenue in 2Q22 (3 years ago).

 

[MKWVentures]

AMD and TSMC really feel the heat of Intel. Nobody is in nobody's rear view mirror. The semiconductor industry needs competition to innovate and disrupt. I do think that AMD and TSMC will continue to lead but I am rooting for intel, absolutely.

Apple's volumes are very high and they ship systems not just chips so comparing Apple to AMD or Intel is difficult. The Apple SoCs are also incremental, they are not completely new designs. AMD and Nvidia do more new design work than Apple thus it takes more time. Multi die may help that but multi die brings new challenges. Apple is a fierce competitor, they are a systems company with the best SoC, absolutely. Apple is also responsible for 25% of TSMC's revenue, ain't nobody gonna beat that.

Is it really a competition between Intel and TSMC? Intel pushed TSMC to deliver faster.... and they did. how do you think N2 wafer volume will compare to Intels external 18A volume in 2026? @Daniel Nenni: when does a design on 18A need to start in order to be high volume in 2027? is it too late?

 

[MKWVentures]

In FY25, TSMC will very likely double Intel's revenue, after surpassing their revenue in 2Q22 (3 years ago).

Intel is one of TSMCs best customers. TSMC is Perhaps Intel's most important supplier. someday this may change. today is not that day

 

[siliconbruh999]

Intel is one of TSMCs best customers. TSMC is Perhaps Intel's most important supplier. someday this may change. today is not that day

As of rn Yes due to the N3B/E contracts but over time it's due to reduced I don't think anyone would like to keep wafers external while having internal wafer shop and loosing money on Internal wafer shop.