TSMC To Dominate In Next-Gen CPU Offerings, Including Intel’s Nova Lake & AMD’s EPYC Venice
The semiconductor industry is once again turning its attention to Taiwan Semiconductor Manufacturing Company (TSMC) as demand for its 2nm (N2) process node continues to grow. Recent reports suggest that not only are mobile and high-performance computing (HPC) customers rushing toward TSMC’s N2 process, but now both Intel and AMD are expected to integrate it into their next-generation CPU lineups, further solidifying the foundry’s central role in the global chip supply chain.
According to a Morgan Stanley note shared by @QQ_Timmy, AMD and Intel are preparing to adopt TSMC’s 2nm node for their future products, namely AMD’s EPYC Venice data center CPUs and Intel’s Nova Lake client CPUs. This development underscores the growing dependence of every mainstream segment of the tech industry on TSMC’s cutting-edge manufacturing capabilities.
AMD and Intel could join the N2 customer list in 1H26, we believe. As AMD announced
— 駿HaYaO (@QQ_Timmy) October 1, 2025
officially, its next-generation AMD EPYC processor, code-named “Venice,” has already been
taped out and will be brought up on TSMC advanced 2nm process technology. According
to our industry… pic.twitter.com/2LwmXidztU
AMD has long been a close partner of TSMC, and it comes as no surprise that the company is positioning itself as one of the first customers of the N2 process. CEO Lisa Su already confirmed AMD’s adoption of the 2nm node earlier this year, with the upcoming EPYC Venice CPUs set to leverage the advanced technology for the next wave of data center computing. Given AMD’s success with EPYC Genoa and Turin architectures, Venice is expected to continue the trajectory of performance-per-watt leadership with a significant jump in efficiency and density gains brought by N2.
Perhaps more surprising is Intel’s adoption of TSMC’s 2nm process for its upcoming Nova Lake architecture. According to Morgan Stanley, Intel will outsource the compute tile of Nova Lake to TSMC, marking a pivotal moment in Intel’s long-stated “internal-first” manufacturing strategy.
The report suggests that Intel’s decision is largely due to subpar yield rates of its own 18A process node, which has limited its ability to fully rely on internal fabs for such an important product family. While Intel has publicly maintained confidence in its 18A node, the move toward TSMC for Nova Lake shows that customer satisfaction and timely product delivery remain higher priorities.
Intel’s strategy of balancing between Intel Foundry Services (IFS) and external foundries like TSMC is not new, but using TSMC’s most advanced node for Nova Lake highlights the industry’s skepticism about Intel’s ability to meet aggressive timelines internally.
The semiconductor community is closely watching Intel’s Panther Lake (PTL) lineup, which will be the true test of how far Intel’s 18A node has progressed. Panther Lake is expected to debut later this quarter, with mass production scheduled for Q1 2026, serving as an early indicator of whether Intel’s foundry business can regain ground against TSMC.
Meanwhile, TSMC continues to expand its N2 capacity, preparing to supply not only AMD and Intel but also a broad range of partners across the mobile, AI, and HPC markets. If adoption scales as analysts predict, TSMC could maintain a dominant lead into the late 2020s, further strengthening its status as the backbone of global semiconductor manufacturing.
With AMD and Intel both leaning on TSMC’s 2nm node, do you think Intel can still establish its foundry ambitions long-term, or will reliance on TSMC become the industry’s new normal?
