NVIDIA is going to retain its top position as TSMC's biggest customer in 2027, but AMD EPYC Venice could outpace its Vera CPUs.
TSMC Sees Growing 2.5D Advanced Packaging Demand As CPUs Heat Up The Agentic AI Race - NVIDIA Tops The Customer List While AMD EPYC Catches Vera
CPUs are quickly becoming the major consumer of advanced packaging technologies as their demand has grown proportionally with the rise of Agentic AI. So far, 2026 has been a roller-coaster ride for the entire tech segment, and 2027 is shaping up to be just as intense. In 2027, major firms will either introduce new products or ramp up the volume of existing products as AI continues to demand compute capacity.
In a report by Morgan Stanley, it is stated that NVIDIA will retain itself as TSMC's major customer for CoWoS capacity. TSMC is expected to achieve a wafer capacity of 200,000 wafers per month in 2027.
NVIDIA is utilizing TSMC's CoWoS packaging solution for two key products, CoWoS-L for AI GPUs such as Blackwell and Rubin, and CoWoS-R for Vera CPUs. The CoWoS-L capacity is expected to reach around 910,000 units, a 40% increase versus the previous year, and Vera is expected to see a doubling in shipments. This would lead NVIDIA to generate 52% higher revenue versus the prior year.
Nvidia uses TSMC’s CoWoS-L as the single source for all its AI GPU products (e.g. Blackwell and Rubin). Its 2027 CoWoS-L consumption could reach ~910k, up ~40% Y/Y. Strong CoWoS-R bookings by Nvidia also suggest room for AI GPM products (such as doubling). Taken together, we estimate Nvidia’s 2027 forecast for Nvidia’s data center revenue to rise 52% Y/Y.
via Morgan Stanley
But right now, NVIDIA faces some big competition. While its Vera CPUs are already in volume production at TSMC, so is AMD's next-gen EPYC platform, codenamed Venice. Venice is based on the upcoming Zen 6 architecture, which is expected to deliver some solid gains in performance & efficiency. We talk more about those here.
Morgan Stanley has projected that NVIDIA's Vera CPUs are expected to reach 5.75 million units by 2027. That's a big figure for a new CPU launch, & NVIDIA has already said to become the biggest CPU supplier in 2026.
But at the same time, the report expects EPYC Venice CPUs to hit 6.75 million units, 17% more units than NVIDIA's Vera (and 5.4 times the volume versus 2026). AMD also uses an advanced 2nm process node by TSMC, while Vera is based on a 3nm process technology. Vera is designed for Agentic AI, while AMD EPYC Venice tackles both AI and HPC.
Based on our CoWoS consumption forecasts, Nvidia’s 5nm Vera CPU could grow to 5.75mn units in 2027, while AMD’s 2nm Venice CPU may reach 6.75mn units in 2027 vs. ~1.25mn in 2026.
via Morgan Stanley
The challenger that lies ahead is not AMD for NVIDIA or NVIDIA for AMD, it's custom silicon. Many AI firms are now venturing into the custom silicon field. OpenAI, Google, Amazon, & others are in talks or already producing custom chips that will intensify the in-house versus external debate. As custom silicon continues to pick up the pace, NVIDIA, AMD, and other chipmakers might be facing a dire situation. There's still a lot of demand for compute, but AI firms making chips themselves is going to further stress the supply-demand gap.
About the author: A Software Engineer by training and a PC enthusiast by passion, Hassan Mujtaba serves as Wccftech's Senior Editor for hardware section. With years of experience in the industry, he specializes in deep-dive technical analysis of next-generation CPU and GPU architectures, motherboards, and cooling solutions. His work involves not only breaking news on upcoming technologies but also extensive hands-on reviews and benchmarking.
Follow Wccftech on Google to get more of our news coverage in your feeds.
