Nvidia locks in over $4 billion in packaging and testing capacity at Berlin GTC; CPO equipment demand to hit an inflection point in 2027
A BofA Securities report says that at the October 21 Berlin GTC, Nvidia confirmed that its Vera Rubin GPU and CPU have entered full mass production, and made capacity commitments totaling more than $4 billion to ASE/SPIL, KYEC, and Amkor. Of this, KYEC received a $1.1 billion prepayment for a new test facility in the United States, while Amkor secured a $3 billion agreement to expand its South Korean interposer capacity by more than 300% by the end of 2028. The report argues that the AI infrastructure bottleneck has shifted from chip design to packaging and testing, that demand for CPO-related equipment is expected to grow tenfold in 2027, and that the adoption inflection point is expected in the fourth quarter of 2027.
Why read this
AI score 60
The BofA report breaks down Nvidia's packaging/testing and CPO capacity commitments secured at Berlin GTC, shedding light on the bottlenecks and capital flows in the AI back-end supply chain.
Topics
Body · AI translation
A series of supply chain signals released by Nvidia at the GTC conference in Berlin is reshaping the landscape of the Asia-Pacific semiconductor back-end industry. BofA Securities believes that the core bottleneck in AI infrastructure construction has shifted from chip design to packaging and testing capacity, while the accelerated deployment of co-packaged optics (CPO) technology has opened up a new growth track.
According to Zhuifeng Trading Desk, a research report released by BofA Securities on October 22 shows that Nvidia confirmed on October 21 at the GTC conference in Berlin that Vera Rubin GPUs and CPUs have entered full mass production, and announced deepened cooperation with several key outsourced semiconductor assembly and test (OSAT) vendors, covering multiple segments including advanced packaging, testing, CPO and substrates. Among them, Nvidia made major capacity commitments to ASE/SPIL, KYEC and Amkor, involving a total funding scale of more than US$4 billion.
The above cooperation arrangements point directly to the most prominent contradiction in the current AI supply chain—insufficient back-end capacity. BofA Securities noted that Nvidia is accelerating the provision of consigned equipment to OSAT vendors to advance capacity ramp-up, while also accelerating CPO deployment, with related equipment demand expected to see inflection-point growth in 2027.
Nvidia locks in back-end capacity with large-scale commitments
The cooperation details disclosed by Nvidia at the Berlin GTC are considerable in scale. According to the BofA Securities report, Nvidia reached an agreement with ASE/SPIL aimed at increasing optical engine (OE) capacity 15-fold by mid-2027 and doubling it again by the end of 2028; signed an advance payment agreement totaling US$1.1 billion with KYEC to support its new test facility in the United States (expected to start production in April 2027) and expand the scale of its Singapore testing business (expected to begin mass production in February 2027); and signed a US$3 billion agreement with Amkor aimed at increasing South Korea interposer capacity by more than 300% by the end of 2028.
BofA Securities believes the core logic of the above arrangements lies in "customer-funded capacity expansion protection"—Nvidia, through advance payments and equipment consignment, transfers the risk of capacity expansion to itself, thereby accelerating the ramp-up pace of OSAT vendors. The report estimates that if ASE/SPIL's optical engine capacity expansion proceeds smoothly, this business alone could contribute about US$1 billion in revenue to ASE in 2028.
Testing segment: wafer probing and system-level testing become new focus
In the testing field, BofA Securities noted that Nvidia's goal is to double testing capacity, but incremental opportunities are mainly concentrated in end wafer probing and system-level testing (SLT), rather than final testing—the latter's capacity is already ample for mass production of Vera Rubin and Mellanox network switch chips.
The key factor driving the rapid growth of testing demand lies in the substantial extension of test time per chip, rather than a simple increase in the number of tests. The report takes a cautious attitude toward KYEC's US investment, noting that its new US facility will initially add only about 100 testers, and its Singapore business scale is also relatively limited, mainly serving HPC customers seeking to reduce geopolitical risk.
CPO acceleration: equipment demand expected to reach inflection point in 2027
Co-packaged optics (CPO) is another important theme of this GTC. BofA Securities stated that Nvidia's plan to accelerate CPO deployment is highly consistent with supply chain feedback—related equipment demand is expected to grow 10-fold in 2027, with the adoption inflection point expected to appear in the fourth quarter of 2027.
Nvidia also announced that it will diversify the packaging source of Vera CPUs from Amkor/TSMC to ASE, and extend ASE/SPIL's role from GPU substrate assembly to CPO processes. BofA Securities believes that this strategic adjustment not only expands ASE's business boundaries, but also creates new market space for tool vendors with CPO equipment supply capabilities. In addition to CPO, equipment demand related to 2.5D packaging (such as CoWoS, FoCoS, EMIB-T) and 3D packaging (such as SoIC, WoW) is also benefiting from capacity expansion demand from multiple ASIC/GPU/CPU/network switch chip customers.
~~~~~~~~~~~~~~~~~~~~~~~~
The above exciting content comes from Zhuifeng Trading Desk.
For more detailed interpretation, including real-time interpretation and frontline research, please join [Zhuifeng Trading Desk▪Annual Membership] Risk Disclosure and Disclaimer
Markets carry risks, and investment requires caution. This article does not constitute personal investment advice, nor does it take into account the special investment objectives, financial conditions or needs of individual users. Users should consider whether any opinions, views or conclusions in this article are suitable for their specific circumstances. Investment based on this is at the user's own risk.