Samsung is making strategic moves in the semiconductor market as it vies for NVIDIA’s attention with its advanced 3nm node technology. In an effort to secure orders for NVIDIA’s next-generation GPUs, Samsung is doubling down on the development and readiness of its 3nm fabrication processes.
Samsung’s Game Plan for Attracting NVIDIA
Samsung Foundry is keenly aware of the competition it faces from TSMC, which is currently the provider of choice for NVIDIA’s GPU production, utilizing TSMC’s 5nm process technology. However, Samsung is not backing down. With sights set on partnership with the prominent chipmaker for its upcoming graphics card families, Samsung has initiated significant internal measures to meet NVIDIA’s stringent requirements.
The company’s plan, internally named “Nemo”, alludes to the directed efforts involving multiple departments to fulfill NVIDIA’s high standards. Samsung’s last contribution to NVIDIA’s product line was the 8nm production of the GeForce RTX 30 series GPUs, which targeted the consumer gaming market. NVIDIA later shifted to TSMC for the Ada Lovelace GeForce RTX 40 series, utilizing a 5nm process.
Ramping Up for 3nm Production
The potential of Samsung’s 3nm process, which employs Gate-All-Around (GAA) architecture, is crucial for Samsung’s ambition to position itself as NVIDIA’s manufacturing partner. This GAA technology is expected to address some limitations inherent in older FinFET manufacturing processes. With mass production targeted for the first half of 2024, Samsung is eagerly refining its technology to meet the requirements and expectations NVIDIA may have for this partnership.
Overcoming Challenges
Despite facing challenges like failing the HBM3E memory qualification process with NVIDIA, Samsung is not deterred. It is proactively working on improving its high-bandwidth memory (HBM) offerings. Looking further ahead, Samsung is also planning the development of its next-generation HBM4 memory architecture, with an anticipated debut slated between 2025 and 2026.
The Future of NVIDIA and Samsung’s Potential Partnership
It remains to be seen whether NVIDIA will choose Samsung as its Foundry partner for future GPU endeavors. In the interim, it is possible that semi-custom products could emerge before NVIDIA fully commits to Samsung’s processes.
Given NVIDIA’s continued demand for innovative and powerful GPUs, it is likely that the chipmaker will pursue a strategy of dual sourcing its chips from multiple semiconductor fabs to diversify and ensure supply chain resilience, much as it has done with other components in the past.
As the tech world watches, only time will tell if Samsung’s strategic push and advancements in 3nm GAA technology will be sufficient to persuade NVIDIA into forging a new partnership for the production of cutting-edge GPUs.
