• AiNews.com
  • Posts
  • Samsung Unveils Turnkey Foundry Service Advanced Nodes for AI Chips

Samsung Unveils Turnkey Foundry Service Advanced Nodes for AI Chips

Author

Alicia Shapiro
June 13, 2024 • Estimated Reading Time: 5 minutes

Futuristic industrial scene showcasing Samsung's advanced AI chip manufacturing process. At the center, a large microchip labeled 'AI' sits on a platform with glowing blue circuits, representing cutting-edge AI technology. The chip is surrounded by two smaller labeled chips, '2nm S SF4Z' and '4nm S SF4U,' indicating advanced node technology for chip production. Robotic arms are positioned on either side, assembling and managing the chips, emphasizing automation in the chip-making process. In the background, data servers, processing units, and high-tech machinery fill the scene, all interconnected by glowing circuits and lights, symbolizing a vast, efficient production network. The Samsung logo is prominently displayed in the upper left corner, signifying the company's role in developing advanced AI chips and turnkey foundry services. The overall setting suggests a highly advanced, automated facility designed for AI chip manufacturing and innovation

Image Source: ChatGPT-4o

Samsung Unveils Turnkey Foundry Service Advanced Nodes for AI Chips

Samsung’s foundry business unit, traditionally focused on manufacturing chips for smartphones, is now shifting its focus to AI and high-performance computing (HPC). On Wednesday, Samsung introduced an updated roadmap for its most advanced chip nodes and unveiled a new turnkey service aimed at attracting companies like Nvidia and AMD to use its foundry for their AI chips.

Advanced Chip Nodes for AI and HPC

At the Samsung Foundry Forum, held in San Jose with the theme "Empowering the AI Revolution," Samsung showcased its new 2-nanometer (nm) and 4nm process nodes, dubbed SF2Z and SF4U, respectively. These nodes are designed to cater to the growing demands of AI and HPC applications.

SF2Z Node

  • Backside Power Delivery Network (BSPDN): The SF2Z node incorporates BSPDN technology, which places power rails on the backside of the wafer, unlike current chips where all components are on the front side.

  • Enhanced PPA and Voltage Drop: This innovation improves power, performance, and area (PPA) metrics and reduces voltage drop, making it ideal for AI and HPC chips.

  • Launch Timeline: SF2Z is set to roll out in 2027, while its predecessor, SF2, will launch in 2025.

SF4U Node

  • Optical Shrink Technology: SF4U offers PPA boosts by scaling down existing die designs to fit the newer node without requiring major architectural changes.

  • Cost Efficiency: This process saves money and simplifies the migration to more advanced nodes.

  • Launch Timeline: SF4U will be available in 2025, building on the already available SF4 process.

Expansion of Samsung Foundry's Focus

Samsung Foundry's AI-related sales have surged by 80% over the past year. The company aims to diversify its revenue streams, targeting over 50% of its foundry revenue from non-mobile sectors. This shift is part of a broader strategy to meet evolving market demands and expand its customer base.

New Turnkey Foundry Service: Samsung AI Solutions

Samsung also introduced a new turnkey foundry platform called Samsung AI Solutions. This platform integrates the strengths of Samsung's Foundry, Memory, and Advanced Package (AVP) business units to provide comprehensive solutions tailored to the specific needs of AI chips.

Key Features of Samsung AI Solutions

  • Bandwidth and Power Efficiency: The platform offers increased bandwidth in a compact form factor, reduces power consumption, and enhances signal integrity.

  • Integrated Supply Chain: Samsung's unique position as the only chipmaker providing memory chips, manufacturing, and packaging services streamlines supply chain management and reduces time to market.

  • Turnaround Time Improvement: Samsung AI Solutions promises a 20% improvement in total turnaround time for customers.

  • Future Integration: The platform will feature co-packaged optics (CPO) in 2027.

Industry Context and Competitiveness

Samsung's move to focus on AI and HPC chips aligns with broader industry trends. The introduction of BSPDN technology and advanced nodes puts Samsung in direct competition with other major contract chipmakers like Intel and TSMC, who are also developing similar technologies for their 2nm or smaller nodes.

Conclusion

Samsung's unveiling of advanced chip nodes and a new turnkey foundry service marks a significant shift in its foundry business strategy. By focusing on AI and HPC, Samsung aims to capitalize on the growing demand for advanced semiconductor technologies. The company's innovative approaches, such as BSPDN and optical shrink technology, promise to enhance chip performance and efficiency, positioning Samsung as a leader in the AI revolution.

With these advancements, Samsung is not only diversifying its revenue streams but also solidifying its role as a key player in the semiconductor industry, offering comprehensive and integrated solutions to meet the complex needs of AI and HPC applications.