Samsung Electronics, a global leader in advanced semiconductor technology, today unveiled its latest foundry innovations and outlined its vision for the age of artificial intelligence at the Samsung Foundry Forum (SFF) in the US. Annual event held at American Equipment Solutions headquarters in San Jose.

Under the theme "Powering the AI Revolution", Samsung announced its advanced process technology roadmap, including two new breakthrough nodes - the SF2Z and SF4U - as well as unique solutions that leverage its foundry, memory and advanced technology AI solutions platform. dr. Siyoung Choi, president and foundry business manager of Samsung Electronics, said: "In an era where many technologies are continuously evolving in relation to artificial intelligence, the key to realizing artificial intelligence lies in high-performance, low-power semiconductors." In addition to our optimization for AI chips, in addition to our proven GAA process, we plan to introduce integrated co-packaged optics (CPO) technology for high-speed, low-power computing, providing our customers with one-stop AI solutions to thrive in this transformational era.

The event featured prominent industry leaders, including Arm CEO Rene Haas and Groq CEO Jonathan Ross, who took to the stage to highlight their work with Samsung to address the new challenges of the AI Strong partnership. Approximately 30 partner companies exhibited at the booth, further highlighting the dynamic collaboration across the US foundry ecosystem.

Empowering Customer AI Solutions With State-of-the-Art Process Technology Roadmap

Samsung announced two new process nodes, SF2Z and SF4U, strengthening its flagship process technology roadmap.

The company's latest 2nm process, SF2Z, includes optimized backside power delivery network (BSPDN) technology, which places power rails on the back of the wafer to eliminate congestion between power and signal lines. The use of BSPDN technology in the SF2Z not only improves the power consumption, performance and area (PPA) compared to the first generation 2nm node SF2, but also significantly reduces the voltage drop (IR drop), improving the performance of HPC designs. The SF2Z is expected to enter mass production in 2027. SF4U, on the other hand, is a high-end 4nm variant that improves on PPA by incorporating optical shrink technology and is slated for mass production in 2025.

Samsung reiterated that its preparations for SF1.4 (1.4nm) are progressing smoothly, with performance and yield targets expected to reach mass production by 2027. Underscoring its ongoing commitment to Beyond Moore, Samsung is actively building future sub-1.4nm process technologies through material and structural innovation.

Continuously Advancing GAA Maturity

With the advent of the artificial intelligence era, structural advances such as gate-all-around (GAA) have become necessary to meet power consumption and performance requirements. At SFF, Samsung highlighted the maturity of its GAA technology as a key enabling technology to improve artificial intelligence capabilities. Entering its third year of mass production, Samsung's GAA process has consistently demonstrated continued maturity in both yield and performance. Using this accumulated GAA manufacturing experience, Samsung plans to mass-produce its second-generation 3nm process (SF3) in the second half of this year and deliver GAA on the upcoming 2nm process. Since 2022, Samsung GAA output has been constantly increasing, and in the coming years it is going to expand essentially.

Highlighting Cross-Company Collaborations for Turnkey Samsung AI Solutions

Another highlight was the introduction of Samsung AI Solutions, a turnkey artificial intelligence platform created through the collaboration of the company's foundry, memory and AVP business units.

Samsung combines the unique strengths of each company to deliver high-performance, low-power, high-bandwidth solutions that can be tailored to customers' specific AI needs. Cross-company collaboration also simplifies supply chain management (SCM) and improves time-to-market, resulting in a significant 20% reduction in total turnaround time (TAT). Samsung plans to launch an integrated, CPO-integrated AI solution in 2027, aiming to provide customers with one-stop AI solutions.

Diversifying Customers and Applications for a Balanced Portfolio Across AI to Mainstream Tech

Samsung has also made significant progress in diversifying its customer base and application areas.Close cooperation with customers has led to an 80% increase in AI sales at Samsung foundries in the past year, reflecting their commitment to meet the changing needs of the market. In addition to leading process assemblies, Samsung also offers specialty wafers and 8-inch wafer derivatives with continuous PPA improvements and strong cost competitiveness. With this balanced portfolio of technologies, the company addresses customer needs in automotive, medical, wearable and IoT applications.

Converging AI and Technology to Advance the Foundry Ecosystem Toward Collective Growth

Building on the momentum generated by the SFF event, Samsung will also hold its annual Samsung Advanced Foundry Ecosystem (SAFE™) Forum on June 13. With the theme "AI: Exploring Opportunities and Futures," the forum will serve as a collaborative platform for ecosystem partners to discuss customized technologies and solutions tailored to AI. At this year's forum, industry leaders including Siemens CEO Mike Ellow, AMD VP Bill Enns and Celestial AI CEO David Lazovsky will provide insights into the future of chip and system design technologies.

Following the launch of the MDI Alliance last year, the forum will also host the first Multi-Die Integration (MDI) Alliance workshop. Samsung will hold extensive discussions with its alliance partners on opportunities for mutual growth and specific cooperation measures, paying special attention to the development of integrated solutions for 2.5D and 3D IC projects. These events will further strengthen partnerships and promote a collective vision.