Samsung arrives with new 2nm & 4nm AI chip tech as market competition heats up

Samsung announced during this week’s event at the company’s HQ in San Jose, California, that it is introducing two new advanced process nodes, SF2Z (2nm) and SF4U (4nm).

The new 2nm process node, according to Samsung, uses a backside power delivery network (BSPDN), a tech that improves power, performance, and area (PPA) and reduces voltage drop. The 4nm, billed as SF4U, offers PPA improvements through optical shrink, allowing existing die designs to be scaled down without major architectural changes.

These new advanced process nodes are slated for mass production in 2025 (4nm) and 2027 (2nm), seemingly focusing on HPC and AI chips instead of mobile processors.

Besides, the South Korean tech giant also launched its integrated Samsung AI Solutions platform to enhance its semiconductor offerings for AI apps, potentially attracting AI chip clients like Nvidia and AMD.

Its foundry business unit, which experienced an 80% increase in sales over the past year thanks to continuous innovation, explained that this platform combines the strengths of its Foundry, Memory, and Advanced Package (AVP) units. This makes it possible to create high-performance, low-power, and high-speed solutions specifically designed for AI chips.

Samsung AI Solutions also simplifies supply chain management, speeds up time to market by 20%, and provides a complete service to help customers advance in AI technology.

As the AI race develops, the market competition for AI chips heats up even more. Chip makers like Qualcomm and AMD, for example, have Snapdragon X Elite and Ryzen AI 300 series, not to mention Intel with its new Lunar Lake that can outdo the previous Meteor Lake.