Intel Unveils AI-Focused 18A Process Node Variants and Foundry Roadmap

Intel Introduces 18A Process Node Variants

Intel has unveiled two new variants of its long-awaited 18A process node, aimed at addressing the growing demand for advanced chip manufacturing, particularly in the AI sector. The company announced these developments at its Foundry Direct Connect 2025 event, showcasing its commitment to becoming a major player in the chipmaker-for-hire foundry business 1

1

.

The first variant, 18A-P, is designed for a broad range of applications and promises an 8 percent performance per watt improvement over the base 18A node. Intel's Chief Global Operations Officer, Naga Chandrasekaran, claimed that transitioning to this newer node should be seamless for customers already evaluating designs on 18A 1

1

.

The second variant, 18A-PT, is optimized for designs requiring through silicon vias (TSVs), which allow for communication and power delivery through silicon. This technology enables chips to be assembled by stacking them on top of one another, a crucial feature for complex multi-die semiconductors used in AI applications 1

1

.

Advanced Packaging and AI Focus

Kevin O'Buckley, SVP and GM of Intel Foundry, presented a concept AI accelerator that showcases the potential of the 18A-PT node. This design features a 3D construction with multiple base reticle dies and compute dials stacked on top, surrounded by high-capacity memory including HBM and LPDDR dies. The system is connected by both electrical and optical interfaces, providing tens of terabits of bandwidth 1

1

.

Intel's focus on advanced packaging techniques, including its EMIB and Foveros technologies, positions the company to deliver complex, heterogeneous systems on a package. The company has also announced a collaboration with Amkor to offer its manufacturing tech to a broader customer base 1

1

.

Foundry Roadmap and AI Integration

Intel's 2025 Foundry Direct Connect event outlined a roadmap with a strong emphasis on AI integration across its new nodes. The company's focus is not on developing its own language models, but rather on creating chips optimized for AI technologies in computer processing 2

2

.

The roadmap highlights four pathways divided into different fabrication nodes, with Intel 14A and 18A nodes at the forefront. Both nodes are noted for their focus on edge computing and AI applications. The 18A node, described as an industry-leading backside power delivery node, has been upgraded with ribbon-shaped transistors for enhanced performance 2

2

.

Future Developments and Market Position

Intel's newly appointed CEO, Lip-Bu Tan, emphasized the company's commitment to building trust in its process and packaging technologies while working to reclaim process leadership. Tan highlighted Intel's unique position as the only company conducting advanced, leading-edge semiconductor R&D and manufacturing in the United States 1

1

.

The company expects to see improvements in its foundry unit's financial performance as it stops outsourcing production of its own chips to TSMC when Panther Lake debuts later this year. However, Intel is also planning another round of layoffs, which could affect up to 20 percent of its workforce 1

1

.

As per the roadmap, the 18A nodes are expected to be available in Panther Lake processors between Q3 of this year and early next year, with the 14A node following in 2027 2

2

. The rapid advancements in AI computing suggest that these future chips could have significant capabilities, potentially revolutionizing the AI hardware landscape.