Intel’s foundry business has secured major design wins with AMD, NVIDIA, and OpenAI on its 18A and 14A process nodes while achieving high yields with EMIB packaging technology.
Intel’s foundry services are attracting heavy industry attention. The company has reportedly won design contracts from AMD, NVIDIA, Marvell, Microsoft, Micron, and OpenAI. These customers are reportedly turning to Intel as an alternative to TSMC, which faces supply shortages that make it difficult to meet everyone’s demand for advanced manufacturing.
TSMC has long been the dominant manufacturer for cutting-edge chips, but Intel has quickly emerged as a strong competitor. The company has only fully launched its 18A node, but that technology is already in volume production. Intel is also working toward more advanced versions, including 18A-P and the 14A node, which are set for risk production in 2028 and volume production in 2029.
Related: ZOTAC RTX 5090 fails in minutes despite intact connector
Production metrics for Intel’s newer nodes are improving. The yield rate for the 18A process has reportedly risen to 85%, up from 65% in the previous quarter. This progress puts Intel close to the yield rates achieved by TSMC, which is reportedly at 90% for its 2nm N2 process. Samsung’s SF2 process, by comparison, is reported to have yields between 50% and 60%.
High yields are essential for mass production because they directly impact the cost per chip. As the 18A ramp continues, Intel plans to flood the market with more Panther Lake and Wildcat Lake options later this year. The company is also expanding capacity on its existing Intel 4 and Intel 3 nodes to support this growth.
On the data center side, Intel expects its CPU segment to grow by 25% to 30% this year. This growth is driven by “agentic AI,” which is expected to drive an additional 50% increase in the coming year. The company’s aggressive capacity expansion and competitive pricing are key factors in this expected growth.
While the list of design wins is impressive, the sheer scale of the semiconductor market means Intel still has a long road ahead. TSMC remains the dominant force, and Intel will need to maintain this momentum to truly challenge its position. The company’s ability to scale production and keep costs low will be the ultimate test of these new design wins.
Related: Tensordyne Napier chip delivers 13x AI speed boost
EMIB Packaging Hits New Heights
Intel’s advanced packaging technology, including EMIB and its variations, is also showing strong performance. These solutions pose a significant threat to TSMC’s CoWoS packaging, which is currently facing supply constraints. Intel is developing glass substrate packaging at its Rio Rancho facility, which will allow for larger reticle dies in a cost-effective configuration.
The yield rate for Intel’s EMIB-T technology has reportedly reached 98%. Just three months ago, the yield rate was reported at 90%. This is the “gold standard” for semiconductor packaging, and achieving such high yields is difficult. If these figures are accurate, fabless chipmakers will likely have more confidence in Intel’s packaging capabilities.
Intel’s EMIB technology has been in mass production since 2017, but the recent improvements suggest the process is maturing rapidly. The technology is already being used by several major companies. Alleged customers for Intel’s EMIB packaging include NVIDIA, Google, and Amazon.