The First Systems Foundry Designed for the AI Era
Shifts in the technology landscape and foundry customer needs require new approaches to solution delivery. Intel Foundry provides unique capabilities to address today’s most challenging systems requirements.
The Landscape of Technology is Changing
AI chips consume more power than 61% of countries.1
The projected compute growth for ML training is >3x per year over the next decade.2
Compute is outpacing memory bandwidth.3
Geographically-diverse supply of advanced semiconductors is critical to supply chain resiliency.
Customer Needs are Shifting
Foundry revenue from HPC/AI designs continues to increase.
Chiplets are on track to surpass monolithic die by 2028.4
Today’s and tomorrow’s solutions require system technology co-optimization (STCO).
The World Needs a Systems Foundry for the AI Era
Compute-intensive applications like AI and HPC demand greater performance improvements than can be delivered by the advancement of semiconductor manufacturing alone. We need new approaches to address rising power consumption, satisfy data throughput requirements, and minimize operating expenses. Foundries must optimize at the system level, from silicon to software and at every layer in between.
Hear more from Foundry Services head Kevin O’Buckley
World-Class Foundry Offerings
Process Technologies
Design your next generation of leadership products with our breakthrough technologies Intel 18A, Intel 3, and Intel 16.
Advanced Packaging and Test
Intel Foundry offers cutting-edge interconnects, leadership in 2D, 2.5D, and 3D packaging, and comprehensive assembly and testing services.
Manufacturing
Leverage robust, geo-diverse, secure, and expanding supply for wafers and assembly & test. Our leadership in sustainability helps customers meet their targets and regulatory requirements.
Intel Foundry Portal
Product and Performance Information
Source: AI GPU (https://www.nvidia.com/en-us/data-center/h100/) Projected Power Consumption (https://worldpopulationreview.com/country-rankings/energy-consumption-by-country) projected against Energy Consumption by Country (https://www.tomshardware.com/news/nvidia-to-reportedly-triple-output-of-compute-gpus-in-2024-up-to-2-million-h100s).
Source: Bjorlin, Alexis, "Deploying AI at Meta Scale", https://www.kisacoresearch.com/sites/default/files/presentations/ai_hw_summit_keynote-distro-final.pdf.
Source: Gholami, et al., “AI and Memory Wall,” IEEE Micro Journal, 2024, https://arxiv.org/abs/2403.14123.
Yang, et al., “Challenges and Opportunities to Enable Large-Scale Computing via Heterogeneous Chiplets.” https://arxiv.org/pdf/2311.16417.