Intel has recently made impressive strides with its 18A (1.8 nanometers) process, capturing the attention of prominent tech companies. According to supply chain sources, Intel has secured a significant foundry contract with Microsoft to manufacture chips using this cutting-edge technology, while also holding discussions with Google, NVIDIA, and others for potential collaborations.

The Intel 18A process stands out due to its innovative PowerVia backside power delivery and RibbonFET gate-all-around transistor architecture. PowerVia separates power and signal layers, optimizing signal routing and reducing resistive voltage drop, enhancing chip performance by roughly 15%. The transistor density of the 18A process is 30% greater than that of the Intel 3 process. Additionally, the SRAM bit cell size has been reduced from 0.03 sq. microns to 0.023 sq. microns, nearing the density levels achieved by TSMC's N2 process. Although TSMC's N2 process boasts a higher transistor density with its high-density standard cells, Intel’s 18A excels in power consumption and frequency capabilities, particularly in high-performance contexts. Furthermore, Intel's Foveros Direct 3D hybrid bonding technology offers copper-to-copper connections with less than 5-micron pitch, surpassing TSMC's SoIC-X technology. This advancement is critical for AI chip and high-performance computing interconnect solutions.

Intel aims to launch its first 18A process products in the latter half of 2025, including the Panther Lake processors for AI PCs and the Clearwater Forest for servers. Panther Lake is designed to deliver up to 40 TOPs of AI computations, with engineering samples already successfully booting operating systems and demonstrating target frequency DDR memory performance. On the other hand, Clearwater Forest leverages RibbonFET, PowerVia, and Foveros Direct technologies to provide a dense, low-power solution targeted at data centers. Intel has announced the release of version 1.0 of its Process Design Kit (PDK) for customers designing chips, with broader mass production anticipated by Q4 2025 and complete availability by early 2026. To increase production capacity, Intel is expediting equipment installation at Fab 52 in Arizona and initiating casting at the Hillsboro, Oregon R&D center.

The current state of the semiconductor industry presents a strategic opening for Intel. Despite TSMC's N2 process expected to reach mass production by late 2025 with approximately 50,000 wafer output per month, the demand remains overwhelming, pushing clients to seek alternatives. Intel's 18A process attracts customers like Microsoft due to its robust performance and domestic manufacturing advantages in the U.S. Companies like NVIDIA, which require high-performance and energy-efficient processes, and Google's TPU series, which could benefit from improved interconnect technology, are evaluating this process. Companies such as Broadcom are also actively testing it. In contrast, Samsung’s SF2 process, scheduled for trial production in Q1 2025, lags in production volume and competitiveness, awaiting further validation. Intel’s edge lies in superior technology maturity and customer trust. Additionally, benefiting from a $2.2 billion subsidy under the U.S. CHIPS Act for expanding manufacturing and R&D, Intel is poised for growth.

Under the new leadership of CEO Lip-Bu Tan, appointed in early 2025, Intel is prioritizing the expansion of its foundry business. Highlighting the 18A process as crucial for regaining technological leadership, Tan plans to enhance Intel's competitive stance through automation, advanced packaging, and comprehensive foundry services. Speaking at the Vision 2025 conference, Tan committed to prioritizing the demand for Intel's internal products while also attracting external clients, with initial external customer orders expected by mid-2025. Intel may adjust its "IDM 2.0" strategy, fostering market orientation while deepening its collaboration with TSMC to mitigate initial production pressures.

Intel’s 18A advancement is pivotal for the AI PC and data center markets. IDC forecasts global AI PC shipments to hit 120 million units in 2025, capturing 35% of the PC market share. Panther Lake, if launched as planned, will be instrumental for Intel in securing a competitive foothold. Nevertheless, achieving a yield of at least 70% is critical for mass production. Intel indicated that Panther Lake’s yields have surpassed those of the Meteor Lake timeframe and are progressing as expected. However, risks associated with high numerical aperture EUV lithography require attention due to their complexity and cost, leading Intel to also maintain traditional process options for flexibility in design.

The advancements made by Intel’s 18A process provide a much-needed competitive thrust in the semiconductor sector. Through the integration of PowerVia, RibbonFET, and Foveros Direct 3D technologies, the 18A can rival TSMC's N2 in terms of performance and efficiency. As its mass production phase approaches in 2025, the 18A process is set to drive Intel’s product enhancements for both clients and servers, potentially reshaping the global foundry market.