Intel’s latest earnings report signals a pivotal moment for its foundry division. The company now says it may pause or even halt the development of its 14A and future advanced process nodes unless it can attract more external clients and reach scale-related benchmarks.
As a semiconductor pioneer, Intel has long been central to the industry’s evolution. Co-founder Gordon Moore’s famed law—that transistor counts double every two years—fueled decades of innovation. Intel’s early chips like the 4004 launched the PC era, and later, its dominance in the x86 market with the Pentium and Core lines cemented its place at the top. More recently, Intel sought to compete with TSMC and Samsung by opening its manufacturing services to outside clients through Intel Foundry Services (IFS).
However, IFS has struggled to gain momentum. In Q2 2025, Intel’s revenue held steady at $12.9 billion year-over-year, but net losses reached $2.9 billion, mostly due to restructuring and operational costs. Foundry revenue totaled $4.4 billion, largely from internal use, with only modest contributions from external customers. Despite investing tens of billions into fabs and R&D, Intel has yet to achieve the necessary scale. Of its 12 active IFS projects, eight are expected to wrap in 2025, though most target mature process nodes.
Leadership Shift and New Priorities #
New CEO Lieu Chen, who joined in March 2025, brings deep semiconductor industry experience. Formerly Cadence Design Systems’ CEO, Chen aims to refocus Intel’s efforts on profitable areas. Under his leadership, Intel has declared that IFS must prove economically viable and will no longer receive unlimited funding.
The Stakes of Advanced Nodes #
Intel’s 18A node (a 1.8nm-class process) is now in risk production. Featuring RibbonFET transistors and PowerVia backside power delivery, 18A promises a 15% performance-per-watt gain and 30% density boost over Intel 3. It can deliver 25% higher clock speeds or 36% lower power at the same speed. Internal products like Panther Lake and Clearwater Forest will debut this node. Performance tests place it ahead of TSMC’s N2 and Samsung’s SF2—though its SRAM density lags behind, potentially limiting AI and mobile use cases.
Next up is the 14A node (1.4nm-class), which plans to integrate second-gen PowerDirect delivery and Turbo Cell technology for a projected 15–20% efficiency boost and 1.3x density increase. High-NA EUV tools will be used for fabrication. While sampling has begun, Intel’s 10-Q warns that unless customer interest and milestones materialize, 14A’s expansion could be suspended.
Market Dynamics and Adjustments #
TSMC continues to lead with its N2 node entering mass production in 2025. With nanosheet transistors and better SRAM density and yield, TSMC has locked in major clients like Apple and NVIDIA. Samsung’s SF2 follows closely, utilizing similar Gate-All-Around tech. Though Intel’s 18A performs well in benchmarks, it lags in ecosystem maturity and customer trust.
In response, Intel has announced 18,000 layoffs (15% of its workforce) and postponed some initiatives, including its Ohio fab, to conserve R&D capital. Consumer chip sales fell 3% in Q2, while data center growth helped offset losses. Intel is also growing its IFS Accelerator ecosystem to fast-track customer designs via EDA and IP partnerships.
Looking Forward #
If Intel steps back from advanced nodes, it could shift U.S. semiconductor strategy. While the CHIPS Act supports Intel’s fab construction, TSMC’s Arizona plants—planned to use N3 and N2—might become the primary U.S. source for cutting-edge chips.
Intel still targets becoming the world’s second-largest foundry by 2030 and claims $15B in external orders from firms like Microsoft and Amazon. It’s also betting on advanced packaging solutions such as EMIB and Foveros to compensate for node limitations and differentiate itself.
In the short term, Q3 revenue is projected at $12.6–13.6 billion, suggesting stabilization. But the road ahead demands that Intel balance innovation, cost control, and customer confidence if it wants to remain competitive in the race for next-generation chips.