Intel’s latest advancements are highly significant, representing a direct response to TSMC’s recent official announcements. Information presented at the VLSI 2026 Symposium reveals Intel Foundry’s Intel 18A-P T1.2 paper, featuring concrete technical details that highlight Intel’s aggressive push.
Intel had previously hinted at an advanced version of its foundational node, suggesting an improvement of around 8% over the standard 18A. The newly disclosed figures are even more compelling and reveal crucial details.
Intel 18A-P: A Deep Dive into its T1.2 Paper and Node Characteristics
While specific details are limited for proprietary reasons, the available information offers valuable insights. This analysis assumes a foundational understanding of lithographic nodes.
The previously mentioned 8% performance improvement has now been officially updated to over 9% performance at equal power, or over 18% lower power consumption at equal performance. Intel demonstrated these figures using a 0.75V measurement on an Arm core subgroup. Crucially, the Intel 18A-P maintains compatibility with the 18A node, continuing to utilize RibbonFET and PowerVia technologies without altering the base geometry – a sensible approach for an evolutionary refinement.
Key specifications that were previously undisclosed are now public, offering a clearer understanding of the node’s true capabilities. While Panther Lake processors are already on the market, demonstrating superior efficiency against x86 competitors and challenging only Apple’s Arm-based solutions, the 18A-P promises further advancements.
Current information indicates that the Contacted Poly Pitch (CPP), which measures the distance between gates and reflects lateral scaling, remains at 50 nm. The Library Height (LH) is also consistent at 180 nm and 160 nm, as this defines standard logic cells. The significant innovations lie within the broader general specifications, aimed at strengthening Intel Foundry Services (IFS) for 2026 and beyond into 2027.
A Joint Optimization Exceeding Expectations for TSMC’s N2P and A16
The primary enhancements stem from improvements in devices, voltages, interconnects, thermal management, and Design-Technology Co-Optimization (DTCO). The technical documentation uses “W” (Watts) rather than “Z,” which is important to understand. W1, W2, and W3 represent effective widths of RibbonFETs; higher wattage implies greater current, thus lower wattage contributes to reduced power consumption and leakage.
As previously known, Intel 18A offers W2 and W3 at 180 nm, and W1, W2, and W3 at 160 nm. The latest leaks reveal that Intel 18A-P introduces W1 Low Power at 180 nm, W1.5 Low Power at 160 nm, and W3P with High Performance (HP) contact at both heights. “Low Power” focuses on efficiency, while “HP” aims to deliver higher current and power for increased performance in CPUs, GPUs, or ASICs.
Further advancements include refinements to Threshold Voltages (VT). ULVT (Ultra-Low Threshold Voltage) enables earlier switching and higher frequencies but increases leakage. LVT (Low Threshold Voltage) is a less aggressive option. While Intel 18A featured four pairs of Logic VT, Intel 18A-P now offers five or more, introducing an additional VT between ULVT and LVT, and reducing the ULVT by 10 mV.
Additionally, the company has narrowed skew corners by 33%, reduced resistance in lower interconnect layers (V0-V2), added jogs in M2-M4, improved thermal conductivity by 50%, matched SRAM Vmin (minimum SRAM voltage), and enhanced NBTI (Negative Bias Temperature Instability), a form of degradation caused by electrical and thermal stress.
In comparison to Intel 3, the strengths and novelties were already present in 18A, with 18A-P adding further refinement without altering the core architecture. TSMC presented its A16 node at the same VLSI symposium, featuring nanosheets, Super Power Rail technology, offering 8% to 10% higher speed, 15% to 20% lower power consumption, and 8% to 10% greater density, with production slated for Q4 2026. However, TSMC omitted several key technical details.
The Intel 18A-P and its T1.2 paper demonstrate improvements that extend beyond a simple refresh or minor optimization. These enhancements position it closer to TSMC’s A16 than to a direct competitor to N2P, which would be its primary market rival. This will likely be evidenced in upcoming CPU releases, where architectural battles will determine the true winners, potentially by 2027 or 2028, contingent on the demands of the AI sector.
The entry Intel 18A-P paper T1.2 revealed: how the blues will compete against the threat of TSMC with its N2 and N2P for 2027 and 2028 first appeared on El Chapuzas Informático.
