Intel debuts Panther Lake chip built on 18A, claims 60% faster

Intel launched its Panther Lake AI chip for laptops at CES in Las Vegas, introducing its first product built using the company’s new 18A manufacturing process.

The Panther Lake chips, called Intel Core Ultra series 3, use a new transistor design and power delivery system enabled by the 18A process.

Intel’s CEO said the company has made good on its promise to ship the first 18A-based products in 2025.

Unlike the previous Lunar Lake chips, which were mainly produced by TSMC, Panther Lake is Intel’s first high-volume in-house chip as it works to regain market share from AMD.

Intel said the new chips deliver 60% better performance than the previous generation.

🔗 Source: Reuters

🧠 Food for thought

Implications, context, and why it matters.

Leak hints at a limited 2025 Panther Lake rollout as 18A issues may persist

Intel planned only one Panther Lake stock keeping unit (SKU) for Q4 2025 1. That single 4 performance cores, 8 efficiency cores, plus a 45W Thermal Design Power (TDP) option does not look like high volume 1.
Staggered timing likely comes from 18A ramp limits 1. Energy‑efficient variants for ultrabooks (thin-and-light laptops) do not arrive before 2026, which leaves 2025 room for AMD Strix Point laptops and Qualcomm Snapdragon X systems 1.
18A carries weight for Intel roadmaps plus outside foundry customers (chip designers that outsource manufacturing) such as Arm (which licenses chip architectures) and maybe Nvidia 1. The first wave serves as a credibility test over immediate sales 1.

AI developers can use Intel’s Neural Processing Unit (NPU) with OpenVINO for on-device generative AI

Independent software vendors (ISVs) can build on-device AI features with OpenVINO 2. The open-source toolkit from Intel supports PyTorch, TensorFlow, plus Open Neural Network Exchange (ONNX) models with optimized inference pipelines 2.
OpenVINO supplies Python and C++ application programming interfaces (APIs) with tools for generative AI 2 3. Developers can deploy lightweight models locally instead of cloud services 3. Automatic device discovery spans CPU, GPU, and NPU 3.
Intel needs strong co-marketing with original equipment manufacturer (OEM) partners plus clear software development kit (SDK) docs 1. Wider Panther Lake availability is expected in 2026 when more SKUs ship 1.