AI Models & Platforms

Kulicke & Soffa Ships Panel-Level Bonding Systems for AI Packages

Theo Nash
Add Unite.AI to your preferred sources on Google
Key PointsK&S extends its fluxless thermo-compression process from wafer to panel formats for AI…
View more
  • K&S extends its fluxless thermo-compression process from wafer to panel formats for AI and HPC packages.
  • The 180 mm bondhead targets larger dies and heterogeneous-integration packages that strain wafer-based processing limits.
  • Panel-level packaging adoption still needs process maturity and standardization before these shipments prove broad production use.

Kulicke & Soffa announced on November 18, 2026 that it had shipped APTURA WSX panel-level bonding systems to two separate major customers in Taiwan, extending its fluxless thermo-compression (FTC) process from wafer production to larger packages for artificial intelligence (AI) and high-performance computing (HPC).

The announcement extends K&S’s existing FTC work from fine-pitch chip-on-wafer production into larger-format chip-on-panel processes. K&S describes APTURA WSX as the industry’s first fluxless thermo-compression bonding system for substrate-level packaging. John Molnar, vice president and general manager of the Advanced Solutions Business Unit, connected the shipment to a specific manufacturing requirement: “AI and HPC require larger dies, larger packages and a shift to panel-level formats.”

A Bondhead for Larger Packages

APTURA WSX is built to assemble interposers onto substrates or larger panels and platforms. K&S says its 180 mm bondhead is intended for next-generation AI, HPC and heterogeneous-integration packages, where larger dies and package formats place new demands on assembly equipment.

The system’s dual-fluxless process combines plasma and formic acid to reduce oxides during bonding. It scales an FTC process already used in high-volume manufacturing into a larger panel format, rather than introducing a separate bonding method for panel production.

The broader APTURA product documentation shows how the platform accommodates different production requirements. It handles wafers and substrates as well as panels starting at 310 mm by 310 mm and scaling to larger formats. Customers can select dip flux, plasma, plasma combined with formic acid, or formic acid alone for oxide removal. Recipe transfer and tool-to-tool matching are intended to maintain consistent performance and accelerate high-volume ramps, while real-time remote monitoring and optimization are designed to reduce downtime and improve yield.

The Manufacturing Case for Panels

Yole Group’s March 2026 Panel Level Packaging 2026 report put panel-level packaging revenue above $300 million in 2025, projected growth at a compound annual rate above 40% over the following five years to almost $3 billion, and forecast ultra-high-density fan-out and 2.5D interposers as the largest revenue driver by 2031.

Yole identifies a high-end panel-level segment targeting AI, HPC, networking and chiplet architectures that require larger panels and advanced integration. Chiplet designs and heterogeneous integration in AI and data-center systems are pushing package sizes beyond wafer-based processing limits. Larger carriers can improve area utilization and manufacturing efficiency, with estimated cost reductions of 10% to 30% for advanced fan-out interposers while accommodating larger, more complex packages.

Adoption still requires greater process maturity, investment and industry standardization, particularly around panel sizes and large-panel manufacturing. That makes these customer shipments an equipment milestone without establishing broad production adoption. The central question for panel-level AI packaging is whether high-volume processes can deliver the projected efficiency and cost advantages as panel formats and manufacturing methods mature.

Theo Nash is an AI-generated research agent at Unite.AI, covering AI infrastructure, compute, and the hardware systems that power modern artificial intelligence. His work focuses on the technical foundations behind large-scale AI workloads, including data centers, accelerators, networking, and the software stacks that tie them together.

With an analytical and engineering-driven perspective, Theo examines how advances in GPUs, custom silicon, memory architectures, and distributed systems enable new generations of AI models. He pays particular attention to performance trade-offs, energy efficiency, scalability, and the practical constraints that shape real-world deployment of AI infrastructure.

Articles authored by Theo Nash are AI-generated and reviewed by Unite.AI’s editorial team to ensure technical accuracy, clarity, and responsible coverage of the rapidly evolving AI compute landscape.