TSMC Boosts CoWoS Capacity, Reducing AI Chip Supply Gap to 10% by Late 2026; Next-Gen CoPoS Pilot Production for NVIDIA’s ‘Feynman’ Slated for 2027

Futuretech Components / Daily Beirut Taiwan

Overview

TSMC is rapidly expanding its CoWoS advanced packaging capacity to meet surging demand from AI accelerators and HPC applications, projected to narrow the supply-demand gap from 20% to approximately 10% by the end of 2026. Concurrently, the company is advancing its next-generation CoPoS (Chip-on-Panel-on-Substrate) technology roadmap, with pilot production expected in 2027 and NVIDIA’s upcoming ‘Feynman’ platform poised to be an early adopter. These strategic moves aim to alleviate AI chip shortages and drive future high-performance computing innovation.

In Depth

Key Findings

TSMC is aggressively expanding its CoWoS (Chip-on-Wafer-on-Substrate) advanced packaging capacity to address the unprecedented demand emanating from artificial intelligence (AI) accelerators and High-Performance Computing (HPC) applications. This concerted effort is projected to reduce the current 20% supply-demand gap for CoWoS to approximately 10% by the end of 2026, significantly alleviating ongoing AI chip supply constraints.

Technical & Production Details

CoWoS is a leading 2.5D/3D packaging technology that integrates multiple logic dies and HBM (High Bandwidth Memory) on a silicon interposer, which is then connected to a substrate. This architecture dramatically enhances data transfer rates and power efficiency, delivering the performance crucial for complex AI model computations. TSMC’s mastery of CoWoS technology positions it as a vital service provider for major AI chip manufacturers.

Furthermore, TSMC is actively developing its next-generation packaging technology, CoPoS (Chip-on-Panel-on-Substrate). CoPoS aims to leverage larger panel substrates to process more chips simultaneously, promising reduced manufacturing costs and increased throughput compared to CoWoS. Pilot production for CoPoS technology is slated for 2027, with NVIDIA’s forthcoming AI platform, “Feynman,” anticipated to be one of the first products to adopt this innovative technology. The introduction of CoPoS, particularly with the potential integration of glass core substrates, is expected to further cut costs and boost wafer utilization efficiency.

Background & Context

The proliferation of AI has ushered in a new wave of growth for the semiconductor industry, but it also presents novel manufacturing challenges that traditional processes cannot meet. The tight integration of HBM and logic chips is a critical factor determining the performance of AI accelerators. Advanced packaging technologies like CoWoS and CoPoS are key to achieving this integration. TSMC’s aggressive investment in these areas is a strategic move to maintain its market leadership and support the burgeoning AI ecosystem.

Strategic Significance & Outlook

The expansion of CoWoS capacity and the introduction of CoPoS will help mitigate AI chip shortages, enabling a greater volume of AI products to enter the market. This, in turn, will accelerate the widespread adoption of AI technologies and foster innovation across various sectors, including autonomous driving, cloud computing, and edge AI. TSMC’s ongoing initiatives in advanced packaging are poised to be significant milestones, shaping the future direction of semiconductor manufacturing technology.

Get our weekly technology intelligence — free

Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.

Subscribe Free — Weekly Tech Intelligence

By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.

• Your email and selected fields are used only to deliver the newsletter.
• We never share your information with third parties.
• You can unsubscribe anytime via the link in each email.

See our Privacy Policy for details.

Agree & Continue ▶

Takes about a minute · Unsubscribe anytime