TSMC plans to raise monthly CoWoS output to 200,000 wafers by 2027
According to Taiwanese sources, TSMC intends to raise monthly output of its CoWoS advanced packaging modules from the current 13 million wafers (end of 2026) to 20 million in 2027, and potentially to 24-26 million by the end of 2027. The trajectory: 2024 — 30,000 wafers, 2025 — 70,000, 2026 (as of October) — 130,000. Demand for CoWoS (Chip-on-Wafer-on-Substrate) is critical for high-performance AI chips.
AI-processed from 36Kr (36氪); edited by Hamidun News
TSMC is aggressively expanding CoWoS packaging capacity to meet demand for artificial intelligence chips. According to supply chain sources reported in July 2026, the company plans to hit several capacity milestones.
Current and future output
Growth in monthly output:
- 2024 — 30,000 wafers
- 2025 — 70,000 wafers
- 2026 (as of October) — 130,000 wafers
- 2027 (target) — 200,000 wafers
The market is optimistic, expecting 240-260 thousand wafers per month by the end of 2027. CoWoS is critical for high-performance chips that require dense 3D die packaging.
Why it matters for AI
Packaging capacity is a bottleneck in the AI chip supply chain. Faster production at TSMC addresses the shortage that has held back new accelerators for neural networks and LLM services. All of TSMC's new production lines run 24/7.
What it means
TSMC's capacity expansion signals that demand for AI chips remains strong and will keep growing at least through the end of 2027. That is a good sign for makers of GPUs and specialized accelerators.
Key Questions Answered
What are the key details on Current and future output?
Growth in monthly output: 2024 — 30,000 wafers 2025 — 70,000 wafers 2026 (as of October) — 130,000 wafers 2027 (target) — 200,000 wafers The market is optimistic, expecting 240-260 thousand wafers per month by the end of 2027. CoWoS is critical for high-performance chips that require dense 3D die packaging.
Why it matters for AI?
Packaging capacity is a bottleneck in the AI chip supply chain. Faster production at TSMC addresses the shortage that has held back new accelerators for neural networks and LLM services. All of TSMC's new production lines run 24/7.
What it means?
TSMC's capacity expansion signals that demand for AI chips remains strong and will keep growing at least through the end of 2027. That is a good sign for makers of GPUs and specialized accelerators.
Test Frontier LLMs & Autonomous Agents in One Workspace
Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.