TSMC Ties Capacity Allocation to Credit Limits Amid AI Financing Concerns
To address growing concerns over AI circular financing, TSMC is evaluating customer credit limits based on financial statements and requiring cash payments or prepayments for orders exceeding those thresholds. Additionally, the company is strictly prohibiting capacity subleasing or auctioning while vetting data center construction progress. This move highlights a significant shift in the semiconductor supply chain toward stricter financial risk management as the market watches for potential AI bubbles. By tying production allocation directly to balance sheet health, TSMC is insulating itself from the credit risks of heavily leveraged AI and ASIC startups. As of June, TSMC's customer advance payments reached 234.23 billion New Taiwan Dollars, reflecting the scale of upfront capital involved. TSMC Chairman C.C. Wei noted that capacity planning now comprehensively incorporates cloud service provider capital expenditures, financials, power availability, and component supply.
## BACKGROUND
AI circular financing refers to high-risk business structures where tech giants invest heavily in AI startups that simultaneously use those funds to purchase chips and cloud infrastructure from the same investors. Special Application Integrated Circuits (ASICs) are customized chips designed for specific workloads like AI training and inference, which are increasingly sought after alongside general-purpose GPUs.