TSMC focuses on AI, 3nm production capacity under pressure

TSMC, the Taiwanese semiconductor manufacturing giant, is reallocating its production resources to prioritize chip deliveries for the artificial intelligence sector and its strategic clients. This strategic decision comes in a context where 3nm production capacity remains limited, creating potential bottlenecks in the supply chain.

The demand for high-performance chips for AI applications, such as training large language models (LLM) and inference, is growing strongly. TSMC, as the main foundry supplier for companies like NVIDIA, AMD and Apple, is facing the challenge of managing high demand with a production capacity that cannot keep up.

The 3nm technology is crucial for the realization of increasingly powerful and energy-efficient chips, necessary to support the growing computational requirements of AI. The limited availability of this technology could impact the delivery times and costs of new AI accelerators, affecting the entire ecosystem as a result.

For those evaluating on-premise deployments, there are trade-offs between performance, costs, and data control. AI-RADAR offers analytical frameworks on /llm-onpremise to evaluate these aspects.