AI · SEMICONDUCTOR · FACT CHECK
SKD WIREMostly True
Anthropic, the US AI developer behind the Claude family of large language models (LLM), recorded cumulative losses amounting to approximately 107 trillion won, according to a claim circulating in Korean tech media that has been rated as largely accurate.
The figure of 107 trillion won reflects Anthropic's losses, a sum that underscores the enormous capital expenditure required to compete at the frontier of AI development. Training and operating large language models demands massive computing infrastructure, and Anthropic — like its closest rivals — has spent heavily on GPUs and data center capacity to keep pace in the generative AI race.
The claim, reviewed against 11 primary sources, was judged to hold up in its core assertion. The losses are tied to Anthropic's aggressive investment cycle rather than a collapse in demand; the company continues to attract enterprise customers for its Claude models even as spending on compute outpaces revenue.
For Korea's semiconductor and hardware industry, Anthropic's loss profile is a double-edged signal. Sustained AI investment at this scale has been a key driver of demand for advanced memory such as HBM and for cutting-edge foundry capacity — segments where Samsung and SK hynix compete directly. Any slowdown in frontier AI spending would ripple through these supply chains.
The overall assessment of the claim is: Mostly True.
