In January this year, the world's emerging artificial intelligence AI markets were rocked when Chinese company DeepSeek released a model they claimed to be significantly cheaper than the dominant US alternatives.
When announcing the launch of their "R1" model - an open-source large language model that works in a similar way to services such as ChatGPT - DeepSeek said they were able to train the model for just 6m, compared to the more than 100m US giant OpenAI spent producing ChatGPT-4. The massive cost savings were achieved because DeepSeek's models require the use of far fewer advanced chips.
The news sparked a market rout on Wall Street, with investors fearing that major US tech firms could be undercut by cheaper rivals in China. Concerns over the future demand and value of advanced ships saw the American chipmaking giant Nvidia shed almost 600bn in market value on a single trading day - the biggest single-day loss in US history.