The first month of the year often leads to a change in direction for markets as investors assess their annual outlook and consider longer term trends. However, the inauguration of President Elect Donald Trump in January, added to uncertainty as the market attempted to price in the policy promises being made and, towards the end of the month, actual action being taken. The most market sensitive of these were threats of tariffs on various regions and at differing magnitudes. The rationales for tariffs have been wide ranging although the threat has mostly been targeted at countries that have particularly large goods trade surpluses with the US. Trump is expected to utilise tariffs as a key diplomatic tool to extract concessions.
Other areas of focus have included immigration and more socially orientated policy changes. However, a spotlight on US tech dominance, including the Stargate AI project totaling $500bn of investment was a major win for the incoming administration. Importantly, this included inward investment from Japan and the Middle East, on top of the support of US tech giants, a major vote of confidence by the business world.
Nevertheless, towards the end of the month, markets were rocked by the revelation that Chinese AI company DeepSeek was able to deliver similar results to US located ChatGPT and Meta large language models. Critically, DeepSeek was able to achieve this with a lower use of high powered chips widely used in US AI datacenters. Shares in Nvidia, the main provider of these chips fell over 10% on the news, and contagion spread across most of the US tech sector. The performance of the DeepSeek model puts into question the high levels of investment slated by US tech giants to deliver AI services. While more analysis will need to be done to prove the performance of the Chinese AI models is genuine, the episode illustrates the risk to investors in these stocks, which remain highly valued.
Competition within AI may be bad news for markets in the short term, however, more efficient delivery of solutions should ultimately prove positive for the economy and users of AI. Indeed, if the technology can be deployed faster and more cheaply it may bring forward the development and benefits that are already expected.
