James Carthew: Are trusts really backing AI winners – or funding the losers?
The likes of HgCapital and Lindsell Train have come under fire for their AI holdings, but is the upside potential as powerful as they claim?
I recently attended HgCapital’s (HGT) capital markets day where the main focus was the effect of AI on its portfolio. The issues raised are crucial to HGT’s success and to the future of a number of other trusts.
Lindsell Train’s (LTI) recent results illustrated the threat that agentic AI poses to investors’ portfolios after reporting a sharp NAV fall and dividend cut. However, HGT, LTI, and its sister trust Finsbury Growth and Income (FGT) are resolute in their stance that they are invested in companies that are AI beneficiaries – not AI losers.
At the top of the AI stack are the model creators. Who is winning that race depends on who you ask, and as new model releases tend to leapfrog the competition, it is hard to pick a clear winner.
Meta (Llama) and Alphabet (Gemini) are already listed. An investment in newly-listed SpaceX gives some exposure to Grok, while IPOs for OpenAI (ChatGPT) and Anthropic (Claude) are on the way.
However, it would be remiss to ignore the Chinese competition such as Alibaba’s Qwen, private company Deep Seek, Moonshot AI’s Kimi, and ByteDance Seed, whose parent company is held by several Baillie Gifford trusts and Fidelity China Special Situations (FCSS).
There is a significant overlap between the model owners and the hyperscalers that own the data centres on which these models are developed and run on. All the big US players are pouring money into data centres filled with the latest tech.
McKinsey forecasts that AI’s demand for global data centre capacity will rise from… read more here