Investors are wobbling on AI. The recent results from the technology giants show that the vast spending continues, but there are mounting concerns about AI adoption, and whether companies will see a pay-off on their investment. One crucial factor may be the role of China and whether it can launch lower-cost AI alternatives.
Julian Bishop, manager of the Brunner Investment Trust, points out that previous technological leaps have seen companies with monopolistic tendencies emerge – Meta in social media, Alphabet in search, Nvidia in chips, Microsoft in enterprise software. This has allowed them to make super-normal returns.
It is not at all clear that AI businesses will have the same characteristics. His view is that AI may end up being a deeply competitive business in which it is very hard for a single company to dominate. In particular, he says, there are a raft of low-cost Chinese competitors.
China has built its own AI ecosystem that increasingly runs independently of its US peers. Earlier this year, DeepSeek teamed up with Huawei to launch a new AI model that doesn’t need Nvidia chips. The top Chinese AI models can now run on Chinese hardware alone.
China has taken a different approach to its AI development. It has chosen to make it “open source”, which makes it far easier for countries and companies to pick it up and run with it.
It also avoids the potential problem of companies and governments switching off access to AI. This would have been unthinkable until recently, but the current US administration has shown itself increasingly inclined to meddle in how and where its technology giants distribute their products. That means the Chinese approach has greater appeal.
Will Mackintosh Whyte, co-manager of the Rathbone Strategic Growth Portfolio, says: “China is increasingly competing with the US on AI on cost and increasingly competitive on model capability; however the US retains advantages at the absolute frontier and in access to leading-edge computing power.
“It is also important to note that while China is closing the intelligence gap, claims of equivalent intelligence at dramatically lower cost aren’t yet evident and the US remains ahead once you assess cost per completed task rather than headline cost per token.”
Nevertheless, he says that China doesn’t necessarily need to beat the US at building the world’s best frontier model to create enormous economic value from AI. “Cheap, capable models deployed widely across industry could drive productivity and support Chinese and, indeed, global corporate earnings,” he adds.
Dale Nicholls, manager of Fidelity China Special Situations, agrees. He says that despite the way it has been framed by governments, it isn’t a question of one side ‘winning’ the AI war.
“‘Winning’ implies there is a finish line and that all cannot benefit from the developments. The two ecosystems have different advantages. The US remains at the frontier in areas such as advanced chips and model development, while China brings enormous manufacturing scale, engineering talent and a very competitive cost base for power and infrastructure,” he says.
“Chinese companies have also demonstrated an ability to innovate around chip constraints and develop domestic alternatives. Importantly, AI investment by Chinese hyperscalers remains considerably below that of their US counterparts, suggesting there is still substantial room for the ecosystem to develop.”
In terms of which is the better investment, that prize has gone to US AI to date, but it is not clear that it will be able to sustain that advantage. Direct comparisons are difficult, but the iShares AI Innovation and Tech Active ETF is up 40.65% for the year to date*. The Global X China Robotics and AI ETF is down over 25% over the same period**. The Chinese hyperscalers Tencent and Alibaba are down 39% and 40% respectively over one year***. The CNBC Mag 7 index, a proxy for the US hyperscalers, is up 14% over the same period^.
That said, some parts of the AI trade in China have done very well. Sharukh Malik, manager in the Asian equities team at Guinness Global Investors, has been taking profits on those companies that have performed best. He is playing the AI trade through industrials that are benefiting from the power build out, which he sees as “a very interesting space in China that many investors aren’t really paying enough attention to”.
This is also a priority for Roderick Snell, manager of the Baillie Gifford Pacific fund. He is taking a barbell approach, maintaining investments in companies such as TSMC, Samsung and SK Hynix, but also investing in Chinese raw materials companies. He says the new industrial age will be more dependent on energy, resources and construction and this is where he is focusing some of the portfolio^^.
He is also looking to the next phase of development – robotics. “We participated in the September IPO of Mech Mind, a company that develops 3D machine-vision systems – effectively the eyes of robots. Although still early, we think physical AI could become the next EV industry for China, as the country combines relatively low-cost electricity and increasingly capable AI models with one of the world’s deepest manufacturing ecosystems.”
As it stands, robotics companies in the US have come up against higher manufacturing costs, which has stalled progress.
Even if Chinese AI cannot make the extraordinary advances of the frontier models, it can disrupt the profitability and adoption of US AI models. LLMs already look like a crowded space. AI may not be a winner-takes-all technology. With that in mind, investors should be wary of assuming it can deliver the extraordinary profitability of previous technological developments.
*Source: BlackRock, 30 September 2026
**Source: Mirae Asset Global Investments (Hong Kong), 21 October 2026
***Source: Google Finance, 22 October 2026
^Source: CNBC, 21 October 2026
^^Source: fund factsheet, 30 September 2026
Darius McDermott is managing director of FundCalibre and Chelsea Financial Services. The views above are his own and do not constitute financial advice.