Schroders Investment Management has provided its perspective on the potential disruption caused by Chinese start-up DeepSeek in the artificial intelligence space, and how it could impact stock markets and tech companies.
Stock markets suffered a sell-off on 27 January as DeepSeek claimed new advances in the field of AI. The Nasdaq index fell by 3%, and the shares of some AI-related companies were down much more.
DeepSeek has reportedly developed large language models comparable to market leaders but at significantly lower training costs. If these reports are accurate, this could imply a reduced demand for high-performance semiconductors required for the computational workloads associated with AI, say Paddy Flood, portfolio manager and technology sector specialist, and Simon Webber, head of global equities, at Schroders.
They say this conclusion hinges on whether DeepSeek’s cost data is truly comparable to the incumbent players and, more importantly, that other factors remain unchanged.
“Increased efficiency in compute power does not necessarily translate into reduced demand for chips. Jevon’s Paradox – a well-established economic concept – suggests that improvements in resource efficiency often drive greater consumption of that resource. In this context, higher compute efficiency could spur further adoption and expansion of AI, potentially offsetting any direct reduction in chip demand,” say Flood and Webber.
If increased compute efficiency leads to reduced demand for chips/AI equipment, companies such as Nvidia and other compute infrastructure providers could face headwinds. However, this outcome is far from certain, particularly given Jevon’s Paradox.
On the other hand, this development could prove favourable for software companies. Lower AI costs might make these technologies accessible to a broader base of customers who were previously deterred by high price points. For software providers embedding AI capabilities into their products, this could bolster adoption while preserving profitability, say Flood and Webber.
“Additionally, large hyperscale companies such as Microsoft, Meta, and Google could stand to benefit. Concerns have been growing around the potential returns on their substantial AI-related investments. If this situation results in reduced spending requirements for these companies, it could lower their capital expenditure needs and drive significant increases in free cashflow generation.”
Flood and Webber say many uncertainties need to be resolved, such as clarifying DeepSeek’s cost structure and whether cheaper infrastructure will really lead to less spending in the global AI race. This uncertainty does create risk, but the severity of these moves is likely to create opportunities for active investors, particularly in technology and industrials.
Schroders’ group chief investment officer, Johanna Kyrklund, says the full implications of DeepSeek’s technology still need to be understood. However, last week’s sell-off highlights that markets are vulnerable to a misstep by one of the large US mega-caps, or by the emergence of new competition.
Highly concentrated equity markets can pose a danger to investors. The level of index concentration now far surpasses that of the late 1990s.
Kyrklund says it is not prudent for an investment portfolio to have a high exposure to a handful of companies. “Understanding the underlying stocks is crucial, and an active approach is needed to manage the risks.”
Disclaimer: The information in this article does not constitute investment or financial planning advice that is appropriate for every individual’s needs and circumstances.
