(Shutterstock)
Revelations of the cost-effectiveness of Chinese AI model DeepSeek have whipsawed tech stocks over the last 10 days. The VIX equity volatility index jumped from around 14 to an intraday peak of 22.5 before receding to around 16/17. The Nasdaq fell as much as 3.6% in two sessions before recovering 1.6% in the subsequent three sessions. Markets appeared to be startled after becoming accustomed to resilient US growth and a structurally positive AI story. We think this episode serves as a warning of volatility to come with tariffs arriving and more US policy uncertainty looming.
But investors should remember that in every selloff, opportunities might also emerge. This is therefore a good occasion to reiterate some key messages about managing portfolios during such periods of elevated volatility and uncertainty.
Relying on your core. The recent bout of volatility illustrates the logic behind using a 鈥渃ore鈥� component to gird portfolios against the risk of uncertainty and volatility disrupting investors鈥� financial goals. We believe that this 鈥渃ore鈥� should be a portfolio diversified effectively across asset classes, geographies, and sectors, and left alone to grow consistently for the long term. Combining asset classes with different drivers and uncorrelated returns tends to lead to smoother growth, smaller drawdowns, and faster recoveries. This approach can also help investors pursue tactical opportunities with greater confidence. With various event risks potentially producing further episodes of elevated volatility鈥攇eopolitics and climate change adding to tariffs and US policy risk鈥攐ver the next few months, investors should use volatility and elevated yields to build out their 鈥渃ore.鈥�
But volatility also favors active management. To complement a strong 鈥渃ore鈥�, we would also add a level of tactical finetuning to meet the investment needs of the day. Active investing tends to be critical in volatile and uncertain periods, less developed markets, assets with less liquidity like small caps, or in thematic investing. In particular, periods of investor risk aversion, when accompanied by high market volatility and elevated uncertainty tend to favor active management, more judicious security selection, and harnessing volatility to generate additional portfolio income. With the DeepSeek concerns still simmering in the background, investors should be prepared to act on any outsized market movements, especially in the tech space.
AI theme remains intact. We think the recent market correction was overdone and recommend taking advantage of extreme volatility through structures and buying the dip in quality AI stocks. Initial 4Q2024 earnings releases from the major tech firms indicate that AI fundamentals such as capex and adoption trends remain solid. We expect the Big 4鈥檚 combined capex for 2025 to grow around 25% year-on-year (y/y) despite the emergence of low-cost AI models like DeepSeek. The key counterpoint here is that the lower cost of AI adoption is likely to increase its usage鈥攖o the point of sustaining profit growth of 18% y/y for global tech in 2025.
Given the lack of clarity on the full set of implications of a lower cost of training AI models, it is worth considering a bear case. In this scenario, low-cost models like DeepSeek proliferate, resulting in potentially lower training and inference costs. Here, we would still expect AI training demand to be supported by robust volume growth, driven by increasing contributions from tier-2 platforms, enterprises, and emerging market customers, as well as from Big Tech on high-cost frontier models. In such a scenario we would expect profit margins for the AI compute industry to fall from 50% in 2024 to 35% in 2029. With AI compute stocks currently trading in line with the S&P 500, we still see these offering a fair risk-reward tradeoff even in the bear case.
