In 1957, the Soviet Union launched Sputnik 1, the world's first artificial satellite, into space orbit. The man-made craft sent a radio signal back to Earth for three weeks before its batteries depleted. It subsequently fell back into the atmosphere after three months, having circled Earth more than 1,400 times.
The Soviet rocket that launched Sputnik into space produced a thrust that was a few times stronger than that of an American-made rocket, a claim that was doubted by many US officials at that time.
The success of the Sputnik launch was not anticipated by the US, and the ensuing "Sputnik Crisis" was a period of public anxiety and fear about the perceived gap in technology between the Soviet Union and the US. The Sputnik moment also sparked the beginning of a new era of military, technological, political and scientific developments.
Fast-forward close to seven decades later, the global artificial intelligence (AI) economy got its equivalent Sputnik moment after DeepSeek, a small Chinese AI startup, stole the show from Donald Trump's inauguration with the news it had managed to create a powerful chatbot at a fraction of the cost of industry leaders in the US.
Stocks in the AI value chain, including power generation companies, were punished with declines that exceeded 20% in some high-beta names. Shares of the poster child of the AI narrative, the chip-maker Nvidia, fell by 17% and saw its market capitalisation melt away by almost $600 billion, which is now the all-time record for a publicly listed company in terms of market-cap loss in a single day. On the other hand, some of Nvidia's main customers saw their shares rise slightly.
FAR-REACHING IMPLICATIONS
In our opinion, the implications of DeepSeek's models could be far-reaching, with potential impact on hyperscalers' capital expenditures and efficiencies, but also on geopolitics as it originated from China. According to DeepSeek, its approach offers advancements in performance, cost-effectiveness, and an open-source, technologically advanced framework.
The launch of its models also coincides with efforts to promote US leadership in AI, highlighting demand concerns for specialised hardware.
One can already draw two lessons from the DeepSeek revelation. First, we are clearly living in an era of accelerating scientific and technological progress and more than ever, an astute investor must learn to work with probabilities as there are no absolute certainties. IT companies are vulnerable to innovations that challenge their competitive advantage from one day to the next.
The Trump administration's announcement of the Stargate AI initiative should have resulted in another leap forward for the US domestic sector, only for the euphoria to be upended by an unknown Chinese startup.
Second, China's capacity for innovation should not be underestimated -- it trains more than 4 million engineers every year, compared with just 500,000 in the US.
For now, we remain constructive on cloud computing and AI themes, as we believe it is too early to assess if DeepSeek is a true game-changer. Its impact has yet to be tested on congested networks, and social media's doomsday scenarios on hardware seem far-fetched. It will take more time to see the fundamental consequences for the evolution of the AI ecosystem.
On the other hand, the US government announced broad-based tariffs on Canada, Mexico and China before the market had time to fully digest the impact of DeepSeek. Our views on trade tariffs remain unchanged -- they are largely a negotiating tactic to reach other goals, namely tackling immigration and the opioid crisis.
NEGOTIATING TOOL
The newly appointed US Treasury Secretary, Scott Bessent, has publicly stated tariffs are counterproductive and merely a means of exerting pressure. Tariffs are likely to have less of an inflationary impact in the US than a deflationary impact elsewhere.
The US consumer is too dominant, and foreign manufacturers are likely to grant concessions on prices to keep market share. A stronger US dollar also helps to buffer any potential trade tariff impact.
In the previous column, we opined that equity markets will likely experience higher volatility in 2025 after two years of strong equity returns. These episodes reinforce one of our core investment theses for 2025: the risks of short-term market corrections are higher, though we recommend against rushing into major portfolio changes.
Earnings breadth is likely to expand, which typically coincides with a more diversified market performance. This, in turn, underscores the importance of allocating assets beyond US large-cap IT stocks. We see the best opportunities in cyclically oriented segments, where the recent rotation appears to be gaining traction. In the cyclical space, our preferences lean towards industrials, financials and US mid-caps.
Kean Tan is Managing Director, Senior Advisor and Head of Investment Solutions at SCB-Julius Baer Securities Co Ltd in Bangkok.
