It’s difficult to discuss the economic history of the 21st century without mentioning the “China shock,” which refers to China’s entry into the global market. This development resulted in an influx of inexpensive goods for wealthy nations but also left numerous industries and workers sidelined.
Now, a relatively obscure Chinese hedge fund named DeepSeek is attempting to make waves in the artificial intelligence sector. It has introduced a large language model that rivals the leading competitors, such as OpenAI’s models, at a significantly lower cost. Unlike OpenAI, which treats its technology as proprietary, DeepSeek’s R1 transparently displays its underlying architecture, making it appealing for developers to utilize and expand upon.
The rapid pace of change in the AI industry is remarkable. Recently, the five largest tech stocks heavily invested in AI—Nvidia, Alphabet, Amazon, Microsoft, and Meta Platforms—experienced a combined loss of nearly $750 billion in market value. This situation may be particularly challenging for Nvidia if DeepSeek has succeeded without depending on its most advanced chips.
Investors in technology companies and related sectors, including European chip manufacturers and energy firms expected to benefit from data center expansion, are left uncertain about the future of their investments. These major tech firms had intended to invest around $300 billion in capital expenditures this year. Analysts predict that when Meta and Microsoft release their earnings, they will announce a total investment of $94 billion for 2024.
Despite the unsettling landscape, the situation is still evolving. DeepSeek has yet to realize its full potential and has not achieved “artificial general intelligence,” the coveted human-like capability that companies like Meta and OpenAI are striving to reach. However, the landscape may have shifted. DeepSeek could potentially attract some customers from major U.S. firms and has raised questions about the belief that enhanced hardware is essential for superior AI performance, a belief that has driven investments in Silicon Valley.
What may be detrimental for the largest tech companies could also be advantageous for others. Many businesses prioritize the availability of reliable and sufficiently effective models over possessing the absolute best technology. Not every application requires the highest performing model. Advancements in reasoning capabilities such as those represented by R1 could represent a significant improvement for automated systems that interact with customers and streamline workplace tasks. If such technologies become available at lower costs, corporate profits could increase as a result.
In this sense, the situation might parallel the initial China shock. It may not only lead to disruption but also initiate a significant reshuffle within the industry, albeit with challenges for various stakeholders. Some researchers have found that for each job lost due to the earlier China shock, the purchasing power of U.S. households increased by over $400,000. While the race for AI dominance might currently be on hold, an opportunity for significant changes appears to be underway.
photo credit: www.ft.com
