On February 7th, Schroders Global Investments released a statement saying that, according to reports, the Chinese artificial intelligence startup DeepSeek has developed a large language model (LLM) that rivals market leaders, but at a much lower cost. If true, this could mean a reduction in the demand for high-performance semiconductors needed for AI-related computational workloads. However, this conclusion depends on whether DeepSeek's cost data truly compares with existing industry peers and whether other factors remain constant. An increase in computational efficiency does not necessarily mean a decrease in semiconductor demand. Jevon's Paradox is a well-established economic concept that states that an improvement in resource efficiency often leads to an increase in the consumption of that resource. In this case, higher computational efficiency may stimulate further adoption and development of AI, potentially offsetting any direct decrease in semiconductor demand. Schroders Global Investments pointed out that if an increase in computational efficiency leads to a decrease in semiconductor/AI device demand, companies like Nvidia (NVDA.US) and other computational infrastructure suppliers may face pressure. However, this scenario is not certain, especially considering the mentioned Jevon's Paradox. On the other hand, this development could benefit software companies. Lower AI costs may allow more customers who were previously deterred by high prices to use these technologies. For software suppliers embedding AI functionality in their products, this could promote adoption while maintaining profitability. Furthermore, mega-cap companies like Microsoft (MSFT.US), Meta (META.US), and Google (GOOGL.US) could also benefit. Can these companies recoup their investments in AI technology after significant research and development expenditures? Market concerns about this continue to grow. If this situation leads to reduced expenses for these companies, their capital expenditures may decrease, leading to a significant increase in free cash flow. Schroders Global Investments believe that there are still many uncertainties, such as clarifying DeepSeek's cost structure and whether cheaper infrastructure will actually reduce the expenses required in the global AI competition. This uncertainty does pose risks, but it is likely to create opportunities for active investors, especially in the technology and industrial sectors. The full impact of DeepSeek's technology remains to be seen. However, this case highlights how the market can be easily influenced by missteps from large American companies or the emergence of new competitors. Major stock indices may not provide the diversification benefits they once did. Investors looking to build more resilient portfolios must take a proactive approach and seek investment opportunities across industries and regions for true diversification.