L'intelligenza artificiale cinese scuote i mercati globali

Chinese AI developers have intensified pressure on global tech markets, reshaping the competitive landscape and prompting reassessment among industry leaders. In recent weeks, Z.ai and Moonshot AI, both Chinese firms, have launched advanced artificial intelligence models that have drawn widespread attention. Notably, Moonshot AI's Kimi K3 has sparked significant interest, with Wall Street reacting sharply. The move has prompted allegations from U.S. officials that the startup may have leveraged proprietary American technology in its development. Despite these claims, the model has also received praise from OpenAI's leadership, highlighting the growing influence of Chinese AI innovation. The impact of these developments has been evident in financial markets, where volatility has increased. Although the Nasdaq 100 ended the month on a positive note, it briefly dipped into correction territory during the week. Meanwhile, South Korea's Kospi index plummeted by as much as 38 percent in July before partially rebounding. Analysts attribute this turbulence to concerns over the return on investment for the massive sums spent on AI by tech giants. Many believe the rapid rise of affordable Chinese alternatives is undermining confidence in the long-term viability of current strategies. Tony Sycamore, an analyst at IG Markets, noted that the gap between U.S. and Chinese AI capabilities has narrowed significantly. He stated that Chinese AI and infrastructure are now comparable to some of the most advanced U.S. models, creating a new challenge for investors and markets linked to them, such as those in South Korea and Japan. Sycamore emphasized that the effects of this shift are most pronounced in South Korea, though continued declines in the U.S. tech sector could ripple through to Australia, affecting returns on superannuation funds. Australian businesses, especially startups integrating AI into their operations, are feeling the strain of this evolving dynamic. Jacky Koh, co-founder and co-CEO of Sydney-based startup Relevance AI, has observed firsthand how the competitive environment is shifting. His company assists other firms, including Canva, KPMG, and Autodesk, in developing AI agents capable of performing complex tasks autonomously. Until recently, these agents relied heavily on proprietary models from U.S. providers such as OpenAI, Anthropic, and Google. However, Koh reports that the proportion of work utilizing open-source models has risen substantially. Currently, approximately 20 to 25 percent of Relevance AI's workload involves open-weight models, many of which originate from Chinese developers. These models offer businesses the advantage of self-hosting, allowing them to avoid recurring licensing fees while still managing the computational costs of running the software. This approach contrasts sharply with traditional proprietary models, which remain tightly controlled by their creators and often require substantial financial commitments. Cost efficiency is a primary factor driving this transition. As businesses seek to optimize expenses, the availability of high-quality, low-cost models from Chinese developers presents a compelling alternative. With ongoing advancements in AI technology, the balance of power continues to shift, challenging established norms and forcing both domestic and international players to adapt rapidly. The implications extend beyond immediate economic considerations, signaling a broader transformation in how AI is developed, deployed, and integrated into global business practices.