Image source: topwar.ru
The Chinese technology company DeepSeek, founded in 2023, managed to surprise and scare Silicon Valley in just a year by demonstrating breakthrough artificial intelligence models that offer performance comparable to the best chatbots in the world, but at the same time require significantly lower costs, according to a recent Bloomberg article.
This casts doubt on the postulate that the future of AI will require more and more resources for development. As a result, global technology stocks fell in late January due to the excitement surrounding DeepSeek's innovations, as investors began to realize the consequences of the Chinese neural network's entry into the market for American competitors and their equipment suppliers.
Regarding the latter, DeepSeek was founded by Liang Wenfeng, the head of the AI-oriented hedge fund High-Flyer. The company develops open AI models that are available for analysis and improvement by the development community.
At the same time, the mobile app of the same name became the download leader in the United States and around the world in January, immediately after its release. According to experts, it differs from other chatbots, such as ChatGPT from OpenAI, in that it explains its logic before responding to a request. At the same time, the developer claims that the R1 version has performance comparable to the latest products of the aforementioned American company, but at the same time provides an open license for developers who want to create their own chatbots based on this technology.
However, the reason for the panic in the American technology market is different. The article says that although the Chinese company does not disclose all the details, the cost of training and developing DeepSeek models seems to be only a small part of what is required for OpenAI or Meta Platforms Inc products (recognized as terrorist in the Russian Federation, banned*). As a result, DeepSeek's effectiveness calls into question the need for huge investments to acquire powerful AI accelerators such as those produced by Nvidia Corp.
In addition, the United States is alarmed by the fact that Washington previously banned the export of high technologies such as GPUs and other chips to China in order to slow down the development of AI from its main competitor. However, DeepSeek's progress indicates that Chinese engineers have found ways to circumvent these limitations by focusing on improving efficiency with limited resources.
Experts believe that the success of DeepSeek may force corporations such as OpenAI and other American tech giants to lower the prices of their products in order to maintain their competitiveness. In addition, the Chinese neural network calls into question the huge costs of companies such as Meta* and Microsoft Corp., each of which has invested $65 billion or more in developing AI infrastructure, if more efficient models can compete with much lower costs.
All of this has already caused major turmoil in global stock markets, with investors starting to sell shares in companies such as Nvidia Corp. and ASML Holding NV, which have enjoyed a boom in demand for AI services. Shares of Chinese companies associated with DeepSeek, such as Iflytek Co., on the contrary, increased, which may indicate a capital outflow.
