Depseek has new plans at the door: the Startup China accelerates the launch of its new R2 artificial intelligence model. Last month, its R1 predecessor shocked world markets, causing losses of more than one billion dollars to its rivals.
Investors discovered that the Chinese company almost unknown had created a system capable of competing with American giants, but at drastically lower costs. As reported to Reuters, nearby sources, the company based in the city of Hangzhou decided to take advantage of its position to advance its proposed model for May. The reason is clear: to consolidate the competitive advantage obtained with the R1.
The accelerated development of R2 represents one more element of tension in the complex technological relationship between China and the US, in what has turned out to be an increasingly strategic sector for world security and economy: artificial intelligence. Deepseek’s goal is to improve its system capabilities with advanced functionalities, with a greater multilingual programming and reasoning capable capable of overcoming English limitations, all without requiring high investments.
Hugging the Chinese state
The acceleration of R2’s development exposes how Deepseek has quickly become a strategic asset for Beijing in world technological competition. The transformation is amazing taking into account that in Chinese Government he saw with suspic Clusters Supercomputing with 10,000 Nvidia A100 chips. The calculation potential concentration raised doubts among the regulators, although they decided not to interfere.
The decision was timely in 2022, when USA imposed tariffs on the export of chips A100 to China. With Fire-Flyer II in operations, Depseek found himself in a privileged position: he had one of the few advanced computer infrastructures in the country capable of developing state-of-the-art AI models, becoming a strategic mechanism in world technological competition.
Although initially discreet, government support for Deepseek gave the green light for the company to introduce the tool in Beijing. The support resulted in concrete actions: up to 13 municipal administrations and 10 state energy companies adopted the company’s systems. At the same time, national technological giants: Lenovo, Baidu and Tencent, have integrated Depseek solutions on their own platforms, further expanding their deployment. The official consecration arrived in January, when Chinese prime minister Li Qiang chose Liang Wenfeng, founder of Deepseek, as a representative of the entire artificial intelligence sector for a meeting in Beijing, dispensed with the leaders of much more consolidated companies . However, the Great Dragon shelter was accompanied by precise directives; Chinese authorities ordered Deepseek to maintain a low media profile to avoid attracting excessive international attention that can lead to new restrictions.
The secret of Deepseek’s recipe
While technological giants like Baidu and Tencent rushed to launch artificial intelligence applications for the general public, Depseek chose a completely different path. The company prioritized the quality of its systems rather than in rapid marketingfocusing everything on efficiency and cost reduction. An unconventional strategy that turned out to be a resounding success.
The result of this vision is evident: Depseek created artificial intelligence systems that offer a performance comparable to that of Westerners, but at prices up to 40 times lower. At the beginning of February, the cost of using Depseek services was a minimal fraction of what Openai charged for an equivalent function. A difference so marked that he even forced the most established competitors to review their price lists: OpenAI reduced its prices this month, while Google introduced cheaper subscriptions for its Gemini system.
Article originally published in Wired Italy. Adapted by Alondra Flores.
#Depseek #advance #launch #model #challenges #West