As the debate over TikTok’s future in the United States continues, another Chinese tech innovation—the AI model DeepSeek R1—is challenging assumptions about American dominance in the world of artificial intelligence. DeepSeekrich711, a little-known Chinese AI lab has stunned the global tech community by creating a powerful AI model at a far lesser cost than many experts thought possible.

DeepSeek’s AI assistant app has now reached the top spot in the Apple App Store in the U.S. and other countries. U.S. President Donald Trump said DeepSeek should serve as a “wake-up call for our industries that we need to be laser-focused on competing to win.”

Founded in 2023 by Liang Wenfeng, co-founder of the hedge fund High-Flyer which uses AI to predict market trends, DeepSeek claims its AI can perform tasks similar to those of U.S. tech giants such as OpenAI and Google, but at a fraction of the cost.

Last week, DeepSeek published a research paper stating it spent less than $6 million on computing power to train its model. This cost is a fraction of the multibillion-dollar AI budgets of U.S. tech giants like OpenAI and Google, developers of ChatGPT and Gemini, respectively.

“To see the DeepSeek new model, it’s super impressive in terms of both how they have really effectively done an open-source model that does this inference-time compute and is super-compute efficient,” Microsoft CEO Satya Nadella said at the World Economic Forum in Davos, Switzerland, on Wednesday. “We should take the developments out of China very, very seriously.”

BY Urvashi Sarkar

The achievement by a relatively unknown AI startup becomes even more significant considering the United States has restricted the supply of high-power AI chips to China for years, citing national security concerns.

In their research paper, DeepSeek’s engineers said they had used about 2,000 Nvidia H800 chips, which are less advanced than the most cutting-edge chips, to train their model.

This challenges the core assumption of U.S. AI companies and policymakers that the best AI requires hugely expensive, advanced semiconductors. This belief led the Biden administration to ban U.S. companies from exporting powerful AI chips to China to slow the country’s tech industry. DeepSeek’s research paper suggests either the most advanced chips are not necessary to create high-performing AI models, or Chinese firms can still source chips in sufficient quantities—or both.

On the other hand, China have qualified for their maiden Asian Champions Trophy final. Their journey to the summit clash has been a rollercoaster ride. They started with a defeat to India but bounced back with a 4-2 win over Malaysia. The hosts then faced a 2-3 loss to South Korea and were also routed 5-1 by Pakistan in the pool stage.

The youngster displayed goalkeeping masterclass in the shoot-out, saving all four of Pakistan’s attempts after the match was drawn 1-1 in regulation time.

Domestically, DeepSeek is known for attracting young and talented AI researchers from top universities by offering high salaries and opportunities to work on cutting-edge projects.

Similar to other applications powered by generative AI models, DeepSeek's app allows users to ask questions and get conversational responses. The app’s interface is similar to its competitors like ChatGPT.

Unlike traditional AI training methods reliant on supervised fine-tuning, DeepSeek uses pure reinforcement learning, enabling models to learn through trial and error and self-improve through algorithmic rewards. This has been particularly effective in enhancing the reasoning capabilities of the AI model, called R1. Essentially, the models learn by interacting with their environment and receiving feedback on their actions, similar to how humans learn through experience. This allows them to develop sophisticated reasoning abilities and adapt to new situations more effectively.

Other distinguishing feature is DeepSeek’s open-source approach. Key components are free for anyone to access and modify, although the company has not disclosed the data used for training.

Investors’ confidence in the U.S. tech scene has been shaken, at least in the short term. By Monday, the new AI chatbot triggered a massive sell-off of major tech stocks, as fears mounted over America’s leadership in the sector. Nvidia appears to have suffered the most, as its stock price plunged 17 per cent over the course of the day. It lost nearly $600 billion of its market value on Monday—the biggest one-day loss in U.S. history. The chipmaker, previously the most valuable company in the world by market capitalisation, fell to third place after Apple and Microsoft.

China’s government has, like its U.S. counterpart, made AI a national priority. The country has invested significant amounts of money into AI research and technology, which have poured billions into AI data centres and development over the past two years.

Concerned that China could use AI to match U.S. military and economic power, the Biden administration blocked U.S. chip companies from exporting high-end AI hardware to China. However, Chinese companies have, in some cases, been able to get around these controls by buying AI chips through other countries.

The impressive performance of the DeepSeek model has raised questions about the unintended consequences of the U.S. government’s trade restrictions. Some argue in favour of U.S. trade restrictionsrich711, saying that the measures were only recently implemented and that they will have a greater effect on China’s ability to create AI in the future. Others say these controls have pushed researchers in China to get creative with freely available tools on the internet.