China's newly launched artificial intelligence (AI) startup DeepSeek has overwhelmed ChatGPT, causing a stir in the world's technology community and possibly causing the US tech stock bubble to burst.
China's AI Blockbuster
A young and modest-sized (less than 200 employees) Chinese artificial intelligence (AI) company has just shocked the world technology industry after announcing breakthroughs in AI technology, which could open a new era in the global AI industry.
This has forced American tech giants like OpenAI and Google to rethink their development strategies. Instead of investing in hardware to buy tens of thousands of expensive super-fast chips from Nvidia or building AI factories, American corporations may have to turn to algorithm optimization to reduce costs.
The war in the technology field in general, in the AI field in particular, and between the US and China may be more fierce than ever after DeepSeek appeared and caused a storm at the beginning of the new year 2025.
Although it was only established in May 2023 in Hangzhou (China), DeepSeek has created AI models with outstanding features and is considered to be superior to ChatGPT - the AI platform that has been making waves in the technology market for the past 2 years, or recently Claude of Anthropic.
DeepSeek's technological breakthrough lies in its extremely low development cost, only a fraction of the super expensive AI models of American giants. DeepSeek does not need to use overly powerful chip complexes while still being superior in efficiency.
Some experts even after trying DeepSeek said ChatGPT was like “child's play”.
US tech stocks plunge: Will the bubble burst?
In the trading session on January 27, after China announced a low-cost AI model that caused turmoil in the global market, US technology stocks were sold off heavily.
Nvidia shares fell nearly 17%, wiping out $450 billion in market value.
The factor that plunged tech stocks into the red was Chinese startup DeepSeek, which is raising concerns about competitiveness in the AI field and the US's leading position may be shaken.
Besides Nivdia's plunge, other AI stocks in the US also fell sharply. Chipmaker Broadcom's shares fell 17%. AMD fell 6.4%. Other AI stocks in the US also fell, such as Micron, which lost 8%, Arm Holdings, which lost 9%.
AI-related energy construction stocks such as Vistra and Constellation Energy fell 28% and 21% respectively.
At the end of the session on January 27 (early morning of January 28, Vietnam time), the Nasdaq Composite technology index fell nearly 3.1%. Other indexes also fell sharply. Investors are worried about the AI stock bubble bursting because of the appearance of Chinese startup DeepSeek.
Thus, the risk of the "AI bubble" in the US bursting is greater than ever, but not because of the decline of this trend, but because the stock price has gone too high, many times in the past year and strong competition from China.
China’s low-cost AI model has raised questions among global investors about the high valuations of American tech giants. Many fear the tech market could face a bubble burst, as witnessed in the dot.com bubble of the early 2000s. Capital could quickly flee the AI sector.
Earlier this week, DeepSeek’s new AI model took the top spot on Apple’s App Store charts. The Chinese AI startup beat ChatGPT right on US soil.
On Bloomberg, Vey-Sern Ling from Union Bancaire Privee said that DeepSeek's model could turn the investment thesis for the entire AI supply chain upside down.
Investors also believe that AI is no longer just for a few tech giants, but can be a playground for small startups. Open source will make this game more exciting than ever. And this is an unexpected pressure at the beginning of the new year for American tech giants.
While US tech stocks fell, Hong Kong's Hang Seng Tech Index rose as much as 2% ahead of the Lunar New Year holiday.
With DeepSeek's breakthrough, the US-China tech war will become more intense than ever. The belief that China's AI technology is lagging behind the US is shaken.
The US has been working hard to keep the most advanced chips like Nvidia’s out of the hands of China. But now that’s not so important.
The cryptocurrency market also fluctuated after the appearance of DeepSeek-R1. Bitcoin immediately fell 12%. Many other currencies fell 10-20%.
While today's leading AI models like ChatGPT or its competitor Claude... have to spend hundreds of millions of dollars to build data centers and AI factories for computation, DeepSeek only needs a few million dollars. DeepSeek-V3 is a product of DeepSeek. It is a large language model that even beats GPT-4 and Claude on many tasks. DeepSeek-V3 only needs 2,000 Nvidia GPUs to process the training data, compared to 10,000 GPUs that ChatGPT needed. On January 20, DeepSeek launched a newer model, DeepSeek-R1, which applies the chain of thought method, allowing it to continuously reevaluate its logic while working, thereby helping it solve more complex tasks with high accuracy. Initial evaluations show that Deep Seek-V3 is comparable to OpenAI's GPT-4 and Anthropic's Claude 3.5 (founded by former OpenAI experts and funded by many giants such as Google, Salesforce...). DeepSeek-V3 outperforms many models from other giants such as Meta... In addition to its ability to recognize phrases (rather than individual words) for speed, DeepSeek-V3 is also smart in that it can call on the right “expert” for each specific task and can run on a regular gaming GPU, rather than data center hardware. DeepSeek is built on open source code and does not need a multi-billion dollar data center. What DeepSeek has done is truly terrifying for the American chip giant Nvidia. For the past few years, Nvidia has been a near-monopoly in the GPU chip field, with extremely expensive selling prices and huge profit margins. |
Source: https://vietnamnet.vn/vi-sao-deepseek-cua-trung-quoc-chan-dong-toan-cau-bong-bong-my-co-no-tung-2367138.html
Comment (0)