The unexpected appearance of DeepSeek’s open-source Chinese AI model is shaking up Silicon Valley and global markets. Launched on 20 January, it offers comparable performance to the OpenAI flagship at a fraction of the cost, using less advanced microchips.

The launch of DeepSeek’s new model, just as Donald Trump was sworn in as US president on January 20, was no coincidence. China was sending the message that, despite Washington’s restrictions on the export of next-generation microchips to curb the technological development and competitiveness of Chinese companies, Beijing can be the world leader in AI.

A day later, Masayoshi Son, CEO of SoftBank, Sam Altman, CEO of OpenAI, and Larry Ellison, chairman of Oracle joined the new Trump administration as it announced a private sector investment of up to $500 billion to fund infrastructure for artificial intelligence to outpace rival nations in this business-critical technology.

Impact on global stock markets

The White House announcement did not prevent the stock market plunge of companies such as Nvidia, one of the pillars of the technological revolution generated by AI and the main manufacturer of microchips in the supply of hardware for AI, which saw its shares fall by more than 15%, losing 589 billion dollars in a single day.

It was the biggest loss of value by a stock in the history of the stock market and relegated Nvidia to third place in the ranking, behind Apple and Microsoft.

Nvidia’s fall dragged down other technology giants and several companies linked to chips and semiconductors: Advanced Micro Devices (AMD -10.38%), Marvell (-19.11%), Broadcom (-17.40%) and Grail (-9.16%), Microsoft (-2.17). This turmoil also affected European stock markets, with significant falls in companies such as ASML (-8%), Schneider Electric (-10%) and Siemens (-5%).

Investors were concerned that DeepSeek’s innovative approach would lead to a collapse in demand for graphics processors and other components in data centres, which are essential for the development of AI.

A disruptive business model

One of the keys to the success of the Chinese AI model is the demonstration that the industry is capable of developing efficient and practical artificial intelligence without the need to rely on Nvidia’s advanced and expensive microchips. Thanks to Washington’s restrictions, the company was forced to use H800 model microchips, a less powerful version of those available to US companies.

In addition, DeepSeek claims that with its new technologies, it can reduce algorithm training costs by 75%, noting that it has invested only $7 million in developing its model, compared to the $50 million OpenAI has spent on ChatGPT. As a result, it can offer access to its interface for just $0.14 per million input tokens, far less than the $15 OpenAI charges for the same service.

The other pillar of DeepSeek’s success is its open-source approach, which allows any developer to modify and adapt the software to suit their needs. A business model that contrasts with the more restrictive uses that Western companies impose on their AI applications. It is therefore not surprising that, on Wednesday morning, DeepSeek became the most popular application overall, not just for AI, in the Apple and Google app shops.

DeepSeek’s breakthrough is proving that US technological hegemony in AI is no longer guaranteed. There is little doubt that sooner or later Washington will also impose bans and restrictions on this new Chinese competitor under the pretext of national security, as it has previously done with ZTE, Huawei and TikTok, but, ironically, the increasing aggressiveness of the US administration in the geopolitical framework to prevent the collapse of the empire is accelerating the development of technological, commercial and financial alternatives by other global players.

If you want to discover the best option to protect your savings, enter Preciosos 11Onze. We will help you buy at the best price the safe-haven asset par excellence: physical gold.