Nvidia's oversights

You are at the top, practically untouchable, you underestimate the competition and announce big plans for the future, and then in a few days it all comes crashing down. As much as it may associate you with some local events, the fate of American companies Nvidia and OpenAI is discussed here in the past few days. And that also means the fate of the world.

The reason is the appearance of AI models R1 and R1 Zero of the Chinese company DeepSeek. On January 20th, the company shocked the world by announcing these two models, which are in no way behind the Chat GPT model of the company OpenAI, the undisputed leader in the previous two years. It is not unusual for China to produce a high-tech product of superior quality, but it was unexpected at this time.

Assessing the level of development that China has reached, many experts spoke of a delay of three to four years, the biggest optimists (or pessimists, depending on the position) believed that the delay was not less than six to nine months.

That lag is considerable in the world of artificial intelligence, where the situation has been changing at a dramatic rate in recent years. But now it is already being talked about that China is not lagging behind, but is leading the way in many things. Only that is not the main news either.

Recognizing China as the most serious competitor in dominating the world market, and therefore a competitor for the global leader in everything, the USA introduced sanctions on the export of top technology from the USA to China. Since the American company Nvidia is the world leader in the production of graphics cards, which are crucial for the successful development of artificial intelligence, it was believed that Chinese companies have too great a handicap in this regard and will lag behind for years.

DeepSeek surprised them with its creativity. Faced with the aforementioned handicap, their engineers had to optimize the algorithms so that they required less processing power and therefore less resources. Their model uses the so-called reinforced learning (reinforced learning) which means that through the system of designed "rewards" and "punishment" system, he largely learns by himself. Therefore, it can develop faster than a system that requires more intensive control of living people. The disadvantage of this approach is more frequent errors in the early stages of development, which is compensated by acceleration when the system "grows up" a bit.

And here we come to the main news. On Monday, the shares of technology companies in the US stumbled quite a bit, the stock market loss is about a trillion dollars (one thousand billion), and more than half of that amount goes to the company Nvidia, whose shares fell by about 15 percent, and from the place of the most valuable company in the world, it immediately fell to third place, behind Apple and Microsoft. The reason is the assessment that artificial intelligence systems can now be developed with significantly less resources. If DeepSeek is to be believed, they developed the R1 for a paltry $6 million with less than two hundred engineers and in less than two years. Without Nvidia processors. OpenAI has been around for ten years, thousands of engineers work in it, and billions of dollars are spent on development. Thanks to this, the Chinese AI model provides its services dramatically cheaper.

The Americans say that DeepSeek's claims should be taken with a deep grain of salt, they believe that the Chinese state's assistance was significantly greater, as well as the number of experts involved, so we are left to choose who to believe.

It is interesting that the American and Chinese models of AI differ in that the Americans emphasize the ethics of artificial intelligence development, while the Chinese focus on the protection of user privacy. This should not confuse you because the Chinese see the main market in companies that are more important to protect their data than to care about the interests of society. Also, the Chinese are developing in their own, large market unencumbered by different laws, unlike the Americans who adapt to the systems of all countries where they work, primarily the USA and the EU.

A big game of ping-pong begins, it is known in whose hands are the rackets, and Europe and Russia will serve as tables, until they come up with a better role for themselves.