A small Chinese startup has forced the largest technical companies in America to rethink how to build artificial intelligence.

Dibsic His R1 mode versionL, who, who is compatible or exceeding the capabilities of artificial intelligence systems in the United States, has costs to cost huge sales in technical shares that erase approximately $ 600 billion in the market value of NVIDIA alone.

Shock waves Hit the American technology sector in GUR.With industry leaders in a hurry to analyze how Deepseek achieved these results.

Although there are still open questions, after analyzing the open source code, the consensus, at the present time, is that the Chinese developers are better in building effective models. The fennel of technology was placed in artificial intelligence on their smiling faces and looked at the bright side, while embracing the idea that any progress in artificial intelligence was useful for this industry.

Sam Altman from Openai admitted to perform the impressive model while promising to accelerate the "Better Models" version.

Mark Zuckerberg said from Meta that his company gathered Multiple "War Chamber" Floorized with the engineers of Deepseek technology analysis and a Meta strategy.

Meanwhile, President Donald Trump, who never missed a news course, described Dibsic as a "vigilance" and "positive" development for American technology "because you do not have to spend a lot of money."

The post -deep era

Well, so let's ignore what they say and think about what they will likely do to respond to Deepseek.

It turns out that many of the greats closed to the source are already infiltrating into their DEPSEK methods in their play books-and they will not get the headlines about borrowing from the competition.

For example, really confusion Make up On its search engine, and clear It also made it available for operating it at speed deduction times.

Most of the big names in the American artificial intelligence scene, including Meta, either adapt to Deepseek or think about ways to take advantage of their technology.

With the decline in the initial panic in the market - NVIDIA shares recovered by 9 % today - technology leaders indicate an intuitive economic principle indicating that the penetration of Deepseek efficiency may enhance the demand for artificial intelligence devices.

Known as Javs' paradoxThis concept explains the reason for technological efficiency to expand use rather than reduce consumption.

"Since artificial intelligence becomes more efficient and accessible, we will see its use rises, and turn it into a commodity that we cannot get enough," He said Satia Nadella, CEO of Microsoft, the largest investor in Openai.

Despite the suffering of the most important decrease every one day in the market, NVIDIA sees Deepseek penetration an opportunity.

"The pie has just got much larger, faster. Tweet Jim van, chief researcher in Nafidia on Monday." We, as one humanity, are walking towards Universal Agi soon. "

In other words, if the Jevons' paradox is applied, the Deepseek demonstration can be built high -quality artificial intelligence models with minimal mathematical resources that we will use less graphics processing units in general. Instead, adult men will grow up.

At the other end of the spectrum, as an entry barrier, the increase in developers and new companies will jump to the development of artificial intelligence.

The explosion in total projects is likely to push the demand for the account to unprecedented levels. Of course, for artificial intelligence, not all chips are strict, and the market has decided that Silicon Apple may have a leg on NVIDIA chips in this new world.

For this reason, AAPL increased by 8 % this week, although it was "Apple Intelligence" at the consumer level He mocked as Akomorone.

The mediator is that Apple chips are more efficient in using energy, designed for topical use versus large server farms that use NVIDIA chips, and feature a "uniform memory structure", which means that the CPU and GPU unit and the nervous engine share a quick memory.

This eliminates the need to transfer data between separate ingredients, reduce cumin and increase efficiency of the burdens of artificial intelligence. For models such as Deepseek, which depends on rapid access to complex operations, UMA is supposed to improve performance significantly.

It is clear, in the midst of the innovative dilemma, NVIDIA is unlike Cuda monopolyThe key to operating and developing most of the artificial intelligence models currently available.

Deepseek does not challenge this monopoly - but China is working on it to enhance adoption Huawei ascending A selection of chips.

As, Microsoft does not seem very concerned about changing her business strategy as an infrastructure provider.

However, Openai has applied a simple change to counter users' expectations, while giving users as well as users (who pay $ 20 a month) some of the features that were previously available to professional users (those who pay $ 200 per month) to keep customers.

Another company with a lot of leather in the game is Meta, developer from Lama- The largest and most popular family of Llms Open Source.

Meta has already adhered to investment 65 billion dollars in Amnesty International Infrastructure this year.

Yan Lakhi, the chief of the company's artificial intelligence, looked at the bright side of obtaining Including Through a small startup in China: "For people who see Deepseek and think:" China goes beyond the United States in artificial intelligence. "

"You read this error; the correct reading is:" Open sources models go beyond the royal models. " to publish LinkedIn.

Don't be surprised if Meta adopts Deepseek methods to reinforce Llama-4: "Since their work is published and open source, everyone can benefit from-which is the open search force and open source," Lecun books.

During which Call profits Q4Zuckerberg CEO said that Meta is planning to customize a ten-time computing power to develop Llama-4 from the Llama-3 training resources.

The company may either reduce its spending and apply Deepseek technologies - or maintain spending with the application of these technologies and reach a more powerful model.

The future of artificial intelligence may not depend on the best Amnesty International

Regardless of the wonderful reasoning of Deepseek, in the end, AI still has a Sharia appetite for two things: Power (server farms) and data (for training and learning).

The industry analyst project will demand graphics processing units of 30 % 10x grows In the next five years.

How these costs are transferred to companies and consumers still an open question.

Meanwhile, artificial intelligence models have become open source, such as Deepseek, are so good that people are wondering whether the distinctive prices charged by the royal blog companies are fair.

Who wants to pay $ 20 a month for Openai at the consumer level-we called on a single $ 200 per month for his upscale model-when you can get it for free?

"More companies build open source alternatives to excellent artificial intelligence tools, creating competition that benefits (small and medium -sized companies)." Decipher. "This natural development towards accessible solutions reflects how other technologies have become democratic through market dynamics rather than organization."

For Sirdesai, models such as Deepseek and other open source initiatives are driven forward because they give developers tools to put themselves in markets that appear to be dominated by a few and a few huge companies.

However, it turned out that "decentralized infrastructure and open source development already created competitive alternatives to excellent artificial intelligence tools."

ATUL ARYA, CEO and founder of Blackstraw.ai, who develops artificial intelligence implementation strategies for different companies -NNOTS.

And he told Decipher.

ARYA believes that open source developments such as Deepseek help at the level of the scale and create more fair conditions in the market such as artificial intelligence industry.

He said: "The real driver to reach democracy is the open source community, which is noticed quickly."

Edit Sebastian Senkler and Josh Ketner