Chinese AI startup DeepSeek’s cheaper technology shakes tech world

#Chinese #startup #DeepSeeks #cheaper #technology #shakes #tech #world

Deep Sak, a new Chinese company, is increasing the less expensive AI technology of Deep Sak, billions of investors, increasing that the US is offering a big tech for its construction, as analysts have predicted by this technology industry. The leaders of this week will overcome the expected results.

With the help of low sophisticated chips, NVIDIA’s H800, Deepesic said it had created a million Million within two months AI model for less than 6 million. On Monday, the most downloaded iPhone in the United States was run by the App V3 model.

Startup, founded in 2023, said its AI model is either similar to the cost of cost or improve high -American rivals, challenging the theory that wide computing power and capital for Skylling AI Need work.

The need for such a business has increased the “Magnificant Seven” companies’ market value in November 2022 after kicking AI Boom by about $ 10 trillion.

“Did Dupic really built an open in 5 million? Of course not,” Burnstein’s analyst Stsi Rasgon said. “It seems like a tension to think that the innovations deployed by Dipic are completely unknown by researchers at the top Terra AI in many other AI labs in the world.”

He said that the competition of DPCAC prices blows anything from the competition. Shares of AI Chip Panier Navidia fell 16, Microsoft 3.8 % and TSMC’s US stock declined by 14 %.

Rasgon and other analysts say DiPsic training costs may be higher for the V3 model because the amount of about $ 6 million offered by Startup includes only the amount spent on computing power, while more More commonly known about the costs of making R1 model.

Nevertheless, it estimates $ 250 billion analysts that large US cloud companies will spend this year on AI infrastructure. Investors worried about the slowdown last year have been questioned last year.

Most US -tech companies are ready to report the results this week and next week, with analysts and investors expected to provide further explanation on their strategy.

“(The rise of DPCC) raises the question of whether the current pace of Capix Expenditure/Technology upgrades is necessary, CFRA analyst Angelo Zeno said. The interpretation of US hypersonallers will prove to be key this week whether they are AI The costs will be aggressive, “said CFRA analyst Angelo Zeno.

Zeno added, “They will potentially emphasize the need for maximum computing power when we move towards Agent AI and Physical AI,” Zeno added, referring to independent AI agents. They need very little human intervention for normal tasks as well as robots and self -run cars.

Fallpack

Although the cost of using AI models is falling with rising competition and progress in this technology, Burnstein’s Rasgon said that Dipic stands because it has cost 40 times less than the open model of the open.

Analysts said it could start a price battle for AI services, possibly pressing on tech companies such as Open AI, which caused the high operational costs of running services like Chat GPT. Losing billions of dollars years ago.

“If the adoption of a depester intensifies, it can start a reduction in prices from rivals of similar open source products,” said Gudju Ciula, a senior analyst at Emirate.

“Market leaders like Open (are pushing for profit) are unlikely to decrease in short -term prices. They will potentially double confidence and safet Having. “

Some experts also suspect that the US business will be willing to embrace Chinese AI technology in view of the confidentiality and security of China and the US tensions and data.

DiPsic says it secures user information in China’s servers, which can be an important point in adopting Americans.

However, some investors believe that US tech companies will emphasize the progress of Deep Sak and that cheap AI services are obliged to increase the adoption of technology, which can eliminate the demand for chips.

Mark Malik, Chief Investment Officer of Seburt NXT, said, “Did Dipisic find and searched the more efficient processing model for AI?

“(This) will only make the AI ​​opportunities bigger in the future.”