Deepseek sets the stage of fanatic income from Meta, Microsoft, Tesla | Real Time Headlines

News of the strength and efficiency around China roar The science and technology market on Monday and landing in a timely manner have made our technology investors reason to pay special attention to the income season. The season began after the bells of Wednesday.

Deepseek, an artificial intelligence laboratory in Hangzhou, issued Last week, a reasoning model called R1 was better than the latest model of OpenAI in many third -party tests. In late December, Deepseek launched a free open source large language model. This model only took two months to build less than $ 6 million. The startup was established in 2023.

Although Deepseek’s claims have a lot of doubt, the latest statement has caused panic in Wall Street, because the potential impact of obvious cheap AI models is fast instead of late.

NVIDIA’s stock price Stretch 17 % on Monday, since the common popularity in March 2020, its maximum decline. The market value of short selling has disappeared by nearly $ 600 billion, which is the largest losses in the history of American companies. Chip manufacturer’s stock BroadcomThis has also benefited from AI prosperity, and it also decreased by 17 %. and Oracle,,,,, Dell,,,,, Super micro computer and HP Enterprise Everyone saw a sharp decline because they sold servers with NVIDIA graphics processing units.

Yuan,,,,, Microsoft and Tesla All reports are on Wednesday’s quarterly performance, and they are one of the MEGACAP technology companies. It spent billions of dollars on AI infrastructure. The largest check will pay the GPU of NVIDIA. apple The report results on Thursday, followed by letter and Amazon next week.

For Deepseek, the most direct problem for those companies is less than income, and more is related to cost. They tend to spend a lot of time talking about income calls.

“DeepSeek’s training framework before and after training (if FRIM) has a problem with the pure efficiency of the global high standards and whether the government has already invested in a large number of CAPEX in the AI ​​infrastructure. Considering the innovative method of DeepSeek revealed.

Investors hope to get some preliminary answers in the return report this week and next week. Nvidia did not report until late February, and it is expected that Broadcom will announce the results in March.

This is the significance of how much returns next week.

Headquarters executive officer Mark Zuckerberg Last week, he said that his company planned to invest around 60 billion to 65 billion US dollars In the capital expenditure in 2025, it greatly increased AI infrastructure. Zuckerberg discusses the quarterly update of AI related companies and likes to improve progress in open source Camel family LLMS.

In view of how much Meta spent on the NVIDIA GPU, any change in market dynamics is a strong interest in shareholders. Some technology investors are paying attention to AI’s large expenditure for the real -time income of the company.

Analyst of William Blair Ralph Schackart, an analyst at William Blair, told CNBC, Deepsink’s Recent claim It is said that it is said to be cheaper and faster training an open source AI model. “Surprising everyone, but it is difficult to know if this investment is real.”

Meta’s stock price rose almost 2 % on Monday, compared with NVIDIA and other hardware companies, it was very different from Wall Street. The potential of lower cost may be the gospel of the Yuan. However, like LLAMA, DeepSeek is also open source. Therefore, if developers attract another model, there may be a threat of competition.

Shatcart said that investors can be sure that Yuan will continue to invest in AI.

He said: “There are too many benefits or losses without investing.”

In large software and Internet companies, Microsoft is closest to AI prosperity due to the intimate relationship with the ChatGPT Creator Openai. approximately $ 14 billion Microsoft has pumped OPENAI for this startup. The startup needs a ship of the GPU to build and run its model.

Deepseek directly against Openai. exist Post on X Over the evening of Monday, OPENAI CEO Sam Altman said that R1 said “impressive model, especially in terms of price delivery.” He added that OpenAI will provide better models, but “legally owns legal own New competitors “”

In addition to remittance to Openai, Microsoft itself is a large stock when the GPU’s AI products and Azure Yunyun computing services.

Analysts are Morgan Stanley The rise of Deepseek is the active development of the company because there may be lower capital expenditures.

However, OPPENHEIMER analyst Timothy Horan believes that DeepSeek shows a negative sign of Microsoft. He wrote in a report that he wrote in a report that he wrote in a report: “Bettish is more of the proprietary model. (Openai) and higher cost of hardware.

Microsoft said earlier this month that it plans to spend $ 80 billion In fiscal 2025, the construction of data centers that can handle AI workload.

On his social network X, Tesla CEO X Elon Musk (Elon Musk)He also operated SpaceX and expressed his suspicion that DeepSeek’s model was created based on low budgets that are said to be less than 10 million US dollars.

But Musk praised Deepseek’s technical achievements. Musk has history in the space. He was the co -founder of Openai. At that time, it was launched as a non -profit organization in 2015, but now he has participated in enthusiasm Legal struggle In the case of company and Altman, he was suing to prevent Openai from transforming into a profitable entity.

Musk is also the founder of AI startup and Openai competitors, XAI, Raise billions of dollarsMainly to buy NVIDIA GPU.

In Tesla, Musk has been increasing AI spending because the company is committed to building a team of autonomous cars. Musk has talked about a large amount of AI investment in his company, which is a key advantage. He said in an article on X in April that Tesla will spend $ 10 billion In the year of “merger training and reasoning AI”.

UBS analysts wrote in a comment on Monday: “Model training and reasoning are more efficient, which can help progress and eventually spread.” “Autonomous cars. However, they added that the affordable open source model” It may break the narrative of narrative, that is, the more capital is calculated and infrastructure, the greater the possibility of success, thereby reducing the possibility of any “moat”. “

Tesla’s stock price fell 2.3 % on Monday.

Apple is Reduce As far as capital expenditure is concerned, its competitors are compared with competitors. In the company’s latest fiscal year, it spent $ 9.5 billion in capital expenditure, 14 % less than in 2023.

Unlike some technical competitors, Apple will not run cloud services for rental computing power. So far, the company’s large AI betting involves Apple Intelligence, which is a software kit for iPhone and Mac. It can summarize texts and generate images.

Apple’s kit can run by itself, whether on the user’s device or on the company’s server. In order to train its model, Apple published a study, saying that it uses Google’s exclusive tensor processing unit for Apple Intelligence’s initial model (TPU)More limited chip than GPU, but can handle some AI tasks.

The Apple and OpenAI contract integrates ChatGPT to its Siri Voice Assistant, which is supported by the expensive NVIDIA GPU. After DeepSeek development, analysts may be interested in topics. However, although the two companies did not publicly confirm their transactions with their partnerships, analysts expressed their suspect that Apple was paying for ChatGPT.

watch: Deepseek has made a breakthrough of AI