The report comes from the “Indian internal” news communication of CNBC this week, which brings you the timely news and market reviews of the emerging power of the emerging power and the timely prompting company behind its rapid rise. Do you like what you see? You can subscribe to here.
Big story
This week, on my three-year-old laptop published by CNBC, I run a powerful artificial intelligence model, namely Deepseek-R1.
The new model developed by the Chinese AI Laboratory DeepSeek and released last week has triggered a large -scale decline in US technology stocks. AI models with fierce competition and shocking cost -effective AI models have caused investors to question the billions of dollars that large technology companies are currently spent.
Although I can only satisfy my curiosity about DeepSeek-R1’s short exercise, millions of others in the world may keep in mind a more productive goal.
Deepseek runs on Ganesh’s laptop computer
In the experiment of Deepseek, it may be Indian technology companies, which will provide customers with a strong reasoning AI model for the first time. This model is trained and custody of training and custody without relying on large technology companies.
China’s Deepseek will release its model for free for commercial purposes, and it is also Public technical knowledge From scratch, build such a model. The company said it only spent $ 6 million AI chip cost to develop the model.
Though I have raised questions about exact numbersIt spent hundreds of millions of dollars (sometimes even billions of dollars).
The development may marked the beginning of the development of AI models in India, because the previous methods used to train the big language model requires thousands of energy -intensive and expensive AI chips.
It may also marked Indian technology companies (for example InfosysI used to rely on the AI ​​model created by American technology companies, such as YuanCall up.
Keshav Murugesh, CEO of Business Process Management Company WNSDeepseek’s AI model is the “key progress” of Indian technology companies.
He suggested that the cost of low development costs will enable the new AI model to receive regional language training in India and make the cases that were not thought of economy before. Today, most complex large language models, such as Openai’s O1 and Anthropic Claude Sonnet 3.5, can only output text in a few languages.
Murugesh told CNBC: “By using the innovation behind Deepseek, these companies can greatly reduce costs and accelerate their listing time.” WNSThe New York Stock Exchange, which was revealed in the third quarter of this month, will be listed on the New York Stock Exchange in its third quarter. Use case.
Industry surveys show that the high cost of data privacy and the implementation of large language models is one of the reasons why companies resist AI. If it is confirmed, the benefits of DeepSeek-R1 will quickly eliminate two of the top ten problems and may begin to solve many other problems.
The Indian government also began to subsidize access to AI chips (called graphics processing units), so that the country’s academic and startups began to develop AI models.
Indian IT service companies can also benefit from Deekseek more indirectly. Analysts predict that large -scale corporate customers of IT will redirect budget from AI expenditure to other IT -related expenditures.
“In our opinion, DeepSeek is positive for Indian IT services.” “Although it is not too early to be fully evaluated, it is too early, but it involves lower cost more than the influence of the influence of the influence of the influence of India. Fast AI development should help release IT budgets, which leads to an increase in expenditure in other fields.
However, this may not be a good news. Analysts of Indian agent Anand Rathi said that domestic companies’ “large number of exposure” of cloud computing giants Microsoft,,,,, Amazon,,,,, Google and Oracle With the low -cost AI model of Deepseek, “may face short -term backwinds.”
Bank of America analysts also warned that the entry of service providers was low, although these risks are currently in a “very preliminary stage.”
Therefore, in this world of globalization and contact, China’s breakthroughs may benefit competitors India. However, Deepseek’s R1 may be much earlier than many people expect.
AT & TDue to AI, a US telecommunications operator with more than 150 million subscribers has more than 150 million subscribers, so it has increased the number of calls it received within one year by 30 %. The company also improves efficiency in software development.
AT & T CEO John Stankey said in a phone call with analysts: “We now spend a new code and get more code.” “This is through application AI and technology. Application and what we can use for generating AI. “” “.”
Although the use of the AI ​​model developed by the United States is expensive, it is cheaper than the US high salary.
Taking into account growth Focus on unemploymentThe Indian government will eventually hope that the cheap AI model will not replace low -cost work in the end.
– CNBC’s Michael Bloom contributed a report. If you want to know, Deepseek-R1 is not.
Need to know
Reduce deficit or increase expenditure? Indian Finance Minister faced The difficult choice of making annual budgetIt will be unveiled on February 1. CNBC’s Lim Hui Jie and Anniek Bao report Nirmala Sitharaman challenges, because the budget is a weak domestic demand and domestic demand in the context of the fifth largest economy in the world. This is a budget. The depreciation of rupees and global uncertainty rises.
Foreign investors are running away from India’s stock market. However, analysts say this is a kind of “health correction.” India’s benchmark stock index Nifty 50 Sensex and Sensex wandered at a low point of more than seven months, and have been firmly in the field of correction since September. A little bit Live valuation, others said that profit reservations For the reason for slides.
India is a “composite machine”. During this period, Indian stock retreat, investors’ interest in India gradually weakened at the end of 2024. However, an investment group manager still lies in the country. He said that the decline in the market is a buying opportunity, and Name three Indian stocks 2025 (((((((2025Applicable to dating households.
What happened in the market?
After Indian stocks were extended, Indian stocks resumed some signs. this Nifty 50 So far this year, the index has increased by 0.7 %. However, this year’s benchmark test is 1.7 %.
This week, the targeted 10 -year Indian government bond yield fell to 6.67 % with its global peers.
On the CNBC TV this week, Neelkanth Mishra, the chief economist of Axis Bank, said that India’s Rupee has been “too” stable in the past two years due to the Indian reserve banking policy. Mishra recommends rupee The duration exceeds the necessary As a result, the currency weakened the lowest historical low to the US dollar.
At the same time, emerging markets are the “bridge” of trading with China US President Donald Trump will be hitFergus Argyle, the investment portfolio manager of EFG’s new capital. Argyle also explained why his emerging market funds are not invested in China and Vietnam. He said he prefers markets such as Mexico and India, which is a form and strategy of his China.
What happened next week?
In addition to India’s budget, the inflation rate of the United States, the euro area and Tokyo will focus on the next week, and Big Tech reports the revenue. Dr. Agarwal of Ophthalmology Hospital listed on Wednesday.
January 31: American Personal Consumption Expenditure Index December December,,,,, Tokyo Consumer Price Index in January, ExxonMobil income
February 1: Indian budget
February 3: The consumer price index of the euro zone in January, HSBC in HSBC in India JanuaryAmerican ISM manufacturing PMI January, China CAIXIN ManuoFacturing PMI January, Bank of Japan’s opinion Abstract
February 4: Letter income
February 5: Dr. Agarwal’s healthcare IPO, Indian HSBC Service PMI January, American ISM Service PMI January, China CAIXIN Services PMI January January
February 6: Bank of England’s interest rate decision, Amazon income
