Big Story
This week, on my three-year-old laptop, I ran a powerful artificial intelligence model called DeepSeek-R1, as reported by CNBC.
A new model developed by Chinese AI lab DeepSeek and released last week sent U.S. technology stocks plunging. The highly competitive and potentially shockingly cost-effective AI model is making investors question the billions of dollars currently being spent by big tech companies.
While my brief foray into DeepSeek-R1 satisfied my curiosity, millions of others around the world may have done the same, with more productive goals in mind.
Among the companies experimenting with DeepSec are likely to be Indian technology companies, which for the first time will be able to offer customers a powerful, rational AI model, trained and hosted without relying on big tech companies.
China’s DeepSec has released its model for free for commercial use and has also revealed the technology behind how to build such a model from scratch. The company said it spent just $6 million in AI chip costs to develop the model.
While some have questioned the exact figure, it compares favorably to the millions – and sometimes billions – of dollars spent by American technology companies.
The development could kick-start AI model development in India, as previous methods for training large language models required thousands of energy-intensive and expensive AI chips.
It could also be a major turning point for Indian technology companies like Infosys,
which previously had to rely on AI models built by US-based tech companies like Meta’s Llama.
Keshav Murugesh, chief executive of business transformation firm WNS
, said DeepSeek’s AI model is a “big win” for Indian technology companies.
He suggested that the low development cost will enable new AI models to be trained in India’s regional languages and enable use cases that were previously considered uneconomical. Most state-of-the-art large language models today, such as OpenAI’s o1 and Anthropic’s Cloud Sonnet 3.5, can only output text in a handful of languages.
“By leveraging the innovations behind DeepSeek, these companies can significantly reduce costs and accelerate their time to market,” Murugesh told CNBC. WNS
, which is listed on the NYSE, announced in its third-quarter earnings call earlier this month that it would soon enable generative AI use cases on a top 10 client for the company, a U.S. insurer.
Industry surveys show that data privacy and the high cost of implementing large language models are among the reasons why enterprises have resisted adopting AI. If the benefits of DeepSeek-R1 are confirmed, it would quickly address two of the top 10 concerns and could begin to address many others.
The Indian government has also started subsidizing access to AI chips, known as graphics processing units, to enable the country’s academics and start-ups to develop AI models.
Indian IT services companies could also benefit indirectly from DeepSeek. Analysts expect IT companies’ large enterprise clients to redirect budgets towards other IT-related expenses rather than AI spending.
“In our view, DeepSeek is a positive for Indian IT services,” UBS equity analysts led by Shalin Kumar said in a note to clients. “While it is still too early to fully assess the impact, rapid AI development with lower costs will help free up IT budgets, which in turn will lead to increased IT spending in other areas.”
However, it may not be all good news. Analysts at India-based broker Anand Rathi say that domestic companies with “heavy exposure” to cloud computing giants like Microsoft
, Amazon
, Google
and Oracle
may face “short-term headwinds” as agile companies move away from DeepSeek’s low-cost AI models.
Bank of America analysts also warned that the low barrier to entry for service providers is increasing competition in the sector, although these risks are currently “very early.”
And so — in this globalized and connected world — a breakthrough in China has the potential to benefit rival India. However, DeepSeek’s R1 has the potential to be a double-edged sword much sooner than many expect.
AT&T
, a major US telecom operator with more than 150 million subscribers, has reduced the number of customer support calls it receives by 30% in a single year thanks to AI. The company has also increased efficiency in software development.
“We’re spending less on developing new code internally right now and doing more work,” AT&T Chief Executive John Stankey said on a call with analysts. “And that’s through the use of AI and technology and what we can do with generative AI.”
While implementing using U.S.-developed AI models is expensive, it is cheap compared to high U.S. wages.
Given concerns about rising unemployment, the Indian government will be hoping that cheap AI models will not displace low-cost jobs.
— CNBC’s Michael Bloom contributed reporting. In case you were wondering, DeepSeek-R1 did not.
Need to know
Cut the deficit or increase spending? India's finance minister faces a tough choice as she prepares for the annual budget due on February 1. CNBC's Lim Hui Ji and Annick Bao report on the challenges for Nirmala Sitharaman as the budget comes against a backdrop of slowing growth in the world's fifth-largest economy, weak domestic demand, a weak rupee and rising global uncertainties.
Foreign investors are fleeing India's stock market. Yet analysts say it's a "healthy correction." India's benchmark stock indices Nifty 50 and Sensex are at more than seven-month lows, firmly in correction territory from their September highs. Some point to high valuations, others to profit booking as reasons for the decline.
India is a "growth machine." A slump in Indian equities during that period led to investor interest in India waning in late 2024. However, one portfolio manager remains bullish on the country. He says the market dips are buying opportunities, and names three Indian stocks to buy for 2025. [For subscribers only]
What happened in the market?
Indian stock markets are showing some signs of recovery after a long decline. Nifty 50
The index is up 0.7% so far this year. Yet the benchmark is down 1.7% this year.
Like its global peers, the benchmark 10-year Indian government bond yield fell to 6.67% this week.
What's happening next week?
Besides India's budget, next week will focus on inflation data for the US, euro zone and Tokyo, while Big Tech will report earnings. Dr. Agarwal's Health Care, an eye hospital, lists on Wednesday.
Jan 31: US personal consumption expenditure index for December, Tokyo consumer price index for January, Exxon Mobil earnings
Feb 1: India's budget
Feb 3: Eurozone consumer price index flash for January, India HSBC manufacturing PMI final for January, US ISM manufacturing PMI for January, China Caixin manufacturing PMI for January, Bank of Japan opinion summary
Feb 4: Alphabet earnings
Feb 5: Dr. Agarwal's healthcare IPO, India HSBC services final for January, US ISM services PMI for January, China Caixin services PMI for January
Feb 6: Bank of England interest rate decision, Amazon earnings
1 Comments
While some have questioned the exact figure, it compares favorably to the millions – and sometimes billions – of dollars spent by American technology companies.
ReplyDelete