Markets Hit 7 Month Low, Longest Losing Streak In 23 Years
Monday was another tough day with the bears fully in control as FII selling that is in the vicinity of $8 billion is continuing on a host of factors
On Episode 492 of The Core Report, financial journalist Govindraj Ethiraj talks to Jaspreet Bindra, founder and MD of Tech Whisperer LTd and Co-founder and CEO of Ai&Beyond as well as Dwarak Narasimhan, Partner at Price Waterhouse & Co, LLP, India
(00:00) The Take
(05:38) Markets hit 7 month low, longest losing streak in 23 years
(08:57) Oil prices fall on fresh Trump call to OPEC to cut them
(10:35) Understanding DeepSeek, where did it suddenly come from?
(19:34) India Inc wants a friendlier corporate tax regime
NOTE: This transcript contains the host's monologue and includes interview transcripts by a machine. Human eyes have gone through the script but there might still be errors in some of the text, so please refer to the audio in case you need to clarify any part. If you want to get in touch regarding any feedback, you can drop us a message on [email protected].
—
Good morning, it's Tuesday, the 28th of January and this is Govindraj Ethiraj, headquartered and broadcasting and streaming like always from Mumbai, India’s financial capital.
The Take
Poor Donald Trump. Barely a week ago, he and his tech bro pals appeared to be on top of the world and in control of it.
They were riding on the back of trillions of dollars of wealth riding in turn on top of trillions of dollars of market capitalisation gain in businesses ranging from ecommerce and search to social networks and electric cars.
If there was any problem that raised its head, it could be swatted down with a threat of tariffs or a stern gaze, as Trump did with a flourish with Colombia and brought the Latin American nation in line to take back illegal immigrants.
Now comes DeepSeek, a free, open-source large-language model launched in late December, apparently developed in just two months in less than $6 million — much lower than what it took for OpenAI in the US.
Last week, the company released a reasoning model that also reportedly outperformed OpenAI’s latest in many third-party tests.
The result: stock prices of almost all tech majors started spiralling downwards on Monday, including AI chip giant NVidia. Losses were expected to cross more than $1 trillion in value.
Why are the stocks crashing ? Because the developments have raised questions about the amount of money big tech companies have been investing in AI models and data centers.
“DeepSeek clearly doesn’t have access to as much compute as U.S. hyperscalers and somehow managed to develop a model that appears highly competitive,” Srini Pajjuri, semiconductor analyst at Raymond James, said in a note Monday reported by CNBC.
As other analysts pointed out, this raises questions about the entire AI supply chain, not just the software but the hardware that powers the software as well.
"On one side: The U.S. and its new Stargate Project, a massive $500 billion pledge with backing from the biggest names in tech to build data centers across the U.S.
On the other hand, a scrappy Chinese startup that’s released an impressive open-source model trained on a shoestring budget, wrote Fortune columnist Nicholas Gordon last week."
In which case what sense would it make for a company like Reliance for example, to invest $20 to $30 billion, Bloomberg estimates, in a data centre in a far corner of the country on the west coast of Gujarat with capacities three times that of all of India’s.
Those questions are yet to be posed because it will take a day or two for the global shock waves to subside.
Meanwhile, there will be questions of how much compute capacity is really required, at what cost and if the Chinese can come up with this killer price, then could others ?
Obviously we don’t know.
Meta CEO Mark Zuckerberg said last week they would invest between $60 billion and $65 billion in capital expenditures on AI in 2025.
He also said he expects Meta AI to “be the leading assistant serving more than 1 billion people” this year and that Meta’s Llama 4 model is expected to “become the leading state of the art model” this year, and that the company plans to “build an AI engineer” that can contribute more code to its research and development efforts.
So what happens now ?
Every tech CEO knows in his or her heart and mind that technology changes can come out of nowhere and overturn business models that seemed solid just a day ago.
Covid kicked off a wave of investment in people and infrastructure at a pace not seen for a while in Silicon Valley as it appeared that our lives were becoming rapidly digital and could stay that way.
A year later, the major tech firms including Meta, Google and Amazon were downsizing furiously, laying off tens of thousands.
And AI happened and stock prices shot up again and once again the Valley was in full control.
The tech bros worked on securing and cementing their future and also indulged in world opinion domination, by standing alongside Donald Trump, who has been only too glad to welcome them.
And then comes DeepSeek.
If Trump and team thought the China trade war would be largely about cheap imports of apparel, toys, furniture and leather, this would be a big shock.
China is now demonstrating that it can do a lot more cheaply, even if people argue it is more than $6 million, it’s still a knock out punch that comes from nowhere.
Thomas Friedman wrote in the NYT in mid December.
Donald Trump’s relentless China bashing and tariffs during his first term as president lit a fire under Beijing to double down on its efforts to gain global supremacy in electric cars, robots and rare materials, and to become as independent of America’s markets and tools as possible.
“China had its Sputnik moment — his name was Donald Trump,” Jim McGregor, a business consultant who lived in China for 30 years, told Friedman.
“He woke them up to the fact that they needed an all-hands-on-deck effort to take their indigenous scientific, innovative and advanced manufacturing skills to a new level.”
Bizarrely, Silicon Valley venture capitalist and Donald Trump advisor Marc Andreessen on Monday described DeepSeek-R1 as "AI's Sputnik moment".
Sputnik was the first artificial Earth satellite that was launched by the Soviet Union in 1957 and then set off the US space race which led to the first man on the moon.
Where will DeepSeek land and how firmly?
We don’t know yet but for now it does look like Trump and his good friends will have to descend to earth from their lofty heights.
And that brings us to the top stories and themes of the day.
Markets hit 7 month low, longest losing streak in 23 years.
Oil prices fall on fresh Trump call to OPEC to cut them.
Understanding DeepSeek, where did it suddenly come from ?
India Inc wants a friendlier corporate tax regime.
Markets Crash
The Nifty 50 has hit a seven month low and along with the Sensex is set for its fourth straight month of losses in January, which would be their longest monthly losing streak in more than 23 years.
Monday was another tough day with the bears fully in control as FII selling that is in the vicinity of $8 billion is continuing on a host of factors that are as now well evident, ranging from weak earnings to a weak rupee.
Sudden surprises like news of a Chinese low cost AI model which has rattled all the tech majors globally also affects India.
The Sensex closed down at 75,366.17, a loss of 824 points while the Nifty 50 ended at 22,829.15, down 263 points.
Both indices were down a little over 1% but the mid and small caps took bigger hits with the BSE Midcap index falling 2.68 per cent and the Smallcap index falling 3.51 per cent.
BSE-listed firms' overall market capitalisation (m-cap) dropped to ₹410 lakh crore from ₹419.5 lakh crore in the previous session, making investors lose nearly ₹10 lakh crore in a single day.
The DeepSeek Fall on Wall Street
Stocks were down sharply on Monday on concern about an artificial intelligence stock bubble popping because of the emergence of Chinese startup DeepSeek, which has possibly made a competitive AI model for a fraction of the cost.
The Dow Jones Industrial Average dropped 122 points, or 0.3%. The Nasdaq Composite shed 2.7%, and the S&P 500 slid 1.6%.
Nvidia and other U.S. technology firms plunged on Monday, part of a global sell-off as Chinese startup DeepSeek sparked concerns over competitiveness in artificial intelligence and America’s leadership in the sector.
Nvidia, the chip designer who has been a major beneficiary of the AI hype, slid more than 13%. With that, the megacap tech stock was on track to notch its worst day since March 2020.
Nvidia’s losses helped drive other AI trades and the broader U.S. market lower. Micron and Arm Holdings dropped more than 8% and 9%, respectively. Chipmakers Broadcom and Advanced Micro Devices lost more than 14% and 5%, respectively.
Power companies who were supposed to benefit because of all the energy the data centers would potentially need were getting hammered too.
Constellation Energy and Vistra, two of the best-known derivative plays tied to the power buildout for AI, plummeted about more than 12% and 19%, respectively.
International markets also felt the impacts. Netherlands-based chip companies ASML and ASM International both pulled back sharply in European trading. In Asia, Japanese chip-related stocks including Advantest and Tokyo Electron
were broadly lower.
DeepSeek launched a free, open-source large language model in late December, claiming it was developed in just two months at a cost of under $6 million — a much smaller expense than the one called for by Western counterparts. Last week, the company released a reasoning model that also reportedly outperformed OpenAI’s latest in many third-party tests.
In a social media post, Marc Andreesen called DeepSeek’s product “one of the most amazing and impressive breakthroughs I’ve ever seen” and a “profound gift to the world.” The Andreessen Horowitz cofounder recently gained notoriety for his support of President Donald Trump.
Oil Prices
Oil prices fell on Monday after U.S. President Trump called on the Organisation of Petroleum Exporting Countries to reduce prices after the announcement of wide-ranging measures to boost U.S. oil and gas output in his first week in office.
Brent crude futures dropped 35 cents, or 0.45%, to $78.15 a barrel by 0726 GMT after settling 21 cents higher on Friday, Reuters reported.
Oil also recovered from sharper losses seen earlier in the session after the U.S. reversed plans to impose sanctions and tariffs on Colombia, after the South American nation agreed to accept deported migrants from the United States, the White House said in a statement late on Sunday.
Sanctions could have disrupted oil supply, as Colombia last year sent about 41% of its seaborne crude exports to the U.S., according to data from analytics firm Kpler, said Reuters.
Meanwhile, Saudi Arabia may raise crude prices for Asian buyers in March to their highest in more than a year after benchmark prices spiked on higher demand from China and India as U.S. sanctions disrupted Russian supply, traders told Reuters.
The March official selling price (OSP) for flagship Arab Light crude may jump by $2-$2.50 a barrel from February, three of four Asian refining sources said in a Reuters survey.
—
How Deep is DeepSeek
Chinese artificial-intelligence company DeepSeek has Silicon Valley marveling at how its programmers nearly matched American rivals despite using inferior chips, the WSJ reported on Monday.
AI models from DeepSeek, the Chinese company, have zoomed to the global top 10 in performance, according to a popular ranking.
On Jan. 20, DeepSeek introduced R1, a specialized model designed for complex problem-solving.
DeepSeek’s advances sparked a selloff led by chip shares early Monday on concerns the huge spending by U.S. tech giants on leading-edge semiconductors and other AI infrastructure was justified.
Futures tied to the tech-focused Nasdaq-100 were down about 3%, with Nvidia sliding 10%.
So what does DeepSeek mean and how did such an option even emerge?
I spoke with Jaspreet Bindra, founder and MD of Tech Whisperer LTd, a consultant in digital transformation and AI and earlier Group Chief Digital Officer with the Mahindra Group.
I began by asking him how much of a surprise this was?
INTERVIEW TRANSCRIPT
Jaspreet Bindra: I think it definitely has been a surprise for everyone. Though now a few experts and pundits are claiming that they did see it earlier and did talk about it. I think the fact that China is not too far away from the US was not a surprise.
But the fact that they were like this far ahead and looking at costs which were less than 3% of the costs that the valley companies are building models at was a super big surprise and I think you see the effect of it when the markets opened in a few hours time.
Govindraj Ethiraj: Got it. And in a general sense why is it or how is it to the extent that a layperson could understand that the costs are so dramatically different in the way the Silicon Valley has done it or scoped it versus how this particular Chinese company has done it?
Jaspreet Bindra: I think in many ways the US's efforts to keep China a distant second in the race by the whole chips ban, the semiconductors ban, etc. etc. has come back to bite it.
What happened was that while the US companies were working with tens of billions of dollars available, the Chinese companies were starved of resources, starved of chips, starved of compute power. And as we all know, and we know this well in India, that necessity is the mother of invention. And so I think they just approached this entirely fresh, built upon some of the stuff that already had come out in the valley.
So for example DeepSeek and specifically in this case is not trained on the raw data of the internet, which is what you know, Chad GPT or Gemini etc. were trained on and therefore required this huge computing power, billions of dollars, many, many months to get the models right. These guys used a new technique which people are discovering called distillation, which basically takes Chad GPT and the results that gives and builds on top of that.
So much like in many other technologies, you're building on top of stuff which has already been done. And I think this is part of the secret sauce. I'm sure there are others which we still don't know about.
You know, the fact that they've been able to juice out more power from cheaper last generation chips, the fact that they've used young, very bright, but very hungry engineers to make this happen are all parts of the mix. I'm sure there's other stuff too.
Govindraj Ethiraj: Right, and you know, so people are saying that the entire chain of AI investments is now looking questionable, which includes obviously the hardware side of it. So how does that play out? Because, you know, even in India, we're talking about these giant data centres, where we want to spend billions of dollars, which we would have typically spent on those very chips to do the same kind of computing that we think will be done.
Jaspreet Bindra: It's an interesting question. And it can play out both ways, frankly, because now if you think of the two main kind of companies, or three main kind of companies, they're the big techs like Microsoft, Google, etc. Now, while they've committed to invest $80 billion, $60 billion, etc., if they discover a way where they can have AI much more cheaply to deliver to their customers, they'll be net gainers. They're not as much in the AI game as they are in delivering AI infused services to their end customers. So in the medium to long term, this could actually be a gain for them. Number two is obviously the infrastructure guys like NVIDIA.
Okay, and there it gets more interesting. I think NVIDIA will fall the hardest immediately. But again, if instead of five people building AI and spending $50 billion with buying those chips, if suddenly because of this AI model production gets democratised, and 100,000 people can build small models, they will still need GPUs, semiconductors, chips from companies like NVIDIA.
And so you could look at a scenario where there are more customers, less chips per customer. What, however, is interesting to conclude is the third constituency, which are, you know, these big startups now like OpenAI, you know, which have been spending so much money to build these chips, they clearly will have to go in a different direction. And I think at OpenAI, they must be having these war rooms right now.
I know Meta has their four war rooms, which have been formed at Meta right now, you know, to figure out what is the direction that they go in.
Govindraj Ethiraj: All of this seems so far away from how things were looking just a week ago. But let me come to the other question which has been raised. This is open source.
And I'd like to use this to ask you two questions. I mean, about open source in general, and what does open source in this context mean versus, let's say, OpenAI, the proper noun, and open source in, I guess, the common noun.
Jaspreet Bindra: There have been many surprises about DeepSea. The cost, obviously, is one surprise. The kind of chips used is another surprise.
The fact that China came out and did it is a third surprise. Okay. And I think the fourth big surprise is that this was all open sourced.
And the company has published a technical paper on it, which is a full disclosure paper, the kind of papers that almost no Western company has published. So it's a full disclosure, they kind of give out the tech specs, they tell exactly what the weights in the model are, so that anyone can take it and, you know, build on it. In a sense, it's a Trojan horse approach, because obviously, all developers would want to build on, you know, the cheapest model, and this is by far the cheapest.
And so, you know, this model will become the basis of all AI rather than an OpenAI model becoming the basis of all AI. And so I think the open source part is important, one for this reason. And the second is that, in technology, it's always a closed source, open source backdoor.
And we know that that played out in IT, you know, with Microsoft and the whole Vintel, and then the open source Unix and others. And then, you know, even in the Apple iOS versus the Android kind of ecosystems. And I think the same thing will play out now in the closed source versus open source.
And at least history tells us that both continue to coexist. You know, the open source has certain benefits, and closed source has certain other benefits. But if I may conclude by saying that, I think what has happened is that usually open source takes much longer to catch up.
Okay, in this case, especially with DeepSeek, even Meta's Lama open source models, they were like semi open source, was still reasonably behind in a GPT class models like chat GPT or Gemini. But this has narrowed the gap considerably very fast at super low cost. And so it's caught up much faster.
Govindraj Ethiraj: Got it. And last question. So does India benefit?
Can India benefit? How should we be looking at this development?
Jaspreet Bindra: Well, my argument and hope is that while this is an existentialist threat for many companies, countries in the West, Indians should be elated by this development, despite our antipathy towards China. I mean, if you follow the sector, there's always been this debate between whether India should make its own large language models or not, you know, and everyone has bathed in and the main barriers to building India specific India context, India language, and large language models has really been the cost. You know, it's been the compute cost, the cost of talent, the cost of data, etc.
And now suddenly, that is a myth, you know, that shibboleth has been broken. And so I think it gives an opportunity for India to build its own not one, but multiple large language models. And secondly, it's great for startups and developers, if they were paying, you know, what they were paying an open AI or equivalent for their APIs.
Now they can potentially get it for, you know, cents to the dollar, and therefore create much cheaper products have lower costs. So I think for both these reasons, it's good for India, if we grasp the opportunity.
Govindraj Ethiraj: Good note to end on. Jaspreet, thank you so much for joining me. Great.
Jaspreet Bindra: Good to be here. Thank you.
Corporate Tax
Corporate tax collections have been strong in the last few years and perhaps this is a good time to examine what is working and not when it comes to taxation, both administratively and otherwise.
There are clear expectations for instance for lower tax on new manufacturing units as we have discussed here and some clamour for lower corporate tax rates overall.
I reached out to E.N. Dwaraknath – Partner, Price Waterhouse & Co LLP and began by asking him how the last year or so has been in tax collections and trends.
INTERVIEW TRANSCRIPT
Dwarak Narasimhan: Anyone would tell you that the broad tax trends have been extremely positive not only in the last year but in the last few years. There has been a fair amount of buoyancy in tax collections. Growth has been robust and that's obviously also backed by the fact that economically we've been doing extremely well.
Our GDP has grown and that is obviously reflected within our tax collections. The overall share of corporate taxes has also been growing which is a positive thing. On its part I think the government has been of course increasing the amount of scrutiny on taxpayers and corporate taxpayers.
Nothing wrong with that. With increased growth comes the responsibility to pay additional taxes. There has been a focus or greater focus with regard to things like assessments like high-pitched assessments, a lot of transfer pricing appeals and assessments which have come up, a lot of litigation on foreign companies on aspects such as prominent establishments and of course some invasive procedures such as search and survey proceedings which continue.
So I think overall extremely buoyant collections on the back of a strong economic growth and good performance by Indian companies.
Govindraj Ethiraj: Dwarak, any trends within that in the terms of what's been more secular and what could have been more let's say in the nature of bumps?
Dwarak Narasimhan: If you look at it industry-wise I think there's been a spurt in manufacturing again which is to be expected. There's a lot more to go still but there has been a spurt in manufacturing and therefore the kind of assessments which have been taking place for manufacturing companies. There have been continued questions with regard to amounts which are leaving the country.
So payments made by a lot of manufacturing companies such as for management fees, for royalties, anything that goes to reduce their profits, anything which is paid outside the country are being questioned a lot more. And that's also only natural because I think the government's thinking is hey guys if you've been manufacturing in India for a period of time then we would expect you to be more self-sufficient and we would expect these kind of payments to reduce over time. So I think there has been a significant spike in the kind of assessments which have been carried out for those kind of companies.
There's been a huge spike in terms of international tax. So the number of foreign companies which have been receiving tax notices, the kind of issues which have been coming up specifically with regard to things like permanent establishment have been immense. A lot of adjustments being made on that count.
Finally on the not so good side, I think the move to faceless assessments has not been without some challenges. I think the amount of information which is being sought by the government is immense. It's extremely voluminous.
So I think if they're able to taper that down to be more specific in terms of the asks of companies, it'll just ensure that A, companies don't waste their time providing a whole lot of irrelevant information and B, the government also targets its focus to something which is relevant and which perhaps can generate greater revenue for them.
Govindraj Ethiraj: You've raised about three or four, I think broadly the administrative approach issues when it comes to tax collections. Is there anything specific that is on top of the mind, either a clause, something that's old or new? We also have a new direct tax code imminent.
How much of it will actually emerge in a week's time, we don't know. But anything there?
Dwarak Narasimhan: Yeah, you took the words out of my heart. So I think there is of course the hope that there will be a new direct tax law, which will come in, at least that's what the finance minister announced six months back. But since that point in time, I think there've been around 22 subcommittees that the government has formed to look into the simplification of the tax law and come out with new provisions.
They put it out, they requested for public consultation in the beginning of October and I believe more than 6,500 suggestions have been received. We still do not know if they're going to come out with a new tax law altogether as part of this budget, or whether they're just going to tinker with the existing law as it is and then propose a new finance bill before the budget session closes. So I'm betting on the second at this point in time, maybe a little bit of tinkering this point just in the budget, but thereafter a new finance bill hopefully comes out maybe before parliament closes or the budget session.
In terms of provisions itself, I think three or four asks, right? One, just simplify the language. It is extremely, extremely complicated, so just simplify it.
Second, remove provisions which are redundant. And this law was written in 1961. The world has changed, businesses have changed, business models have changed.
Remove stuff which is not relevant anymore. What's an example of that? I think a lot of the provisions with regard to taxation of specific kinds of companies is something which is just redundant.
I mean, if you look at the provisions which were introduced for taxation of royalties over a period of time, I mean, those kind of provisions don't exist anymore. So make it more current to what is happening in the world today, which would be my third ask, right? Bring in legislation which is relevant to the kind of businesses and the business models that exist today.
For example, if you say that you want to invest, or the government wants to invest heavily in infrastructure, that will necessarily require, and historically we've seen it, a large presence of foreign personnel in India for a substantial period of time. Now that creates something called a permanent establishment. So can the government come up with some sort of provisions that ease the presence of foreign technicians in India over a substantial period of time?
Will we come up with some suggestions with regulations for Pillar 2, which the whole world has been talking about? And today that one is tricky. Thanks to Donald Trump, exactly.
And I think the US Secretary of Treasury, or his pick for Treasury, has gone out on a limb and he said, any country that introduces Pillar 2 regulations better watch out. So again, is India going to bow to that pressure? Or are we going to go ahead and say, no, this is a commitment, and we're going to go ahead and do it?
The taxation of online companies and a lot of online businesses, things like the significant economic presence, which was introduced three or four years back, need simplification, we need clarifications. So I think making the tax law more relevant to the kind of industries, space, and a sector which has immense potential for the country we want to invest in. Are we going to come up with something for that?
Are we going to do something on green energy? 2070 is the target we've given ourselves to be carbon neutral. But to get there, there has to be substantial incentivisation of businesses to go carbon neutral over the next five to seven years.
Of course, taper it off. Nobody is saying you need to continue with tax breaks forever. But this is the time to do it.
So I think these are some of the things, that I think the government should be looking at.
Govindraj Ethiraj: Thank you so much for joining me and speaking with me.
Dwarak Narasimhan: Thanks, Govind. Pleasure as always.
