On Episode 901 of The Core Report, financial journalist Govindraj Ethiraj talks to Venkat Mangudi, Cyber Security Expert and Founder-CEO at Elytra Security as well as Anand Kulkarni, Director at Crisil Ratings.

SHOW NOTES

(00:00) The Take

(06:55) Markets set to rise further as oil prices head downwards in anticipation of Iran-US deal

(11:58) The G7 meet starts off today, India will also be present.

(12:26) Why is Anthropic pulling its latest LLMs from the market?

(22:49) EV majors are investing billions of dollars as market demand picks up.

—

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 feedback@thecore.in.

Good morning, it's Monday, the 15th of June, and this is Govindraj Ethiraj, broadcasting and streaming weekdays from Mumbai, India's financial capital.

The Take

Should India's software services behemoths have climbed aboard the AI bandwagon earlier, sinking their capital into large language models or the vast data centre infrastructure required to power them? This question has recently animated a broader debate over whether Indian IT firms, sitting on mountains of cash accumulated over decades, missed a generational technological shift. Inevitably, this hand-wringing extends it to a familiar lament.

Why do Indian companies continually fail to invest sufficiently in research and developments or R&D? The autopsies offer some interesting takeaways. India's chief economic advisor, V. Anant Nageshwaran, writing in the Indian Express last week, noted that the deficit stems from a larger captive domestic market, a historical leaning towards trading and arbitrage, early financialization, and third-generation business families losing their entrepreneurial steam. Veteran investor Shankar Sharma pointed the finger at India's capital markets in an article written for the Business Standard last year.

The giant machine called the stock market does efficient capital allocation and hence it efficiently allocates capital to businesses with high valuation potential. No manufacturing or core research business can ever match up to services in terms of returns on equity and predictability, he wrote. He also said, in short, stock markets recoil at risky overseas ventures.

Management teams understand this and shape their strategies to satisfy market demands. National technological goals consequently take a backseat. Contrasting India with the United States, Mr. Sharma argued that stock market booms in both nations have bred long-term uncompetitiveness, hollowing out core research and manufacturing.

The harsh reality, he said, is that only the state possesses the patient capital required for long-term research. Both democracies, he argued, have starved research in favour of democratic populism. Today, Tata Consultancy Services, Wipro, and Infosys are sitting on roughly $5 billion each, give or take.

Should these firms have bet their treasuries on AI when the first signs of that opportunity appeared? Did they even spot it? And if they did, what could they have done realistically and differently? Unfortunately, backward-looking alternate history debates about what Indian IT companies ought to have done are mostly a waste of time. You have to remember that in the 2000s, these firms were toasted for creating a whole new class of jobs running into millions and democratising wealth. Consider the famous story of then Infosys chairman, N.R. Narayana Murthy's chauffeur, who became a millionaire by virtue of his employee stock options.

In the late 90s, Mr. Murthy declared his ambition to create 1,000 millionaires within Infosys by the year 2000. He succeeded, a monumental triumph in a country where wealth creation was, and perhaps still is, viewed with deep suspicion. Distributed ownership, an alien concept to many a traditional business family at the time, became an undeniable force, something that many companies over the years have brought into their compensation packages as well as growth paths.

Since the initial Y2K boom, when Indian coders rescued global computing systems from midnight collapse, these IT firms, now part of a $315 billion industry, have navigated several perilous transitions, including the notoriously difficult pivot to digital transformation. But the latest battle with AI is undoubtedly the most challenging, possessing genuine destructive capability. The IT majors argue that they still need it, albeit in leaner numbers, to operationalise and manage AI within large global enterprises.

Sceptics, however, believe that this run cannot last. And yet, while Indian IT companies have arguably eschewed mega bets on AI, so too have their global counterparts. Accenture and Capgemini are prime examples of large software services majors that have not really embarked on mega AI bets, or at least of the kind people have been referring to in these debates.

All of them are doing precisely what their shareholders, boards, and leadership mandates have asked them to do, which is to deliver high-margin software services, capital appreciation, and steady growth. Broadly speaking, they've executed this playbook flawlessly for more than three decades, if not more. Could they have seen the writing on the wall and destroyed their existing businesses to build a new, akin to Jack Welch at General Electric? Perhaps.

But Mr. Welch's track record in the rearview mirror looks far less stellar than it did at its peak. So one larger question is, where does true innovation originate or lie? Does it spring from within and existing large enterprises? Occasionally, one can point to Nokia's century-long evolution, or the transformations of Nintendo, American Express, and IBM. India, too, has seen trading houses and textile mills evolve into multifaceted conglomerates like the Aditya Birla Group, Reliance Industries, and the Tatas.

But betting on R&D in the context of AI is a fundamentally different wager. When the late Ratan Tata launched the Indica in 1998, it was a major departure for a truck maker driven by a conviction that India needed an indigenous passenger car. Similarly, two-wheeler manufacturers like Hero Bajaj and TVS invested in R&D to stretch every possible kilometre out of a litre of petrol in the 1990s.

These are classic examples of frugal innovation, but in perfect hindsight, this remained incremental. Could India's new generation of tech entrepreneurs have been the deep-tech innovators of this century? Well, the evidence is not encouraging. The most high-profile founders, including some who have since decamped with investor capital, built businesses in edtech, a sector whose very existence is now routinely questioned.

Others are burning vast sums of venture capital on rapid pizza and shampoo delivery. In these arenas, VCs simply replace the stock market in dictating capital allocation. While there are certainly determined entrepreneurs building admirable companies in space tech and defence, they do not yet approach the scale, magnitude, or audacity of what newly-minted trillionaire Elon Musk is executing at SpaceX or some of the other large-language model companies like Chad GPT or OpenAI and Anthropic are doing.

So the only pertinent question perhaps remaining is whether India is cultivating the right educational ecosystem that will produce or throw up such leaders. Are the country's future innovators growing up to think differently, ask difficult questions, and target massive structural problems? Well, India has as much to gain from solving the complexities of indigenous drug discovery and scaling high-quality education for its burgeoning youth as it does from a proprietary LLM bet that may or may not materialise. Looking at the current state of affairs plagued by chronically leaked examination papers and an education system structured like a winner-take-all lottery, it's painfully clear that the country has far more pressing work to do.

And that brings us to the top stories and themes…

Markets set to rise further as oil prices head downwards in anticipation of a Iran-US deal

Why is Anthropic is pulling its latest LLMs from the market?

EV majors are investing billions of dollars as market demand picks up in India

And the G7 meet starts off today and India will also be there.

Markets, G7, Oil and Asian Currencies

The market shot up and saw their best session in two months on Friday in anticipation of a deal between the US and Iran, could Monday see a repeat? On Sunday, a senior Iranian official told Reuters a final draft of the MOU, or the Memorandum of Understanding with the United States, covered a range of issues, from Tehran's nuclear work to reopening the state of Hormuz and US waivers on oil sanctions, with a final deal to be discussed in the 60 days following agreement by the two sides.

As per the known contours of the deal, Iran immediately reopens the state of Hormuz to all commercial vehicles while the US lifts its naval blockade on Iranian ports. Speaking of the US blockade, Secretary of State Marco Rubio defended Washington's position in the state of Hormuz, responding to protests by India after US strikes left three Indian mariners dead. Rubio spoke with India's External Affairs Minister, S.J. Shankar, to discuss the developments in the state of Hormuz.

According to a statement from the US Department of State on Saturday, the Secretary stressed that all commercial vehicles should immediately comply with orders from the US forces as they seek to maintain peace and security in the strategic waterway, according to Bloomberg. Rubio also said that violations of the US-led blockade and the illicit transport of Iranian oil would not be tolerated, according to the statement, and those remarks came after S.J. Shankar protested the attacks by the US Navy in the Gulf. All of this would obviously suggest that India's protests did not really make much headway.

On the contrary, India seems to have been told off. Meanwhile, the good news is that oil prices are falling. In hopes of a peace deal between the US and Iran saw crude oil prices fall nearly 4% on Friday to about $87 a barrel, where it is right now.

If the deal were to materialise, analysts are seeing Brent crude oil prices heading lower. Since Indian indices have inversely and closely tracked crude oil prices, you could expect markets to rise if crude oil prices drop further. Low crude oil prices means a lower import bill and thus less pressure on the rupee and more on that in a moment.

On Friday, the markets did see that strong rally in anticipation of the deal with the Sensex rising 1,695 points to 75,527 and the Nifty 50 rising 461 points to close at 23,622. In the broader markets, the Nifty Mid Cap 100 and Small Cap 100 were up two at 2.4 and 2.8% each. Almost all sector indices were in the green except the Nifty IT, according to Business Standard.

And just to recap the big IPO news from Wall Street, despite a large retail allocation and a huge amount of hype, trading on the SpaceX stock wasn't especially volatile and the positive momentum continued after the market closed for the weekend, according to CNBC. The rocket launch computing and satellite company delivered the largest IPO ever with a trading volume of more than 500 million shares and a closing price above $160, putting its first day market capitalisation over $2.1 trillion. The stock opened at $150 and closed 20% above its telegraphed offering price of $135 per share and continued to rise in after-hours trading, reaching $167 a share, according to CNBC.

Back home, India's retail inflation rose to about 3.93% in May thanks to higher food and fuel costs, according to data released by the government on Friday. There is, of course, the risk of a weak monsoon pushing up prices and thus inflation further up. Sitting where we are in Mumbai, it is still to rain.

We touched on the rupee a little while ago. It hit a one-week high of Rs. 94.95 on Friday before closing at Rs.

95.11, up 0.7% of the day. The move, according to Reuters, was not enough to protect the currency from posting its first weekly decline after rising for the last three weeks. Elsewhere, the Wall Street Journal reiterated that several Asian countries are facing currency pressures.

From Tokyo to Seoul to New Delhi, Asia's currencies are on a losing streak and governments have had enough, according to a Wall Street Journal report, saying that Tokyo has burned through more than $70 billion, defending the yen this year, and finance ministry officials there are warning investors not to test their resolve to do even more. Indonesia's central bank, as we've been reporting, has jacked up interest rates for the second time in three weeks last week, hoping to staunch the flow of capital fleeing the country and to protect a falling rupiah. In South Korea, says the Wall Street Journal, authorities are stepping up scrutiny of foreign exchange trading to tackle what they say is excessive speculation in the Korean won, which has fallen more than 5% against the dollar since the start of the year.

And all of this is, of course, despite that epic AI-fuelled export boom. The Wall Street Journal, of course, has argued even until last week that the United States should finish off the job in Iran, which means effectively not signing any deal and resuming the war. Meanwhile, gold prices are searching for fresh levels, it would appear.

Remember, they've fallen 25% after hitting a record of $5,595 per ounce in Jan, and are currently at around $4,188 an ounce, a six-month low. Analysts told Reuters they see scope for gold to bounce if the conflict in West Asia eases and oil falls to about $80 a barrel. And finally, Prime Minister Narendra Modi and U.S. President Donald Trump will hold bilateral talks on the sidelines of the Group of Seven summit in France, or G7.

According to a White House confirmation, the meeting will mark the first face-to-face interaction between the two leaders since February 2025, and Trump is expected to discuss plans to demine the state of Hormuz with allies during the G7 summit, according to a AP news agency report, quoting a senior U.S. administration official.

What are Anthorpic’s Recent Dramatic Moves?

AI giant Anthropic has suspended its powerful new AI model after U.S. authorities raised security concerns just days following its public release. In a statement published on its website, Anthropic said it was ordered to suspend foreign nationals from using Claude Fable 5, a programme that the company self-described as too powerful.

Anthropic PBC has disabled access to its most advanced artificial intelligence models, including MITOS, following an unprecedented order by the Trump administration to keep the technology out of the hands of all foreign nationals, according to Bloomberg. The U.S. government told Anthropic to suspend access to the Fable 5 and MITOS 5 models by any foreign national, whether inside or outside the United States, citing national security concerns. Anthropic said U.S. national security authorities have not identified specific concerns.

The BBC quoted the company saying their understanding is that the government believes it has become aware of a method of bypassing or jailbreaking Fable 5. Jailbreaking is the process of getting past software restrictions designed to protect a cyber network, allowing hackers to access sensitive information or unblock features. I reached out to Bangalore-based Venkat Mangudi, cybersecurity expert who helps companies build cyber defence systems, and I began by asking him to explain Anthropic's move.

INTERVIEW TRANSCRIPT

Venkat Mangudi: Okay, so it's not just about foreign nationals. Actually, they have disabled it for all users. So the background is that Mythos was touted as a very, very strong, powerful model.

It had identified vulnerabilities in software that were not identified for 30 years. In BSD, for example. BSD powers Mac OS and many of the older systems globally.

Govindraj Ethiraj: What is BSD?

Venkat Mangudi: BSD is Berkeley Systems Division Linux. It's an operating system. And people were using Mythos for finding vulnerabilities as part of Project Glasswing.

Anthropy gave access to very few companies in the US initially. In fact, if you remember, Dr. Sitaraman asking all the BFSI companies to be a little more vigilant about AI threats came right after that. So Mythos was very powerful.

It was even more powerful than the open AI equivalent, which is called ChatGPT 5.5 Cyber. The problem actually is not about Mythos or Fable. It is that the capability of that model, it almost sounded like infrastructure.

Because that model was so powerful, it could reason. It could write code. It could plan.

It could actually set up an entire workflow. For example, if there is a hacker today, and if it's all done manually, which is becoming very rare these days, they are going to try a few things. And then based on those results, they are going to pivot and do something else.

The way Mythos has been projected is that it will do the entire workflow. Like imagine, if you will, an assembly line, a truck assembly line or a car assembly line. Mythos was able to do all the steps unattended, autonomously, and reason it out.

It would reason out, okay, this is the problem. Okay, so I can now go and figure out what the next level of penetration or offensive testing I can do. So, it raised fears, which is why Anthropic wisely did not release it to everybody.

They released it to a few companies, only the large companies apparently, who were doing defensive work. And then there was a lot of pressure globally. They only had released it to a few US-based companies.

All the other countries started raising a hue and a cry. And then recently they released Project Glasswing to non-US organisations. And then all other people, you know, all the others who are not in that select list or elite list that Anthropic used, were left behind.

So, Anthropic then took that Mythos model and created Fable 5. What they did they supposedly applied a lot of guardrails, that it would not do certain sensitive cyber operations. And if a user tried to persist beyond that, it would fall back to the previous model called Opus 4.8. Day before yesterday, I mean, two days after that was done, a researcher was able to find out that they were able to get over the guardrails, right? Disable the guardrails. Fool Fable into doing something malicious. See, the way I look at it, whether it is an offensive work or defensive work, whether it is malicious intent or for cyber security defence, the tool and the capability is the same.

The difference is in the intent. And the AI system cannot determine the intent. It can only determine capability.

And you obviously know Anthropic has openly come out and said that Claude writes 80% of its code. So it is now without a lot of human oversight that was there before. So who is looking at those guardrails?

Who is ensuring that the model is doing the right thing without doing something bad? And the knee-jerk reaction from the US government was national security. That was their reason.

They said, if we release this to foreign nationals, for example, North Korean hackers could open an account in AWS or any other data centre in the US and US would not know about it, or Iranian hackers or Russian hackers. Because their location as provided by the IP address is not in one of those embargoed countries or restricted countries, they can potentially use Fable 5 to do malicious things. That raises the question of national security.

And that is why the US government issued that directive. Anthropic said they did not have enough evidence, so they just disabled it for everybody till they investigated further.

Govindraj Ethiraj: So given the march of AI technology, could someone else catch up with a similar version? And if so, would we be able to control it or manage it in a similar way? When I say we, I mean governments.

And therefore, where is the Senate?

Venkat Mangudi: So the other major models are DeepSeek, OpenAI, ChatGPT, and then the open source, Ollama. Ollama is not trained with as much data sets as OpenAI or Anthropic, or even DeepSeek. So OpenAI and Anthropic have been slugging it out for a long time.

They are the true competitors, so to speak. Gwen and Gemini and Copilot are left far behind. They are there, they are capable, and they also have capabilities for doing such things, but not at the level in which Mythos had.

It surpassed OpenAI by about 30% when it was tested. So yes, there is an opportunity for other vendors to build a better model. Yes, there is an opportunity for, again, these are US-based companies, OpenAI is too, and US will clamp down if they consider it also a national threat.

Govindraj Ethiraj: Right. And what should companies or CTOs of companies doing at this point to ensure that their systems are protected? I'm talking about banks or other kinds of institutions in India.

Venkat Mangudi: Right. Today, none of these cybersecurity large vendors, I mean, there are a few who have got some thoughts, reasons, reasoned out the challenge, but none of the large vendors actually can provide a solution. The way I look at it, and this is well-known among the cybersecurity companies, there is no 100% guarantee that any tool or any firewall or endpoint detection and response system can 100% stop malicious intent.

It is always like how the terrorists and the counterterrorism works. The counterterrorism comes up with a way to track the terrorists and the terrorists find a way to get under that radar and try to do something different. So the hackers are going to persist.

In fact, when ChatGPT launched first, scams went up 162%. I think there was a McKinsey report released a year later, 162% jump in cybersecurity incidents and ransomware attacks. So it is not going to be something that an organisation can effectively stop.

Gone are the days where actually a hacker like in the movie sits and types commands and those days are long over. Last year in 2025, they had scripts that would automate a bunch of things in one shot. September, I started noticing something where it was driven by AI.

I called it Systemic AI Reconnaissance. And I release a report every year called Elytra Threat Risk and Resilience Report. I had predicted in that, that in 12 to 18 months, we would see a spate of SAR, Systemic AI Reconnaissance.

But I was very, very wrong. Because I released it in October and in November, Anthropic released a report that hackers from China used Anthropic Cloud and had attacked about 30 companies. That was the beginning.

They called it GTG 1002. That was the code name. And it just went on up from there.

It's going to get worse at the end of the day.

Govindraj Ethiraj: Right. Thank you so much for joining me.

Venkat Mangudi: No worries.

Why are EVs picking up in India?

Annual electric vehicle car sales volumes are expected to more than double by next year, even as adoption is gathering pace, according to a new report from Crisil Ratings.

Companies are spending some 60,000 crore rupees in catching up with demand, of which close to half will be on electric vehicles. Average monthly volumes are rising 40% to around 26,000 units during the first three months ended May 2026, an all-time high compared with the fiscal 2026 average, according to Crisil. The reasons are fairly apparent.

The running cost of petrol and diesel vehicles has increased 7% to 8% this May, thanks to a rise in fuel prices, which has improved the total cost of ownership by a good 300 basis points. And given, of course, the geopolitical uncertainties and if fuel prices are hiked further, this advantage could be reinforced. But the demand surge actually started earlier.

In the three months through May, electric vehicle penetration rose to 6%, up from the last year's average of 4.6%. Moreover, the acquisition cost of electric cars has declined by 10% to 15%, thanks to product innovation and scale efficiencies. And there are, of course, more models. The report also points out that despite the progress, challenges remain.

While public charging infrastructure has scaled up substantially in recent years, it remains highly urban-centric and uneven across regions. So how are auto companies geared for this growth and how strong are their financials to finance the growth? Since at this point, EVs are still in investment mode. I reached out to Anand Kulkarni, director at Crisil Ratings, and I began by asking him how he was seeing auto majors ramping up for the EV opportunity.

INTERVIEW TRANSCRIPT

Anand Kulkarni: Let me take that question first on the investments that these companies are making. So when we tried analysing the OEMs who are investing in EV space and tried analysing which part of the investments are going where, so broadly the numbers are over the current fiscal and the next fiscal, overall CAPEX outlay including ICE as well as EV is estimated to be around 60,000 crores. When we tried drilling down into what is the breakup of that, so around 40% of that is expected to be towards EV portfolio expansion or supply chain localisation, scaling the production, some of these things are happening.

Now the point is in the near term obviously EVs are comparatively margin dilutive because they have not yet achieved a certain scale. So typically such investments have a medium to longer term horizon. Any new segment where the adoption is gradual will have a certain improvement trajectory.

So we believe at current stage where the scale is slightly limited and higher upfront fixed costs are there or competitive pricing strategies are required, these will be slightly margin dilutive initially. However going forward the margins are expected to improve as the volumes ramp up and we've shown that the volumes are expected to double by next fiscal and the penetration is also increasing. So with such ramp up the operating leverage will gradually come into play.

Additionally localisation of supply chain will also drive some cost efficiencies. What is happening is the large part of the cash flows of these OEMs is still coming from the ICE segment and which is quite strong at present. The balance sheets are also quite strong and the investments are done in a phased manner, it is not a lumpy investment.

So with all that I think OEMs are very well placed to absorb these investments even with some short term margin pressure and the cash flows and balance sheets are quite strong with a clear expectation that scale will restore profitability for electric segment as well over a period.

Govindraj Ethiraj: You're saying the total investment at least the way you've calculated is about 60,000 crores in this year and the next.

Anand Kulkarni: Correct.

Govindraj Ethiraj: Okay, now you're also saying that basically the demand surge which I think even the Federation of Automobile Dealers Association data showing started even before the war started.

Anand Kulkarni: Absolutely that's correct. So today obviously the fuel price hikes and its impact on EV is the talking point but when we looked at slightly before that so the penetration that is electric cars as a percentage of total cars has increased materially in fiscal 2026 itself used to be in the range of let's say two two and a half percent which has already increased to around 4.6 percent in fiscal 26. If you look at the average for the last three months ending May 26 it has already jumped from that 4.6 to more than six percent around 6.1 percent. So all of this has largely happened even before the recent fuel price increases fully played out. So we believe that it is more structural in nature fundamental in nature and the reason is the favourable total cost of ownership the way we calculate. So the total cost of ownership for electric vehicles has been improving and it has been favourable as compared to ICE vehicles because it has multiple factors it has vehicle acquisition cost annual running cost maintenance and what you've seen over last two fiscals is there is around 10 to 15 percent decline in the acquisition cost itself and these are because of either product innovation scale efficiencies which has been passed on by the OEMs and these are benefiting consumers. So the TCO has been a primary driver all of that we believe has played out even before the recent fuel price hikes came in.

For future I mean the story that we have highlighted for future there are further couple of important structural drivers. One we believe that wider availability of EV models then the availability is increasing. Today already we have around 20 models available.

There are several launches expected to be there over the next fiscal or so. Many of them will be in sub 15 lakh category as well where the mass or the volumes are there. We believe the model count total model count will be around over 35 and all of these will increase the consumer access which has been one point which was kind of constraining the adoption in the past.

Further the range anxiety which is also an important consideration for EVs. Now that has been addressed by in some different form that is the increasing range per charge. So we are seeing premium models now offer 500 to 700 kilometre per charge while the mid-range vehicles also offer around 300 to 450 kilometre per charge.

So all of these are also addressing some of the concerns around range anxiety and there are some extended battery warranties or innovative ownership structures on batteries. All of this we believe are structurally important and favourable for the adoption of EVs. Hence we believe the volumes will more than double by next fiscal and penetration will increase to around 8 to 10 percent.

That's what our analysis says, Govind.

Govindraj Ethiraj: Right and I don't know if you have this number but the 60,000 crore figure that you talked about how does that compare to the investment that's being made in the I-space or petrol and diesel vehicles?

Anand Kulkarni: It is not exactly like-to-like because the initial sometimes investments that are required towards setting up some of these facilities are different which may not immediately yield the result. So there is no like-to-like comparison. But it's higher than what the automakers are investing in I-space facilities?

40 percent of that 60,000 crore going towards EV is certainly higher and an increasing trajectory. So I'm sure EVs is getting more and more focused in their overall investment theme.

Govindraj Ethiraj: Okay so you overall investment of which EV is about 40 percent?

Anand Kulkarni: That's correct.

Govindraj Ethiraj: Got it. We've seen very strong demand in the auto space in since September last year because of lower prices particularly for ICE vehicles at the entry level. Are there any other price trends that you're seeing or price trigger trends? I know you referred to it but is there anything else?

Anand Kulkarni: So I think post that GST driven benefit that ICE vehicles got around September and after that I think that was more transient in nature and we believe the EV improvement trajectory resumed after that. So we have commented accordingly that it is more fundamentally in structural nature. We believe this trajectory will continue going forward for electric four-wheelers.

Govindraj Ethiraj: Right. Anand, thank you so much for joining me.

Anand Kulkarni: Sure. Thank you, Govind.