The recent correction in artificial intelligence (AI) stocks is unlikely to mark the end of the AI investment cycle, but investors should expect a period of consolidation and become more selective, according to Arvind Sanger, Managing Partner at Geosphere Capital Management.
Sanger said investors are likely to get greater clarity after the US midterm elections, particularly around AI infrastructure and semiconductor demand. He also expects opportunities to emerge beyond the AI theme, including specialised IT services, financial companies and domestic demand businesses in India.
Sanger said several new risks have emerged for AI stocks. Apart from concerns around memory pricing due to developments in China and political uncertainty over data centre and power infrastructure in the US, investors are now focusing on whether AI infrastructure companies can continue funding heavy investments.
He said concerns over companies such as Oracle and CoreWeave have shifted attention towards balance sheets, making investors more cautious after a strong rally in AI-related stocks.
On India, Sanger said geopolitical developments remain a key risk. He warned that crude oil prices could rise if geopolitical tensions escalate after the US midterm elections.
Despite that, he continues to prefer export-oriented manufacturers, specialised financial services companies, select banks, hospitality and travel businesses, along with niche mid-cap opportunities. While valuations in the mid- and small-cap segments remain elevated, he believes IPO-related liquidity absorption is creating stock-specific opportunities.
This is an edited transcript of the interview.
Question
Let’s begin with AI. We’ve seen these subsequent corrections. You know, I mean, we’ve seen similar falls in AI stocks in the past and subsequent corrections, followed by recoveries thereafter. Once again, earlier this week, the markets went through that phase of AI jitters. What did you make of it? Overdone, or is it different this time?
Answer:
I think there are two or three differences this time. One is that you have more memory and other stuff coming out of China that is creating a headwind for one leg of the stool. There are many legs of the AI stool. There’s the power leg, there’s the chip leg, there’s the memory leg if I can pick three big ones. Then in the memory segment, there’s concern about China. On power, there’s concern about political midterm noise about Democrats and the left being opposed to data-centre build-out and the power build-out. And then on the chip side obviously Nvidia, but there’s competition from AMD and others. And I think the real story, in my opinion, that has been added to this is the whole concern about the balance sheets. First, you have this concern about Oracle. Now you have the concern about CoreWeave. Can they keep the spending going? So, there’s, clouds on the horizon. The markets have been on a tear, and this was an overdue breather. I don’t think the AI story is over, but some of these issues might cause a little bit of digesting before we can say it’s all clear. So, I don’t think it’s going to be a straight bounce back, but I think after the midterms, we will see a better picture, certainly on the power side and on the chip side.
Question
Gartner – I don’t know if you read that – forecasted that 80% of the specialised AI cloud providers will fail to survive independently by 2030 due to cost pressures. But broadly, do you think it’s time to take money off AI stocks right now?
Answer:
I wouldn’t say broadly, but yes, selectively, definitely, because into any rally, you have to recognise at this point that the earnings growth that has been coming from the AI companies is kind of unsustainable at the level at which it’s going. So that is true, but at the same time, you look at some of these valuations, and they’re trading at very low multiples of earnings. So, the real question is, how close are we to peak earnings? And our view is that for some parts of the AI story, it’s hard to tell, and the market is going to treat them as guilty till proven innocent now because the mood has changed. Therefore, I would say that yes, a little bit more diversification beyond the AI-only theme that has been dominating through most of 2026 makes a lot more sense to us.
Question
Looking to get some special edge from you since you’ve got a front-row seat to what’s happening in US tech as well. And you’ve had a cautious view on IT services in India and IT services, or software services generally overall, right, for a while. Here in software services, we’ve had corrections in AI. We’ve had bounces in IT services as well over the last two years, 7% to 10%. Now this one is turning out to be about a 10% rally so far. What’s your sense? What’s the mood like in this space in terms of what’s happening, fundamentally, how fast are things moving in terms of disruption, etc.? And on the other part, the move that you’re seeing, since you also manage money.
Answer:
I would say that I think putting a broad brush on IT services may no longer apply. I think there are clearly going to be companies that are much better positioned to help implement AI, because companies are recognising that they can’t just rely on an LLM to do all the work for them. They want to build their own LLM, keep their data to themselves, so that they’re not displaced by a cloud provider or by OpenAI. Therefore, there is an implementation story that is becoming much more important for most corporates, and therefore there is a role. But I’m not sure how well a broad-based IT services company will do versus some of the niche specialty players. Overall, IT services probably has a good runway for the next couple of years, but the specialised ones are the ones that are going to be winners. The real problem for the broad-based ones is, when you look beyond the next two or three years, a lot of the business that is mass market is going to get disrupted. That’s why I would say IT services is not quite as negative, but I wouldn’t go out and load the basket with all the major IT service companies.
Question
Within India, what would you load the basket with then? Our markets, yes, have pulled back a considerable bit, but the economy actually appears to be in fine fettle. All the high-frequency data is looking up. Yes, we have got our share of challenges on the macro front because of the higher crude oil prices and a deficit monsoon as well. But if you’re looking at India, what’s on your pecking order?
Answer:
One of the things is that we haven’t talked about crude. There is a risk that after the midterms are over, Trump could go back and start a hot war again. And if that happens, Iranian retaliation could cause a much bigger move in crude prices. So, we have to keep that in mind. That’s not that far away. That’s only three weeks away before we could see something happen there. So, having said that, our preference has been for either companies that are manufacturing or selling to the global market, or domestic unique-demand stories. And I think the unique demand stories that we’re finding most interesting are specialty financial services, NBFCs, as well as one or two banks, as well as places such as hospitality and travel. There are a lot of interesting midcap names. And I’ll admit midcap and smallcap stocks are not cheap, but there is still enough value to be found. And you’ve got money being sucked up by IPOs. You had NSE come a few weeks ago. Now you have Reliance Jio coming, and that does suck a little bit of liquidity out of the market. When that happens, we find opportunities in some of these niche names, which we think give a lot of significant upside.
For the full interview, watch the accompanying video
Arvind Sanger sees better opportunities beyond AI-only trades, favors specialised IT and NBFCs
Catch all the latest updates from the stock market here
NOTE TO READERS
Disclosure: Reliance Industries Ltd, which owns Jio, is the sole beneficiary of Independent Media Trust that controls Network18, the parent company of CNBCTV18.com.