Is AI an existential threat to India’s outsourcing industry?

Welcome back to The AI Shift, our weekly exploration of how AI is changing the labour market. This week we’re looking at IT outsourcing in India. Shares in Indian IT firms have plunged since the launch of Anthropic’s new suite of professional AI tools last month. Are investors right to worry that recent breakthroughs pose a deep threat to these companies and workers? Or does India have an opportunity here? We asked the FT’s excellent and well-connected Mumbai correspondent, Krishn Kaushik, what’s really going on. Here’s a lightly-edited transcript of our conversation.

John: IT is very important for the Indian economy, and everything we hear is that AI is going to disrupt software. So what’s the mood music? And what’s actually happening — what’s the concrete stuff looking like?

Krishn: IT is a massive sector in India: it’s about $300bn worth of revenues, it employs over 6mn people, it’s the largest white-collar sector in the country. So it’s extremely imperative for India that the sector does well.

When we talk to people about the impact of AI on this sector, there’s a huge disconnect. I think that is coming from a sense of desperation from the industry leaders, because they don’t want to send out a message that’s gloomy — or even slightly gloomy — because that can create panic in the industry, and among the clients as well. So they are putting on a very brave face. They keep saying that if there is any impact on jobs, it will be temporary; that it means people will be re-skilled to do better jobs; and that AI is not a threat but an opportunity. But I think that is the typical industry-speak you will hear from anybody who wants to dissuade anyone from thinking there is any threat.

When you talk to everybody else, whether that’s the independent experts, whether it’s people in the civil society, whether it’s the market people, you can see that there’s a sense of panic. Even today, they’re not sure how AI will threaten the entire sector, but they think there is something truly amiss.

Sarah: What’s the argument the industry leaders make?

Krishn: What they want to project is: this IT sector has insight about various industries and companies because they’ve been working with them for the last two decades. They say ‘we can give you the diagnostic answer for how you need to leverage AI, and let us build those AI tools for you’. Now it’s up to those enterprise clients to decide: do they need those IT services companies any more, or can they build those tools themselves, and leverage AI how they want?

Sarah: So it’s a question of whether the big clients continue to use the Indian companies to provide the AI-enabled service, or do they just think: we can figure out the AI stuff and we don’t need those guys any more at all?

Krishn: Absolutely, that’s the biggest fear these IT services giants have. But they insist: you need us. We understand your sector better than you do. We understand how your company can be helped by AI better than you do.

They are now trying to get into partnerships with these AI giants. Tata got a new partnership with OpenAI last month. A few days earlier, Infosys got a new partnership with Anthropic. So what they are trying to do is say: let us use your tools that you are building, and then build on that to sell to our enterprise clients.

But there’s so much happening, the agentic tools are coming at such a fast pace, and again the repetitive work that was once required — that was the bread and butter for many of these companies. The IT service sector is not just giants, there are smaller companies as well. So a lot of the repetitive work was happening at the lower end of the spectrum — those are the jobs that are really threatened. Because even if the giants can sell something, it’s the lower end of the spectrum that will start feeling the pinch very quickly.

Sarah: It sounds like either way, there’ll probably be fewer jobs required in those companies? Even if they don’t lose clients, they might be trying to make their business more efficient with the use of AI?

Krishn: There will be some job losses, I don’t think even the people putting on the bravest face right now can deny that. When we talk to the industry experts they point out that there have already been job losses. At least 20,000 jobs have been lost in the past six months. For a sector that employs about 6mn people, 20,000 is not that a big number, but that shows you there is some rationalisation happening.

Sarah: Does it make the Indian companies quite vulnerable, if they’re building their tools and services on top of the models from the big American companies?

Krishn: There are no large LLMs that India has built. India is building some small ones, some voice models . . . but the sort of LLM tools that enterprises need, India doesn’t have those. There is definitely a vulnerability for the IT services companies.

But that doesn’t necessarily mean there’s a vulnerability for India’s talent pool. Because while the IT services are trying to find what is the best business model for them, and how can they best retain clients, what is happening in parallel is a lot of these top companies are opening their new back offices in India. And these back offices are also doing some frontier research.

Somebody gave me an example of why it is more useful for a company like say McDonald’s to have a back office here. McDonald’s in the US will find it very tough to attract top AI talent, but McDonald’s in India with a good salary can definitely attract good AI talent. You don’t need to be at the frontier of tech to be a tech employer in India. You can be from any field and hire good tech talent at scale in this country. So that’s where the reskilled jobs, many of them, are going. They’re called global capability centres (GCCs). The GCC sector is becoming a massive employer.

John: Can you tell us a bit more about the GCCs? When did that trend begin and is it related to AI or would it have happened anyway?

Krishn: The first GCCs, they were called business process outsourcing back in the day. Initially they were just doing back office work — whether it’s accountancy or HR. But then as more and more of such jobs started getting automated slowly over the last two decades — thanks to machine learning — a lot of that was happening in these back offices. So as companies started realising that we have this tech talent, and if you provide the right resources such as AI frontier models, they can start innovating for us as well in house. They slowly evolved into R&D centres.

There is an Airbus GCC, there is a Boeing GCC, there is a Google GCC and there is a JP Morgan GCC. All these top players. Many of them have thousands of people. Many Indians are coming back from the US offices, many are coming from the IT sector, so they have this talent that is very high skilled. How to leverage this is by giving them more frontier responsibilities than they had earlier.

Sarah: So moving away from the commoditised stuff and into much more skilled work?

Krishn: Yep. It’s moving up the ladder.

So what have we learned?

Sarah: John is on holiday this week, so you’ll have to be content with my reflections I’m afraid. What struck me most from our conversation with Krishn was that we need to update our mental model of the sorts of work which multinationals are offshoring or outsourcing to India. The business model of the country’s IT services companies does seem imperilled (especially at the lower end) but on the other hand, the sorts of roles that multinationals are now hiring for directly in India are much more high-value and interesting than the back-office roles of old. That seems like good news for India’s tech talent, at least for now.

Recommended reading

1. Our colleague Cristina Criddle has been doing a great podcast series on AI and intimacy (Sarah)

2. Dean Ball wrote a powerful essay about Anthropic, the Trump administration and the death of the American republic (also Sarah)