Generative AI is splitting India’s offshore technology sector into two divergent halves. Traditional IT services and business process work, the mass-employment engine that built the industry, faces shrinking contracts and hiring freezes as multinational clients automate routine coding, data entry, and support. Meanwhile, Global Capability Centers — captive engineering hubs owned by firms like Google and Oracle — are expanding, with GCC revenue projected to cross US$60 billion as they absorb the country’s top talent.
India’s Nifty IT index has lagged the broader Nifty 50 over the past 12–18 months. The split is not about whether AI helps — it is about who captures the value it creates.
A code assistant now writes the function a junior engineer in Bengaluru used to bill three hours for. It writes it in seconds, and a human checks it. That single change — from doing the work to supervising it — is quietly dismantling the economics that made India the world’s back office for three decades.
The work that built the industry was never the high-skill kind. It was the routine cognitive labor: simple coding, form processing, first-line customer support, sold cheaply and at enormous volume. That volume is precisely what generative AI eats first. The same tools, deployed inside captive corporate centers, are making a far smaller group of elite engineers more valuable than ever.
So the question is no longer whether AI changes Indian outsourcing. It is which half of the industry survives the change — and whether the half that does still belongs to India.
ALSO WORTH READING
The cheap half is the vulnerable half
Start with what the technology actually does. Generative AI models train on huge volumes of code, documents, and chat logs, then predict what should come next. A support bot drafts a reply; a coding assistant suggests and repairs whole functions; a document bot pulls fields from an invoice. Humans still govern the output. They just need far fewer people per task.
That hits one tier hard and lifts another. The traditional IT-BPM sector — the IT services and business process work performed by the large Indian majors — is built on selling routine labor at scale. NASSCOM, India’s industry body, segments its revenue figures by service line rather than by vendor type, so the often-quoted 75-25 split between outsourcers and captive centers is not something the official data cleanly confirms. What the data does show is divergence in trajectory.
Revenue at the large IT exporters has slowed as Western clients pursue AI-led efficiency, and India’s Nifty IT index has underperformed the broader market across the past year and a half. The contrast is sharp. GCC revenue is projected to cross US$60 billion, with consulting firms reporting strong pipelines for new engineering, data science, and AI platform centers.
Here is the line most coverage misses. Within twelve to eighteen months, the firms that win will not be the ones selling the most AI-assisted hours — they will be the ones that own the platform, the data, or the intellectual property the hours run on. Volume was the old moat. It is now the exposure.
Value is migrating into a few foreign-owned silos
The mechanism is concentration. As AI compresses margins on commoditized contracts, capital flows toward whoever controls the higher-value workflow — and increasingly that is the captive center wired directly into a Western headquarters, not the third-party vendor.
The competitive map makes the stakes plain. US hyperscalers — Microsoft, Amazon, Google — own the foundational models and cloud platforms. Indian majors like TCS, Infosys, and Wipro are positioning as large-scale integrators of those tools, not owners of them. China’s Baidu and Alibaba stay focused on their home market. The contract terms compound the dependence: outsourcing work executed in India must often meet the EU’s binding AI Act obligations, even though India itself governs AI through light-touch, voluntary guidance.
For Western firms, the echo is familiar but the scale is not. Generative AI is pressuring mid-tier software and support roles in the US and EU too. The difference is concentration: in Western markets displaced workers move across sectors, while in India whole campuses are tied to legacy outsourcing demand. So the question from the opening resolves uncomfortably. The half of the industry that survives is thriving — inside corporate centers India hosts but does not own.
Beyond the headline
The bigger picture
This restructuring is one instance of a wider shift: in AI-era services, value concentrates in a small number of high-skill hubs tied to proprietary platforms. Instead of many vendors competing on cheap labor, a few ecosystems — cloud providers plus elite engineering centers — start to dominate. That makes corporate AI strategy a genuine macroeconomic lever.
The money trail
Capital is moving toward whoever can own data, IP, and upstream workflows. That favors GCCs and product-engineering teams integrated with Western headquarters, where automation savings get reinvested into new platforms. Traditional outsourcers risk becoming thin-margin implementation arms unless they can capture a slice of those value pools through their own platforms or co-innovation deals.
What isn’t being said
Most discussion fixates on near-term job counts. Far less is said about bargaining power. If a growing share of critical AI and software work sits inside US or EU corporate silos on Indian soil, domestic policymakers may struggle to steer national digital priorities without risking capital flight or the relocation of those centers.
What the split changes for you
With slowing IT-services revenue colliding against booming GCC pipelines, the people exposed to this shift face concrete decisions now.
- Western firms with India delivery contracts
Map where your AI compliance obligations actually sit. Work executed in India still has to meet EU AI Act standards, so audit whether your vendor or captive center can demonstrate that — not just claim it. Treat concentration as a risk: disruption at a single Indian campus can ripple through your delivery timelines faster than a diversified vendor base would.
- Indian tech professionals
The premium is shifting from production to supervision and high-order engineering. Competition for GCC roles is intense because those centers absorb top talent across the ecosystem. Watch NASSCOM’s next Strategic Review and the FY2024–25 export figures: if services revenue keeps slowing while GCC headcount climbs, the rebundling of work is structural, not cyclical.
- Investors in IT services exporters
The Nifty IT index has lagged the Nifty 50 for 12–18 months on margin concerns. Track whether the large majors move beyond reselling AI-assisted hours toward owning platforms or IP. Firms that stay headcount-priced in an automation market are the ones most exposed.
Explainer
- Generative AI
- Software that produces new text, code, or images by predicting what should come next, based on patterns learned from massive training data. In outsourcing it drafts replies, writes functions, and extracts data from documents, cutting the people needed per task. The shift it forces is subtle: junior roles move from creating output to reviewing and governing it.
- GCC
- A Global Capability Center is a captive offshore facility owned and run by a multinational for high-value engineering and product development. Unlike third-party outsourcers, GCCs keep the work, data, and intellectual property inside the parent company. Their revenue in India is projected to cross US$60 billion, and they increasingly absorb the country’s most sought-after engineers.
- NASSCOM
- The National Association of Software and Service Companies is India’s main IT industry body, representing software and services firms. It publishes the closely watched annual Strategic Review and IT-BPM export figures used to gauge sector health. Its data segments revenue by service line rather than by vendor type, which is why a clean outsourcer-versus-GCC revenue split is hard to confirm.
- IT-BPM
- IT-BPM stands for information technology and business process management, the umbrella term for India’s combined software services and process-outsourcing sector. It covers everything from application development to customer support and back-office processing. The lower-value, high-volume slice of this sector is the part generative AI automates most quickly.
- AI Act
- The EU’s binding regulation setting obligations for high-risk and generative AI systems, the first comprehensive law of its kind. It applies based on where a system is used, not only where it is built, so work delivered from India for EU clients must comply. India, by contrast, relies on light-touch, voluntary guidance, creating a compliance gap inside cross-border contracts.
