Indian IT Services: Mixed feedback from our IT Trip

研报英文原文证据摘录

Indian IT Services: Mixed feedback from our IT Trip

nd companies want to own IP and capabilities. According to ANSR,

while mature GCC headcount may stagnate, overall growth will be driven by (1)

increasing penetration among Fortune 500 (currently c30–35%), (2) mid-market

adoption, and (3) expansion across geographies (ANZ, Japan, Europe) and new verticals

(retail, QSR, construction). While new GCCs may start leaner and mature ones grow

slower due to AI, the number of GCCs and scope of work should rise.

Token costs ballooning; model orchestration to become critical

A consistent message across our meetings was that enterprises are increasingly facing

ballooning token costs, and are therefore, recognizing that not every use case requires

expensive frontier models; open-source/open-weight models, private deployments and

model orchestration can materially reduce costs while maintaining sufficient accuracy

for many workloads. This suggests a shift from a single-model approach to dynamic

model routing, where cheaper/private models handle most workloads and frontier

models (e.g., Claude/OpenAI) are used selectively. This not only reduces token cost risk,

it also creates opportunities for IT Services in AI FinOps, model orchestration,

governance, private AI deployment and integration.

Overall, read-through for IT Services is mixed

Near-term demand remains soft due to budget reallocation toward AI infrastructure,

tokens, data, cloud, regulatory and cyber spend, alongside weak discretionary spending

(exacerbated by macro/geopolitics). Meetings suggest that GCCs remain a structural

headwind for high-end, IP-sensitive work, with some enterprises increasingly insourcing

AI and engineering capabilities. However, our discussions also suggest that scaling AI