For two decades, India's IT services industry ran on a single elegant arbitrage: take work out of expensive Western offices, staff it with well-trained engineers in Bengaluru or Hyderabad, and deliver the same outcome for a fraction of the price. The margin lived in the labour gap. Hardware was someone else's problem — a pass-through line item, rarely the headline on a quarterly earnings call.
That assumption is cracking. As LiveMint reported on July 21, rising memory chip and server hardware costs are forcing IT firms to rethink how they price large contracts — particularly the AI-integrated, data-centre-heavy deals that now define the industry's growth ambitions. At least two chief executives have flagged the issue publicly, and one of them is drawing a line in the sand.
The CEO Who Won't Hold the Price
Tech Mahindra's Mohit Joshi put it bluntly during the company's post-earnings analyst call on July 16. Memory prices and chip prices are rising significantly, he said, and Tech Mahindra will not guarantee those costs to customers. "If you're seeing a pricing inflation of 20% year-on-year, to tell clients that you will hold the price for a three- or a five-year deal, we think is a forward call, which doesn't really make sense," Joshi said. Selling services at yesterday's prices is not on the table.
This is not routine earnings-call caution. Joshi is describing a structural renegotiation of the commercial relationship between Indian IT firms and their global clients — one that the industry has spent years avoiding because holding prices steady was itself a competitive weapon.
LTM Ltd's Venu Lambu is watching the same dynamics from a more immediate vantage point. In January, LTM won a $330 million contract from India's income tax department to modernise its software with AI features. The ramp-up has stalled — not because of project complexity or client readiness, but because of memory chip shortages traced to supply disruptions linked to the West Asia conflict. "Whatever is the delay ramp up, I am expecting it to start ramping up in Q2," Lambu said, "but the larger issue about hardware, shipment timelines and the memory prices is something that we still need to see how to navigate over the next one or two months."
A government contract for AI modernisation, delayed by a geopolitical event in a region thousands of miles away, through a supply chain that Indian IT firms neither own nor influence. The sentence contains the whole problem.
The Playbook That No Longer Works
Large IT deals have always carried thinner margins than a firm's overall business. The industry's standard response was to offset the compression during the contract's execution phase — deploy people in low-cost locations, extract efficiencies over the multi-year term, and recover the early losses as the engagement matures. Hardware was either excluded from the contract scope or treated as a predictable commodity with stable enough pricing to model across a five-year horizon.
Neither assumption holds today. Data centre operators have been stockpiling memory components in anticipation of AI-driven demand, creating shortages that have sent prices sharply higher. The AI deals that Indian IT firms are pursuing most aggressively — the ones that replace legacy infrastructure with GPU-accelerated compute, that embed machine learning into tax systems and banking cores and logistics networks — are precisely the contracts with the heaviest hardware component.
The timing compounds the difficulty. Automation tools are simultaneously compelling IT firms to offer deeper cost savings on smaller contracts, compressing the revenue base from which they might otherwise absorb hardware shocks. Geopolitical pressures push enterprise clients to tighten technology budgets. The industry faces thinner deal economics at exactly the moment its input costs are spiking.
Services Integrators in a Hardware-Owning World
The structural fault-line runs deeper than any single earnings cycle. Indian IT's competitive identity is as a services integrator — the firm that assembles, deploys, and manages technology built by others. That model worked when the technology layer was software-defined and labour-intensive. It grows precarious when the value chain shifts toward infrastructure ownership.
Amazon Web Services, Microsoft Azure, and Google Cloud own their hardware stacks. When memory prices rise, they absorb the shock across an asset base that also appreciates in value. They can cross-subsidise cloud pricing through other revenue lines. An Indian IT firm quoting a fixed-price AI infrastructure deal owns none of that buffer. It is exposed to spot hardware price movements in a contract priced at the time of signing, often 18 months before the hardware actually ships.
Analysts working on IT services pricing have argued that Indian firms quoting flat-rate pricing on AI infrastructure-heavy deals without hardware cost pass-through clauses are absorbing a risk category that traditional IT contracts were never designed to carry. The contracts were designed for a world where the value being sold was human expertise — auditable, scalable, location-portable. Hardware is none of those things.
The Semiconductor Dependency Beneath the Surface
India imports virtually all of the semiconductors that power its digital economy — predominantly from Taiwan and South Korea, with GPU supply concentrated among a handful of manufacturers. Every AI-era IT deal embeds a foreign-currency hardware cost that services revenue, earned in dollars and euros, cannot fully hedge. When that supply chain experiences a shock — a war in West Asia disrupting shipping lanes, stockpiling behaviour among hyperscalers, export controls on advanced chips — the cost lands on Indian IT's income statement.
This is the dimension of the hardware cost story that earnings-call commentary cannot fully capture. It is not a pricing negotiation problem; it is a structural exposure that persists regardless of how cleverly contracts are written. MeitY's IndiaAI Mission has acknowledged the underlying vulnerability through its compute aggregation initiative, which aims to pool GPU resources for shared national use — an implicit recognition that hardware dependency is a strategic liability, not just an operating one. But a shared GPU pool for research and public-sector AI workloads is a different instrument from the private-sector compute capacity that IT firms need to underwrite commercial contracts at scale.
The deeper argument — made with some force by analysts tracking India's semiconductor strategy — is that IT services competitiveness in the AI era ultimately resolves into a hardware sovereignty question. As long as the chips are made elsewhere, every large AI deal that Indian firms sign carries an embedded geopolitical risk premium that no contract clause can fully neutralise.
What the Sector Does Next
The immediate commercial response is already visible: firms like Tech Mahindra are pushing hardware pass-through clauses into new contract negotiations, separating the services margin from the hardware cost line. This is standard practice among US-headquartered systems integrators — firms like Accenture and IBM have long distinguished between their professional services economics and their infrastructure procurement terms. Indian IT firms, accustomed to competing on total-cost simplicity, have been slower to adopt it. The hardware inflation of 2025-26 may accelerate that adoption by making the alternative — absorbing the cost — visibly untenable.
The harder question is whether contract restructuring is sufficient or merely symptomatic. IT exports constitute a significant share of India's GDP and represent its largest source of services trade surplus. A sustained compression of operating margins across the sector's major players reduces the earnings base, dampens the attractiveness of Nifty IT-heavy portfolios to foreign institutional investors, and ultimately translates into pressure on employment in one of the economy's most productive and high-wage sectors. The labour-arbitrage era insulated Indian IT from hardware cycles. The AI era has ended that insulation — and the industry, along with the policymakers who depend on its foreign exchange contribution, is only beginning to reckon with what that means.
