Coforge Limited and Hexaware Technologies Ltd are leading the industry response to a critical AI cost challenge as businesses burn through annual token budgets in months or even days. According to latest reports from Mint, every client that Coforge knows has burned their annual budget of tokens in months, if not days, forcing the company to rethink its AI pricing and delivery models. The token usage focus comes as prices determine AI usage for companies integrating the technology into their daily operations, with tokens being the fundamental unit of data processed by large language models such as OpenAI's ChatGPT or Anthropic's Claude. For Fortune 1000 companies, running AI tools in large quantities can generate massive token bills, paid either directly to AI vendors or passed through tech services firms.
Coforge Limited has announced an expanded push to help hi-tech enterprises shift from isolated AI features to AI-driven business outcomes across engineering, operations, and decision-making. According to the latest company update, the strategy centers on Coforge Nuuron, a composable enterprise AI operating layer that connects business context, institutional knowledge, workflows, and decision intelligence. Paired with the Momentuum blue operating model, this approach embeds small teams of Forward Deployed Engineers and AI agents within client settings to accelerate the move from pilot projects to production adoption. The company positions this initiative as part of the next phase of AI transformation, where success is measured by business outcomes rather than the mere presence of AI features. During its webinar on "Operationalising Enterprise Autonomy" on 17 September, Coforge's management emphasized managing token consumption through architecture and model choice, rather than raw token usage alone, focusing on cost per unit of work with model mix and the use of open-weight AI models to reduce token costs.
The AI revolution continues to fundamentally reshape traditional IT services revenue models, with Coforge targeting software, semiconductor, and electronics players that need to convert AI investments into measurable gains. As reported by Business Standard, all IT services companies are experiencing pressure as revenue from businesses that have supported them for decades faces strain due to AI-improved productivity requiring fewer people for the same work. Clients are negotiating deals at significant discounts during renewals, demanding that AI benefits be passed on to them. For Coforge, 6-7% of its topline in the first quarter was outcome-based, highlighting the emerging nature of AI-driven revenue streams. The company's outcome-led approach is intended to strengthen its industry positioning as a trusted partner for autonomous, AI-led enterprises.
Hexaware Technologies has implemented a comprehensive token cost management strategy, with each new proposal now carrying a token-based pricing option according to Mint. CEO R. Srikrishna explained that "deflation for our industry equals to token cost" and that "the money that goes to token is going to come off labour spend." The company plans to fix token costs for some outcomes because of clients' lack of knowledge on token costs, and will help clients figure out their actual token expenditure versus original estimates. Pennsylvania-based Epam Systems Inc. has started factoring in token costs directly into deal structures, while Tech Mahindra Ltd is testing pricing models linked directly to client token consumption. Infosys Ltd and Wipro Ltd are ensuring clients use different AI models to optimize token usage, and HCL Technologies Ltd is promoting tiered models where small language models are used to reduce overall token cost. Cognizant Technology Solutions Corp. has brought in dedicated rate cards for AI work, and LTM Ltd. uses a quota-based system where clients purchase fixed AI units.
Industry analysts have praised the strategic moves of both companies in addressing the AI token cost crisis. Ashis Dash, vice-president at Systematix, noted that "Coforge's sovereign AI offering is positioned to capture demand for control at source by improving AI efficiency, selecting fit-for-purpose models and scaling usage according to task requirements." Motilal Oswal Financial Services analysts emphasized that "the ability to price and monetize AI services in a token-cost environment could emerge as a stronger differentiator" than AI revenue disclosure, though they added that "the commercial framework remains at an early stage and will require validation through client adoption over the next two to three quarters." For the financial year 2026, Coforge and Hexaware reported revenues of $1.87 billion and $1.54 billion, up 29.2% and 7.6% respectively, with Coforge growing the fastest among India's 13 largest IT firms getting more than a billion dollars in revenue.
Singh's assessment aligns with broader industry perspectives on AI adoption challenges. According to Business Standard, Cognizant CEO Ravi Kumar has similarly noted that enterprise AI adoption is low due to high token consumption without proper return on investment linkage. Kumar has also suggested that the threat to IT companies from frontier model firms is overblown, as companies require them for higher AI adoption rates. The latest developments demonstrate that regular AI-led deals are hard to come by as clients remain cautious amid macroeconomic uncertainties, making outcome-led approaches increasingly critical for market positioning.
