Accenture’s $29 Claude cap raises a question for Indian IT: Who will pay for rising AI costs at work?

Accenture’s reported decision to limit its centrally funded Claude AI usage to $29 (roughly ₹2,800) per employee per month has raised a broader question for the technology industry about who should pay when workplace AI becomes expensive to run. According to a report by 404 Media, effective October 1, employees exceeding the allocation must identify a project or team budget to cover additional usage.

Some employees reportedly incurred monthly costs of $500 to $1,000, particularly through intensive use of Claude Code. The move does not amount to a blanket ban on the tool, but signals a shift towards tighter spending controls. For India’s IT services industry, where companies are integrating AI into software development and client projects, the issue extends beyond employee subscriptions to how businesses calculate, manage and recover AI costs.

MUST READ | Meta launches Muse Spark Open model to run on Mac Studio: How it can benefit users

Indian IT firms are rethinking AI costs

The challenge is already visible among Indian technology services providers. A September report by Mint said Coforge was helping clients manage token consumption through model selection and architecture, while Hexaware was adding token-based pricing options to new client proposals.

Coforge has argued that businesses should focus on the cost per unit of work delivered, rather than simply measuring token consumption. Hexaware, meanwhile, has indicated that clients could account for token costs separately in commercial proposals.

DON'T MISS | Google announces $15 million for Indian MSMEs, expands free AI training for teachers

These approaches highlight how AI expenses could become another budget item alongside employee costs, software licences and cloud infrastructure. However, they do not show that Indian IT companies are introducing employee-level spending caps like Accenture’s reported arrangement.

Why AI usage can become expensive

Tokens are the units of data AI models process to understand prompts and generate responses. Intensive coding tasks and multi-step AI agents can consume far more computing resources than routine chatbot queries.

The Financial Times reported on October 15 that companies are exploring measures such as using smaller models for simpler tasks, routing requests to more cost-effective models and tracking AI spending against productivity gains.

For IT services firms, the commercial question is whether the time saved through AI justifies its usage costs. Companies may absorb the expense, allocate it to client projects or incorporate it into pricing arrangements.

As enterprise AI adoption expands, access to powerful tools alone may no longer be enough. Businesses will also need to determine whether AI-generated work delivers enough value to justify the bill.