You can’t spell chai latte without AI. That will hurt India

The coffee chain is tapping artificial intelligence to develop in-house alternatives to systems by Microsoft and IBM that track inventory and manage equipment, Bloomberg News reported last week, after reviewing an internal presentation. According to the article, the Seattle-based company has been working for several years to replace Oracle’s point-of-sale system.

This will be disturbing news in Bengaluru and Hyderabad: Maintaining these very technologies for large multinationals like Starbucks is the bread and butter for the 6 million coders employed by India’s outsourcing industry.

The AI adoption craze is looming over what’s promising to be another lackluster earnings season for IT services exporters. Last week, Tata Consultancy Services Ltd., the biggest among them, reported 0.4% growth in revenue over the previous three months after stripping out currency fluctuations, the slowest expansion in a year. While the company has shed 3% of its workforce in the past year to about 594,000, the spending on third-party specialist contractors to bridge the firm’s own skills gaps ate into revenue. Net profit margin shrank.

At smaller rival HCL Technologies Ltd., sales in the three months to June slipped 0.5% quarter-on-quarter after holding exchange rates constant. The management kept its annual revenue growth guidance of 1% to 4% unchanged, but it still ended up shrinking its employee base by nearly 3,300 people — the sharpest contraction in close to two years. For HCL Tech, too, a rise in subcontractor costs mitigated the wage savings.

Live Events

Bloomberg

For 25 years, India’s software services firms have locked global corporate clients into lucrative contracts to implement and maintain packaged software. Before the arrival of AI tools, it wouldn’t have been cost-effective for a firm like Starbucks, whose business is beverage, to take an IBM system out of its shrink wrap and map it to every piece of kitchen equipment, maintenance schedules, and local technicians across a labyrinthine network of 40,000-plus stores globally. That’s the kind of stuff around which Indian IT vendors have built a $250 billion exports powerhouse.
Similarly, making sure that a multinational can safely add a new local payment method — or correctly reflect a discount or tax change — has been a lucrative annuity for Indian programmers. They specialize in testing for various scenarios that could make the cash registers go down even for a minute. Largely hidden from public view, they keep global supply chains working 24×7 by managing the data pipelines that sync third-party inventory tools with an enterprise’s own resource planning software.To be sure, these long-term, multimillion-dollar orders haven’t completely dried up. TCS shares jumped Monday after the company disclosed that it would be expanding the role it has played in managing the infrastructure and applications for ABB. The new mandate is to design and run the Zurich-based engineering giant’s network as a modern, AI-driven service. HCL Technologies recently won a 5.5-year, $1.14 billion contract to build an AI-driven operating model for a large European engineering and manufacturing conglomerate it didn’t name.

Still, the pricing of large outsourcing deals in the age of AI remains under a question mark. After all, clients will fully expect their suppliers to use fewer humans — and more AI — to keep their tech infrastructure running smoothly. Accordingly, they will pay them less than before.

As for customers embedding artificial intelligence in their own workflows, they’ll probably pay the upfront cost of gathering the unstructured data scattered around their firms and labeling everything correctly. But after a quarter or two, AI agents will use the cleaned-up data to write their own code. The annuity business will have a slow fade, with lumpy AI-related work helping to mask the decline for some time.

Bloomberg

Worse, as clients like Starbucks open their own direct engineering hubs in places like Bengaluru and Nashville — using AI to let small, in-house teams do the work of large code-writing armies — the middleman’s markup becomes an obvious target for cost-cutters.

While the stock market is still giving a thumbs up to any order wins, the NSE IT Index finished June 10% lower than five years ago. Even during the worst of the Global Financial Crisis, pessimism didn’t run this deep. Maybe the gloom is overdone, and US clients will eventually curb their enthusiasm for AI. They may come to realize that even as their token budgets go through the roof, their corporate data and workflows are slipping out of their control and going to frontier AI labs.

However, it’s also possible that investors have read the tea leaves right, and it’s the outsourcing firms that are yet to wake up and smell the coffee.