Companies are scrambling to curtail soaring AI costs For the heaviest users, tokenmaxxing is losing its appeal June 18th 2026 “It’s going to be an absolute nightmare,” says an executive at a big American tech company. He is talking about an emerging problem for businesses that use artificial intelligence. AI agents—bots that can read, interpret and act—use masses of processing power and have started to run up huge bills. As they proliferate, the problem will grow. Big companies, the executive points out, typically use hundreds of software programs. If each of those offers agents (as they probably will), AI costs could spiral out of control. Budget management is a new worry for AI adopters. Not long ago employees were encouraged to binge on the technology. Burning through vast numbers of tokens—the chunks of text that models process, which are often used as a unit of pricing—became a badge of honour; techies dubbed it “tokenmaxxing”. Companies showed off staff’s AI use on internal leaderboards. Meta’s display awarded top users titles like “Token Legend”. Such incentives partly explain the boom in AI spending. So does a change in the way enterprises use the technology. Token-heavy applications, such as reasoning models and agents, are growing more popular. In some cases agents build their own agents, sending costs higher still. Ramp, a corporate- credit-card provider, analyses its clients’ transaction data to shed light on their AI use. It reckons their spending has risen 13-fold in the past year. In April Uber said that it had spent its annual AI budget in four months. Other firms have similar problems. One reportedly spent $500m on AI tokens in a month. Sam Altman, the boss of OpenAI, has called mounting customer costs “a huge issue”.
For now, the problem is concentrated. The top spenders tend to be tech firms, because they are early adopters and because AI is particularly good at writing software. Ramp reckons that the 1% of clients that spend the most on AI per employee are racking up average monthly bills of about $7,450 per person; the median client pays $11. And although the big spenders’ AI bills are low compared with the cost of hiring a developer in San Francisco, they are high compared with employing one in Delhi. Companies are responding in various ways. Although AI laggards are still racing to adopt the technology, for the heaviest users tokenmaxxing is out. Some, including Meta and Amazon, have scrapped their leaderboards. Many are thinking more carefully about model choice. Plenty of tasks do not require pricey, cutting-edge models. Aran Khanna of Archera.ai, which helps companies reduce their cloud costs, points out that in some cases Sonnet, a lagging-edge model from Anthropic, can cost a twentieth of what Opus, a leading-edge one, does. And Kimi, an open-source model from Moonshot AI, a Chinese startup, can cost a twentieth of that. Another approach is spending caps. Uber has limited its employees to $1,500-worth of tokens each month per coding tool. How firms choose to allocate tokens depends on where they can generate the most value from AI, says Rachel Laycock of Thoughtworks, a consultancy. That often means companies apportion the most tokens to their core business function, such as engineers in a tech firm, she adds. Some software vendors are experimenting with new pricing plans to ease clients’ concern over costs. Intercom, one such supplier, offers a service whereby the customer pays only for queries that are resolved by its IT- support agent. The big cloud providers have launched cost-management services, too, including budgeting tools and ways to triage queries to the most appropriate model. For the model-makers, the balancing act is trickier. They want their biggest customers to use as many tokens as possible without balking at the cost. Today the cost of providing AI services is subsidised by the labs. Indeed, OpenAI’s plan for winning customers from Anthropic reportedly involves drastic price cuts. But eventually the model-makers will need to turn a profit, which will mean raising prices. Pressure to do so will only intensify later
this year if, as expected, Anthropic and OpenAI go public. Their customers, then, can look forward to more jaw-dropping AI bills. ■ To track the trends shaping commerce, industry and technology, sign up to “The Bottom Line”, our weekly subscriber-only newsletter on global business. This article was downloaded by zlibrary from https://www.economist.com//business/2026/06/14/companies-are-scrambling-to-curtail- soaring-ai-costs
Tata’s big bets are yet to pay off The Indian conglomerate’s boardroom drama highlights a speculative strategy June 18th 2026 Jamsetji Tata, the Victorian founder of his family’s business empire, dreamed big. He had four ambitions: to open a world-class hotel in India; to build the country’s first steel plant; to power Bombay (now Mumbai) with hydroelectricity; and to establish an Indian scientific institute to rival those in Europe. Though sceptics scoffed, all were eventually realised. Today the Tata Group is one of India’s largest conglomerates, selling everything from salt to software. It is still dreaming. But investments in its new ventures—an airline, a digital “super-app”, smartphones and semiconductors—are absorbing vast amounts of capital.
Concerns about capital allocation have contributed to a boardroom dispute. Sixty-six per cent of Tata Sons, the group’s holding company, is owned by Tata Trusts, one of the world’s largest charities, established by the family over many years, which funds health care, education, poverty alleviation and culture. In February Noel Tata, the Trusts’ chairman, withdrew his support for Natarajan Chandrasekaran to have a third five-year term running Tata Sons. The two met to clear the air before a board meeting in May, at which some of the group’s lossmaking new entities gave presentations. At issue are the future of these ventures, whether the company should remain private, how to find an exit for a minority shareholder and a potential succession plan for Mr Chandrasekaran. None of these issues was resolved at the latest board meeting on June 12th. The minority shareholder is the Mistry family, another Indian business dynasty (like the Tatas, from the Parsi religious minority), which owns 18% of Tata Sons via the Shapoorji Pallonji Group, an infrastructure company. The SP Group is heavily indebted and wants to sell its stake. (Noel Tata is married to Aloo Mistry, sister of the late Cyrus Mistry, former chair of Tata Sons. In 2016 Ratan Tata, Noel’s half-brother, who ran the business for many years and died in 2024, forced Cyrus out, replacing him with Mr Chandrasekaran.) Four years ago this newspaper examined “the world’s biggest bet on India”: instead of deploying capital in foreign businesses such as Jaguar Land Rover, a British carmaker, and Corus Steel, an Anglo-Dutch smelter, Tata was refocusing on India’s ascendant economy. The bet is yet to pay off. An analysis by The Economist finds that Tata Sons’ return on capital employed (ROCE), a measure of profitability, has been roughly stable for the past three years, but that its new ventures are dragging performance down. The ROCE of listed businesses, including Tata Motors and Tata Steel as well as Tata Consultancy Services (TCS), a big IT firm, was 25.1% last year, up from 17.9% in 2019 (see chart). TCS typically provides the largest chunk of Tata Sons’ income. In the 2025 financial year, TCS’s dividend made up over 80% of its revenue.
But unlisted new businesses have little to show for the amounts lavished on them so far. Air India, the national flag-carrier privatised in 2022, is making heavy losses. It had been mismanaged in state hands, but the closure of Pakistani airspace to Indian airlines after a brief conflict in May 2025, a fatal air crash last June and the Gulf war’s disruption of jet-fuel markets have added to its woes. Tata Digital, an attempt to build a super-app selling Tata’s consumer range, is being scaled back. The one bright spot has been Tata Electronics, which manufactures iPhones in India alongside Foxconn, a Taiwanese firm. It is now modestly in the black, but will need more investment to embed itself in supply chains for phones and semiconductors. With ASML, a Dutch firm, it is building an $11bn chip factory in Dholera, in western India. Agratas, a battery-making venture, is likewise hungry for capital. The Indian government has provided grants for Tata’s chip and battery factories. A grant from the British government has also contributed to a battery plant in England to supply Jaguar Land Rover. Yet the unlisted entities’ combined ROCE is -5.5%. Worse, TCS is looking less secure. Since the end of 2024 its share price has fallen by more than half, largely on investors’ fears that generative artificial intelligence will undermine Indian IT outsourcers. Its earnings have not yet suffered: operating profit has risen by 12% in the same period. Even so, as
well as turning around Air India and taking its next steps into high-tech manufacturing, Tata Sons must find a strategy for India’s largest IT firm to adapt to AI. Mr Chandrasekaran thus faces competing demands: providing dividends for Tata Trusts, buying out the Mistrys and investing in the new ventures. The difficulty is compounded by another long-running dispute, over whether Tata Sons should remain private. The Reserve Bank of India, the central bank, has demanded that it list on the stock market. An RBI regulation requires non-bank financial corporations, which carry out banking functions but are regulated differently, to be listed. Though the holding company is in the investment business, Tata Sons argues that it is not a bank and does not intermediate between savers and borrowers. It has applied for a waiver, which the RBI has so far denied. A listing would allow both Tata Sons to raise capital for its new ventures and the SP Group to sell up. The company would probably be valued at $120bn- 150bn, even with a conglomerate discount analysts put at around a third. But Tata Trusts resists the idea. A public offering could mean the family loses control and could threaten Tata Sons’ particular role in India’s industrial development. It would, argues N.A. Soonawala, a former vice-chair of the firm, undermine a structure that has evolved over a century. Jamsetji Tata did not live to see all his dreams realised: he died in 1904 with only the hotel completed. The hydro plant did not start up for over a decade. Tata’s modern bosses may not have as long.■ To track the trends shaping commerce, industry and technology, sign up to “The Bottom Line”, our weekly subscriber-only newsletter on global business. This article was downloaded by zlibrary from https://www.economist.com//business/2026/06/18/tatas-big-bets-are-yet-to-pay-off