Uber and others go slow on AI, will billions spent by OpenAI, Anthropic go waste?

Google, OpenAI, Anthropic, Meta, Amazon and Microsoft are investing billions of dollars in AI. But of late, people have started asking the question: Is it a bubble? The trigger behind the question is the rising cost of AI that is forcing even top companies in the world to go easy on it.

In Short

• Google, OpenAI, Anthropic are spending billions of dollars on AI
• At the same time, AI has become too expensive for users and companies
• Experts say tech giants will earn returns on their money, but in long term

Earlier this week, Google announced it was going to raise $80 billion through equity, the first time the company is doing so since 2006. Reason? It needs money to invest in AI. It is the same across other tech giants. They are all scrambling and scrapping the barrel to find money they can invest in AI. This year just four — Google, Meta, Amazon, and Microsoft — will in total invest close to $700 billion in AI data centres and AI infrastructure. At the same time, AI labs such as OpenAI and Anthropic are burning money.

For over a year now, many have called this spending a sign of an AI bubble. That is because while the money is getting invested, returns are not there yet. At the beginning of 2026, it seemed that the bubble proponents would be proved wrong as Claude surged in popularity and use. But a few months in, now we are seeing a different trend.

AI has suddenly become too expensive for companies and users. With token-based billing arriving as the AI companies search for revenue that justifies their spending, even big companies like Uber and Microsoft are restricting the use of AI by their employees. So much so that this week OpenAI CEO Sam Altman seemed surprised at how quickly companies were pulling back on AI use. “It's kind of a meme now (that) my company spent my entire 2026 budget in Q1,” he said. “All of a sudden (AI costs) are a huge issue.”

In a recent study, Bain & Co found that 44 per cent of large firms that are funding more AI investment are basing them on the savings made from the previous round of expenditure. But those savings “haven’t yet materialised for some of them.”

This brings up a question, or rather a dichotomy. If the AI companies and tech giants continue pouring money into data centres etc, and if the users — enterprise as well as personal — find AI too expensive, how will the cost and revenue align? And will they ever align?

Deja vu and a vision for future

So, what happens to all of this investment? Will billions of dollars go down the drain? Amit Das, founder and CEO of Think360.ai, says, “Some of it almost certainly will. Always does.”

Das believes that the situation around AI could be similar to how fibre-optic cables were laid down decades ago. He tells India Today Tech, “Every major technology wave creates over-investment. During the dot-com era, billions were spent on fibre-optic networks. Many companies disappeared, but the infrastructure remained and became the foundation of today's internet economy.”

AI companies are possibly focusing on building long-term infrastructure – think AI data centres – that would start producing returns only after a few years.

Ambika Sharma, chief strategist at Pulp Strategy, concurs. “The AI infrastructure being built today is the same (fibre cables) class of investment. The returns will not all come from the companies building it. They will come from everything that runs on top of it,” she tells India Today Tech.

The long-term is what it is. Or as celebrated economist J M Keynes said, “In the long run we are all dead.” In the short term, though, experts expect teething pain for tech giants that are spending billions of dollars. That is because the AI market is beginning to go through a maturing phase.

End of an experiment, results expected

Vishal Sirohi, CEO and co-founder at Island Computing, says the pulling back from AI by some companies is a sign that AI is now no longer the next big thing, but rather, just another piece of tech.

“The apparent contradiction is only temporary. Over the last two years, AI adoption was heavily subsidised by model providers and investors,” he says. “As enterprises scale usage, they are measuring AI against the same standard as any other technology investment – revenue growth, productivity gains, cost reduction, and business outcomes.”

Essentially, companies are now getting real about AI and looking at big words like ROI and KPAs. Sudeepta Veerapaneni, chief innovation officer at Deloitte South Asia, says now we are past the phase of experimentation with AI.

“The next phase is about economics, scale, and measurable business outcomes,” Sudeepta tells India Today Tech. “Enterprises are no longer asking how many AI tools they can deploy, rather they are asking how AI can improve productivity, accelerate decision-making, optimise costs, strengthen customer experience, and create new sources of value.”

For a couple of years now, AI tools have been hyped up as magic. Now they have become more mainstream and can no longer work their charm on the basis of their exotic and mystical qualities. Instead, they have to prove they can work as well as humans, preferably at a lower cost. And as companies start scaling AI across their workforce, they are looking for ways to better optimise costs, particularly now when OpenAI and Anthropic have started charging for AI use on the basis of token consumption.

The changing economics has somewhat dented the immediate appeal of AI at workplaces. But Ambika Sharma believes this is a transition necessary for the long term. “We are seeing a market maturing in real time. Uber burned through its entire 2026 Claude Code budget in four months. Walmart shifted from unlimited AI token access to fixed per-employee allocations,” she says. “These are not companies retreating from AI. They are companies learning to govern it. There is a difference.”

According to experts, the promise of AI remains intact. But early 2026 is showing that this promise will be realised only after a few years, as the industry matures on both AI demand as well as AI supply side. This also means that in the process some companies will lose a lot of money. The winners are not guaranteed irrespective of how much money tech giants spend.

Sudeepta puts it succinctly as she says, “The future of AI will not be defined by who spends the most on AI. It will be defined by who can turn AI into an enduring organisational capability that creates measurable business advantage.” She highlights that to help companies navigate the messy short-term transitions that her firm has created GenW.AI, a tool that can make deployment of AI at workplaces more methodical.

For tech giants like Google, OpenAI, Meta and others, there is no easy way out. They have to punt and then hope for the best because they have been caught up in a moment where sitting out may mean losing badly in the long-term. Meta boss Mark Zuckerberg highlighted this predicament last year. “If we end up misspending a couple of hundred billion dollars, I think that that is going to be very unfortunate obviously,” Zuckerberg has said. “But what I’d say is I actually think the risk is higher on the other side.” For now tech giants have chosen to risk their short term profit while hoping for rewards in the long term. Whether it works out for them or not is something we will get to know in 2032.

- Ends