A potential artificial intelligence bubble will deflate faster than earlier tech cycles but ultimately give way to an even stronger rebound once enterprise adoption aligns with the enormous infrastructure build-out underway, according to NTT DATA Inc. chief executive Abhijit Dubey.
Dubey said in an interview with the Reuters Global Markets Forum that even though concerns about supply chains and overheated valuations are rising, the overall trajectory of the technology remains unmistakably upward. He said there is absolutely no doubt that, in the medium to long term, AI will remain a dominant global trend.
“Over the next 12 months, I think we’re going to have a bit of a normalization … It’ll be a short-lived bubble, and (AI) will come out of it stronger,” he said.
Register for Tekedia Mini-MBA edition 19 (Feb 9 – May 2, 2026): big discounts for early bird.
Tekedia AI in Business Masterclass opens registrations.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab: From Technical Design to Deployment (next edition begins Jan 24 2026).
The tech industry is grappling with unprecedented demand for compute capacity. Dubey said supply chains are almost fully spoken for for the next two to three years, with pricing power shifting toward chipmakers and hyperscalers. Their valuations have stretched across public markets, echoing the kind of concentrated exuberance seen in earlier cycles.
The broader environment shows why bubble talk is spreading. AI has triggered the biggest technological shake-up since the early internet era. The scale of capital flowing into data centers, model training clusters, power projects, and high-end chip manufacturing has reached trillions of dollars. Yet the revenue model for many players is still thin.
Companies supplying the infrastructure are booming, but many downstream firms have yet to translate user interest into meaningful profit. Even the largest AI developers report rising costs outpacing monetization, fueling the belief among investors that spending has run too far ahead of earnings. For some market strategists, this mismatch is a classic early-bubble symptom.
Dubey said that despite the unease, the long-term fundamentals remain robust. He highlighted that AI has already caused shortages of memory chips, attracted attention from regulators, and raised questions about labor markets. His firm has begun rethinking recruitment as AI reshapes global employment structures. He said there will clearly be an impact and that over a five- to twenty-five-year horizon, there will likely be some dislocation. Even so, he said NTT DATA is still hiring across multiple locations.
“There will clearly be an impact … Over a five- to 25-year horizon, there will likely be dislocation,” he said.
The labor shift was a major theme at the Reuters NEXT conference in New York, where speakers discussed how AI could reshape work and job growth. May Habib, CEO of Writer Inc., said customers are increasingly focusing on slowing the pace of hiring. She described conversations with clients who immediately think about reducing headcount once they adopt new AI systems, recounting a CEO who asked how soon he could cut thirty percent of his team.
“You close a customer, you get on the phone with the CEO to kick off the project, and it’s like, ‘Great, how soon can I whack 30% of my team?’,” she said.
Yet adoption patterns are more uneven than many executives acknowledge. A PwC survey of the global workforce released in November said daily use of generative AI remains significantly lower than what corporate leaders claim internally. Workers who have AI skills are earning far more, with average wage premiums hitting fifty-six percent, more than double last year. But access to training is uneven. PwC said about half of non-managers reported access to learning resources compared with nearly three-quarters of senior executives.
That contrast reinforces concerns that organizational readiness is lagging far behind investment levels.
This gap between spending and usage is at the heart of the bubble argument. Power-hungry data centers, GPU clusters, and model-development budgets are expanding at a pace usually associated with markets that already have mature demand. Instead, enterprises are still testing pilots, rolling out small-scale deployments, and debating return-on-investment metrics.
Investors have started to question how long companies can keep pouring in cash before they see widespread revenue payback. Some funds have already trimmed exposure to high-cost AI builders because profit margins remain narrow or negative.
At the same time, regulators are sharpening their focus. The escalating energy demands of AI infrastructure have been flagged by policymakers across major economies, and global antitrust agencies are examining chip supply chains and cloud dependencies. Heightened scrutiny adds another layer of uncertainty at a moment when valuations are already stretched.
Even so, Dubey’s view points to a cycle unlike the dot-com bust or the crypto downturn. He believes the correction will be shorter because AI is already embedded in business planning across industries and sits at the center of long-term digital transformation strategies. Once the supply bottlenecks ease and enterprises scale their deployments, he expects revenue models to catch up.
That rebound, he suggested, will be stronger and more durable than the early-phase gains that have characterized the latest boom.