Zoho founder Sridhar Vembu has weighed in on IBM CEO Arvind Krishna's warning that the multi-trillion-dollar AI data centre buildout could be a bubble, saying his company would focus on building core AI capabilities rather than chasing what he described as an investment frenzy. Responding to a social media post highlighting Krishna's remarks, Vembu said: "IBM CEO Arvind Krishna says the multi-trillion-dollar AI data centre buildout is a bubble." He added that Zoho is investing in capabilities such as data curation, reinforcement learning and, most importantly, compiler infrastructure to ensure AI-generated outputs can be verified. "We will not chase the investment bubble. This is just our normal prudence. To some people that would sound defeatist, but we will talk in five years," Vembu said. Vembu was reacting to a post by technology author David Linthicum, who endorsed Krishna's concerns over the massive spending on AI infrastructure.— svembu (@svembu) In the post, Linthicum argued that the industry is repeating a familiar pattern seen during the dot-com, blockchain and metaverse booms, where companies assume that building large-scale infrastructure will automatically generate corresponding demand and profits. Citing Krishna's calculations, Linthicum noted that a 1-gigawatt AI data centre currently costs around $80 billion to build. According to the estimates, a company investing in 20-30 gigawatts of capacity could spend nearly $1.5 trillion, while total industry commitments aimed at pursuing artificial general intelligence (AGI) could reach about 100 gigawatts, translating into roughly $8 trillion in capital expenditure. To justify such investments, the industry would need to generate about $800 billion in profits merely to cover financing costs, Krishna reportedly argued. Linthicum contended that current enterprise adoption remains slow, consumer AI is still evolving, and demand may not be sufficient to support the scale of infrastructure being planned. He warned that the sector could eventually face write-downs, strategic pivots and project shutdowns if expected demand fails to materialise.