Zoho’s Sridhar Vembu backs IBM chief’s AI bubble warning, says firm won’t chase data centre spending spree

Zoho founder Sridhar Vembu has endorsed concerns raised by IBM CEO Arvind Krishna over the scale of investment flowing into artificial intelligence infrastructure, saying his company intends to prioritise foundational AI capabilities rather than participate in what he described as an investment-driven boom.Vembu's comments came in response to recent remarks attributed to Krishna, who warned that the multi-trillion-dollar buildout of AI data centres could prove unsustainable if demand fails to match the industry's ambitious projections. Reacting to a social media post discussing Krishna's views, Vembu said Zoho would continue investing in areas that directly improve AI systems, including data curation, reinforcement learning and compiler infrastructure that can help verify AI-generated outputs."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 wrote. The debate centres on the enormous sums being committed to AI infrastructure by technology companies seeking to support increasingly powerful models and pursue advances in artificial general intelligence (AGI). The debate gained attention after IBM CEO Arvind Krishna questioned whether the scale of planned AI infrastructure spending could be justified by future demand. Technology author David Linthicum echoed those concerns, while Zoho founder Sridhar Vembu said his company would focus on building AI capabilities rather than pursuing large-scale infrastructure investments.Technology author David Linthicum, whose post prompted Vembu's response, argued that the sector may be repeating patterns seen during previous technology booms, including the dot-com era, blockchain investments and the metaverse push. According to Linthicum, companies risk assuming that large infrastructure investments will automatically translate into sustainable demand and profitability. Drawing on estimates attributed to Krishna, Linthicum noted that a single one-gigawatt AI data centre can cost roughly $80 billion to build. At that scale, companies pursuing tens of gigawatts of computing capacity could collectively spend trillions of dollars on infrastructure.Krishna reportedly argued that industry-wide commitments tied to AGI development could eventually require around 100 gigawatts of capacity, implying capital expenditure of approximately $8 trillion. Such investments would need to generate hundreds of billions of dollars in profits simply to support financing costs. Supporters of the spending wave contend that AI adoption is still in its early stages and that future demand for advanced computing power could justify current investments.Against this backdrop, Vembu's comments reflect a more cautious approach to AI development. Rather than focusing on large-scale infrastructure, he suggested that long-term value will come from improving the quality, reliability and verifiability of AI systems.Also read: Ubisoft co-founder Claude Guillemot dies at 69 in plane crash