Artificial intelligence companies are pouring extraordinary sums into massive data centers. One estimate now puts a 100-gigawatt artificial general intelligence effort at about $8 trillion.

International Business Machines Corp. (NYSE:IBM) CEO Arvind Krishna recently told the “Decoder” podcast. He said filling a 1-gigawatt data center costs about $80 billion and called it “today’s number.”

Krishna said a company committing 20–30 gigawatts could face about $1.5 trillion in spending using current costs, telling “Decoder” host Nilay Patel that the estimate reflects announced AI infrastructure plans. Krishna also described the short lifespan of AI chips and said the hardware needs to be used within about five years before being replaced.

Don’t Miss:

Krishna said an $8 trillion investment would require about $800 billion in profit to cover interest payments alone. Patel referenced comments from OpenAI CEO Sam Altman, who has said OpenAI could generate a return on its capital spending.

OpenAI has announced roughly $1.4 trillion in long-term buildout agreements with partners including Nvidia (NASDAQ:NVDA), Broadcom (NASDAQ:AVGO), Oracle (NYSE:ORCL) and Alphabet (NASDAQ:GOOGL, GOOG)), CNBC reported.

Krishna told the “Decoder” podcast that expectations around AGI spending remain beliefs rather than confirmed outcomes. He told Patel the current wave of infrastructure commitments illustrates how quickly companies are scaling capacity for AI, adding that the pace reflects industry activity rather than a guaranteed path to AGI.

Trending: Deloitte’s #1 Fastest-Growing Software Company Lets Users Earn Money Just by Scrolling — Accredited Investors Can Still Get In at $0.50/Share.

Krishna said he gives the current technologies only a “zero to 1% ” chance of achieving AGI. He said today’s large language models do not reach the level associated with AGI and that additional advances will be required. He said current tools already offer significant value in enterprise settings.

Several leaders have questioned AGI projections. Palantir Technologies Inc. (NYSE:PLTR) Chief Technology Officer Shyam Sankar told the “Interesting Times with Ross Douthat” podcast in late October that extreme AGI narratives often function as “a fundraising shtick.”

OpenAI co-founder Ilya Sutskever said in an interview last month on the “Dwarkesh Podcast” hosted by Dwarkesh Patel that scaling current models by “another 100x” would not be transformative and described the direction of progress as a return to the “age of research.”

See Also: Missed Tesla? EnergyX Is Tackling the Next $200 Billion Opportunity — Lithium

Today’s large language models are “a dress rehearsal” for AGI and fall short of the capabilities sometimes implied, researcher Gary Marcus told the Axios AI+ Summit earlier this month.

Krishna told “Decoder” podcast that current AI tools can deliver major productivity gains but said AGI will require combining large language models with forms of hard knowledge. Even with that approach, he said he remains “a maybe” on whether it would be enough.

Read Next: Buffett’s Secret to Wealth? Private Real Estate—Get Institutional Access Yourself

Image: Shutterstock

Up Next: Transform your trading with Benzinga Edge’s one-of-a-kind market trade ideas and tools. Click now to access unique insights that can set you ahead in today’s competitive market.

Get the latest stock analysis from Benzinga:

This article IBM CEO Arvind Krishna Breaks Down Why A 100-Gigawatt AGI Push Could Cost $8T, Says That’s ‘Today’s Number’ originally appeared on Benzinga.com

© 2025 Benzinga.com. Benzinga does not provide investment advice. All rights reserved.