<div id="readability-page-1"><p>IBM (IBM) Chief Executive Officer Arvind Krishna recently shared concerns about the long-term financial return of large investments in AI. In an interview with the Decoder podcast, he said that building and operating the AI infrastructure needed to reach major goals like artificial general intelligence, or AGI, could end up being far more costly than many expect. At the current pace, industry-wide capital expenditures could reach as much as $8 trillion.</p><h3>Claim 50% Off TipRanks Premium and Invest with Confidence </h3><ul><li>Unlock  hedge-fund level data  and  powerful investing tools  designed to help you make smarter, sharper decisions  </li><li> Stay ahead of the market with  the latest news and analysis  so your portfolio is always positioned for maximum potential</li></ul> <p>According to Krishna, filling just one gigawatt of AI data center capacity may now cost around $80 billion. That estimate includes advanced chips, supporting hardware, and power. Several companies are planning to build 20 to 30 gigawatts worth of capacity, which could result in costs above $1.5 trillion. He also said most of this equipment may need to be replaced every five years, making the reinvestment cycle both fast and expensive.</p> <h2><strong>AI Infrastructure Expansion and Investor Response</strong></h2> <p>Meanwhile, companies across the sector are continuing to grow their AI footprints. OpenAI <strong>(PC:OPAIQ)</strong>, which works closely with Microsoft (MSFT), has entered into long-term infrastructure agreements worth around $1.4 trillion. That includes a $300 billion deal with Oracle (ORCL) for cloud services and separate partnerships with Broadcom (AVGO) and Nvidia (NVDA), each planning for 10 gigawatts of capacity.</p> <p>Alphabet (GOOGL) recently raised its capital spending forecast for 2025 to between $91 billion and $93 billion, up from $85 billion earlier. Amazon (AMZN) is also expecting to spend around $125 billion in 2025. Overall, industry-wide AI infrastructure investment is projected to hit $380 billion this year.</p> <p>This wave of spending has drawn a mixed response from the market. Recently, Oracle shares fell 13% after reports of delays at OpenAI's data centers, though the company was quick to reject the report. Broadcom also dropped more than 11% after posting earnings that showed lower profit margins on its custom AI chips.</p> <h2><strong>Implications for Investors</strong></h2> <p>Krishna's remarks come at a time when capital allocation is under more scrutiny. He believes that current language-based AI models may have a &#34;zero to 1%&#34; chance of reaching AGI without further breakthroughs. In his view, meaningful progress in this space would need to combine today's language models with more complex forms of knowledge. He also stated that generating adequate yearly profit to justify the capital needed would be a challenge. Using his numbers, a total investment of $8 trillion would require around $800 billion in profit each year just to cover the cost of that capital.</p> <p>While many companies expect AI to unlock major gains in productivity and automation, there is still uncertainty about when, or if, those gains will match the spending. UBS (UBS) data shows that financing agreements for AI data centers have jumped to $125 billion in 2025, up from $15 billion during the same period last year. Around half of the total $1.5 trillion needed through 2028 may come from private credit markets.</p> <p>We used TipRanks' Comparison Tool to align all the big AI spenders to gain a broader perspective on each of the companies and the AI industry as a whole. In the list you'll find MSFT, GOOGL, META, and AMZN.</p></div>

AMZN

OpenAI

<p>IBM CEO Arvind Krishna questions the financial returns of AI investments, highlighting the potential $8 trillion industry-wide expenditure for AI infrastructure. He doubts current AI models can achieve AGI without breakthroughs, suggesting a "zero to 1%" chance. Companies like OpenAI, Alphabet, and Amazon are increasing AI spending, but investor response is mixed due to high costs and uncertain returns. Krishna emphasizes the challenge of generating profits to justify such capital investments.</p>

- IBM CEO Arvind Krishna expressed concerns about the high costs associated with AI infrastructure, estimating that industry-wide capital expenditures could reach $8 trillion.  
- He highlighted that filling one gigawatt of AI data center capacity may cost around $80 billion, with several companies planning significant expansions that could exceed $1.5 trillion.  
- Krishna noted the challenge of generating sufficient profits to justify these investments, as current AI models may have a low chance of achieving artificial general intelligence without further advancements.

Zero to 1% Chance": IBM CEO Questions AGI Returns on $8 Trillion AI Push