Big technology companies are preparing to invest $725 billion in artificial intelligence infrastructure in 2026 alone, according to new projections. This figure includes major players like Amazon, Alphabet, Microsoft, and Meta Platforms, all of which are increasing their spending compared to last year. The total is up 77% from $410 billion spent just last year, showing a rapid expansion in AI-related infrastructure projects.
IBM CEO Arvind Krishna has expressed skepticism about the scale of this investment, pointing to a mismatch between spending and expected returns, according to 24/7 Wall St.. Krishna discussed his concerns on the In Good Company podcast in May, using a simple calculation involving power and cost to make his point. He noted that 1 gigawatt of AI data center capacity could cost between $60 billion and $80 billion in semiconductors. Global commitments already total around 100 gigawatts, which would require an investment of $6 trillion to $8 trillion. Krishna said that such a level of spending cannot be supported by the revenue growth expected in the AI sector. He emphasized that this isn’t a criticism of AI’s value, but rather a concern about how quickly the industry is investing without matching returns.
Krishna believes that AI models will eventually become commodities, with low switching costs allowing customers to move freely between providers. He thinks that success in the AI space will depend more on how well companies integrate AI into existing services than on building infrastructure alone.
IBM, under Krishna’s leadership, is not investing heavily in AI data centers but instead focuses on consulting and hybrid cloud services. The company’s second-quarter financial results showed a revenue increase of 6% and projected full-year growth in free cash flow. IBM is betting that enterprises will be cautious about locking into large-scale AI infrastructure, which aligns with Krishna’s broader view.
While many companies are chasing the growth in AI infrastructure, not all will benefit equally from this trend.
CoreWeave, a data center startup, reported strong revenue growth in the third quarter despite not yet being profitable, according to a report from CNBC. The company posted $2.58 billion in revenue for the third quarter, with a net loss of $626 million. CoreWeave is one of several companies racing to build data centers capable of running AI models, competing with giants like Amazon and Google. The company has been focusing on renting Nvidia chips, which has helped improve its pricing and margins. Despite rising competition from companies like SpaceX and Meta, CoreWeave continues to see strong demand for its services.
Shares in CoreWeave have risen significantly this year, outperforming the broader market.
Jabil Inc. is also expanding its role in AI infrastructure, though it has a different approach than some of the larger players, Seeking Alpha reported. The company is now involved in more aspects of data center production and support, increasing its reach across the industry. Jabil’s expansion shows how the AI infrastructure market is evolving beyond just building data centers.
Investors are closely watching how these companies perform in the coming quarters as AI spending continues to rise.