A research communication from financial giant Goldman Sachs warns that large technology companies are increasing capital expenditures to drive the development of generative artificial intelligence, but have yet to demonstrate a sustainable business model. The investment bank estimates that about $1 trillion will be spent on data centers, semiconductor, grid upgrades, and other AI infrastructure in the coming years.
Image Source Note: The image is generated by AI, provided by image authorization service Midjourney
The report states that even with the emergence of so-called "killer applications," it is unclear whether generative AI can deliver the financial returns investors expect. At the same time, countries like the United States, which have been leading the technology process, are now facing hardware shortages. More concerning is that power constraints and shortages may require massive grid transformations in the country.
Jim Cavigliano, head of global equity research at Goldman Sachs, said: "So, the key question is: What will AI solve for the $1 trillion problem? Replacing low-wage jobs with extremely expensive technology is the opposite of the technological transformation I've been closely monitoring in the tech industry for nearly 30 years. Many people are trying to compare today's artificial intelligence with the early stages of the internet. But even in its early stages, the internet was a low-cost technology solution that made e-commerce possible, replacing expensive existing solutions."
Cavigliano added that due to the complexity of building AI chips, coupled with NVIDIA's market dominance, there is no guarantee that costs will naturally decrease. "The market is too complacent about the certainty of cost reductions."
Daron Acemoglu, an economist at MIT, said, "In the next 10 years, only about a quarter of AI-related tasks will be cost-effective, meaning AI will only affect less than 5% of all tasks." He also believes that we cannot yet determine whether AI models will become cheaper over time. He also estimates that AI will only increase the productivity level in the United States by 0.5%, while increasing GDP growth by 0.9%.
Cavigliano continued: "If significant application cases do not become more apparent within the next 12 to 18 months, investors' enthusiasm may gradually wane. But it is more important to focus on corporate profitability. Continuous corporate profitability will enable negative return projects to continue experimentation. As long as corporate profits remain strong, these experiments will continue. Therefore, we expect that companies will not reduce spending on AI infrastructure and strategies before entering the more difficult stages of the economic cycle."
Key Points:
📌 It is estimated that about $1 trillion will be spent on data centers, semiconductors, grid upgrades, and other AI infrastructure in the coming years.
📌 Goldman Sachs warns that despite the substantial investment in the development of generative AI, no sustainable business model has emerged yet.
📌 Countries like the United States are facing hardware shortages, power constraints, and shortages, which may require massive grid transformations.