Morgan Stanley’s report states that the demand for computing capacity will exceed supply. There is a possibility of market fragmentation due to the AI policies of America and China. The decline in AI stocks is due to technical reasons, not fundamentals that have weakened.
According to a report by Morgan Stanley, demand for computing capacity is expected to greatly exceed supply for many years to come. At the same time, the growing possibility of interference in AI policy in both the US and China could deepen the fragmentation of the global AI market.
Technical decline in the market, fundamentals are strong
On the sharp decline in AI infrastructure stocks in recent weeks, Morgan Stanley believes that the major reason for this weakness is market and technical factors rather than the decline in fundamentals. The report said, ‘We want to clarify that we believe that a major reason for weakness is technical, not fundamental.’
It further says that enterprise limits on token spending will not significantly impact AI revenues. Morgan Stanley highlighted that companies can impose limits on the use of computing resources by employees, as has been seen in some high-profile cases. However, the data suggests that this is unlikely to be a significant problem. According to the report, the average monthly token spend by enterprise users is currently very low, at less than US$11.
Mathematics of Competition and Cost
At the same time, the economics of using AI remain highly attractive, as AI can save approximately $55 in labor costs for every $2-3 spent on tokens. ‘Companies that fail to adopt the most beneficial AI capabilities will face significant competitive disadvantages, and we expect this situation to become more pronounced over time,’ the report said.
However, Morgan Stanley warned, ‘Chinese open-weight models could become a major competitive threat to American frontier LLMs (Large Language Models), leading to less spending on computing to train LLMs.’ It further added that both larger language models (LLM) and more efficient models deliver stronger returns on the underlying AI infrastructure. Additionally, enterprise AI use cases are also highly cost-effective, with token costs being very low relative to the benefits generated.
Growing threat of market fragmentation
“We see an increasing likelihood of interference in AI policy by both the US and China, which could lead to further fragmentation of the global market,” the report said.
(Except for the headline, this story has not been edited by Asianet News editorial staff and is published from a syndicated feed.)