Media365
The rapid emergence of low-cost artificial intelligence (AI) models is expected to accelerate, rather than reduce, global spending on AI infrastructure, according to a report by global brokerage Jefferies.
The brokerage said the launch of Chinese AI company Z.ai’s latest GLM-5.2 model represents another major milestone in the AI race, intensifying competition with leading Western AI developers while significantly lowering inference costs.
Jefferies described the development as “another DeepSeek moment”, highlighting the growing challenge posed by Chinese AI firms. It noted that GLM-5.2 delivers performance comparable to leading enterprise AI models at a fraction of the operating cost, making AI deployment more affordable for businesses.
Lower inference costs are expected to encourage wider enterprise adoption of AI and increase demand for computing infrastructure, including servers, data centres and memory chips. The report also noted that falling token costs are prompting more companies to run AI models on their own infrastructure instead of relying solely on cloud-based services.
Despite concerns that cheaper AI models could dampen investment, Jefferies said there is currently “zero sign of AI capex slowing.” Hyperscalers and AI developers continue to invest aggressively in data centres and computing infrastructure to support rising AI workloads.
The brokerage, however, cautioned that the biggest long-term challenge remains whether companies can generate adequate returns on the massive investments being made in AI infrastructure. It added that these concerns remain largely theoretical for now, given the strong pace of ongoing investment.
Reflecting its bullish outlook on AI hardware, Jefferies has added South Korean memory chip maker SK Hynix and Japanese flash memory company Kioxia to its model portfolios. It has also increased its allocation to Samsung Electronics, while trimming exposure to internet companies such as Alphabet and Alibaba.
The report further highlighted Taiwan as a key beneficiary of the global AI investment cycle. It said the island’s economy, exports and semiconductor capital expenditure continue to receive a boost from expanding AI infrastructure demand, led by companies such as TSMC.
Jefferies believes that as AI models become more affordable and widely adopted, the demand for high-performance computing infrastructure and advanced memory chips is likely to remain strong, reinforcing the long-term growth outlook for semiconductor companies.
