As interest rates fluctuate, bond yields rise, and Chinese AI models intensify competition, market attention is shifting from technological development and infrastructure expansion to investment returns and long-term debt-servicing capacity. The AI industry continues to attract capital but must also prove that its business models can generate profits.
According to Bloomberg, DBS CEO Tan Su Shan said the market is reassessing the risks of AI investment and demanding higher returns and better financing structures. Companies with weaker financial positions could face credit-rating downgrades, but this remains a healthy development that can strengthen investment discipline.
DBS handled about SGD25 billion in AI-related transactions in 2026, mainly financing hardware in areas including semiconductors, memory and data centers. However, the bank held only a small portion of those transactions.
Chris Wood, global head of equity strategy at Jefferies Group LLC, said the AI investment boom is increasingly dependent on debt, exposing data-center developers to greater risks.
On revenue prospects, Wood warned that open-source models have lower costs, meaning more capital could flow to China. The market is growing concerned that if large language model (LLM) providers fail to develop viable business models, hyperscalers expanding data centers to meet computing demand from OpenAI and Anthropic will face further pressure.
Data from the OpenRouter platform showed that models from Chinese Big Tech companies processed 76.12 trillion tokens in the week ending September 27, far exceeding the 14.2 trillion tokens processed by US models. Chinese models had processed just 4.37 trillion tokens at the end of April, marking rapid growth in only a few months.
Article edited by Jerry Chen
