The narrative of “AI at all costs” seems to have lost some steam recently. Once heralded as an unstoppable force for growth, artificial intelligence now faces critical questions about sustainability, particularly as funding models and market sentiment shift.
Advertisement
One indicator worth watching closely in comparison with the dotcom era is the source of funding for hyperscaler capital expenditure. We are now seeing evidence that companies have started to rely on debt rather than free cash flows and earnings to finance their spending needs. This shift has contributed to a change in sentiment around an overheated AI boom and triggered a sell-off in the US equity market.
Meanwhile, closer to home, Asian stocks have also begun to consolidate after an extended rally, with the Hang Seng Tech Index retreating from a multi-year high in October. Nonetheless, China’s approach to AI, backed by strong policy support and a deliberate effort to balance competition with innovation, could propel Asia’s AI-driven rally further.
China has a different vision for AI from that of the United States. While US firms focus on artificial general intelligence and high-performance models, China is emphasising efficiency, adoption and ecosystem impact. The key question now is how China can balance intensifying competition with a more sustainable innovation trajectory.
Unlike the capital-heavy strategies of Western tech giants, Chinese firms are demonstrating that advanced AI can be built without top-tier hardware. Models such as DeepSeek R1 and other open-source entrants reflect a shift towards cost efficiency and accessibility. This has spurred a wave of open-source development among domestic players, prioritising openness, replicability and shared progress over pure proprietary control.
Advertisement
This approach is shaped by necessity. Limited access to cutting-edge AI chips has prompted China to pursue a more commercially viable path focused on application-level success. Thus, a vertical strategy involving smaller models trained on proprietary data may be best suited for resource-constrained enterprises in China.