Special Feature: Global Wealth Management Forum · 2025 Shanghai Suhewan Conference On October 19, financial frontline news reported that the Global Wealth Management Forum · 2025 Shanghai Suhewan Conference took place from October 18 to 19 in the Jing'an District of Shanghai. Liu Shufeng, co-founder of Hundsun Technologies Inc., attended and spoke at the conference.
Liu stated that we are currently amidst a wave of digital revolution, with large language models rapidly ushering people into the era of financial technology 3.0, beginning to replace cognitive labor. The internet has prepared the groundwork for this era with big data and online capabilities, while the industrial internet and the Internet of Things are pushing parts of the manufacturing industry toward digitization. Moreover, advancements in large language models are further accelerating the overall digitization process.
Liu pointed out three prominent trends in the current digitization of capital markets. Firstly, there is an accelerated integration; whether it's asset management platforms or those centered on custody services, there is a consolidation of trading, data, standardization, alternative investments, and other resource and technology tools into a one-stop service platform, with numerous mergers and acquisitions aimed at this goal. Secondly, there is a comprehensive push towards cloud technology, which lays the groundwork for further intelligent applications of large models. Thirdly, the deepening of comprehensive data utilization is evident; data value is increasingly significant and has permeated into the field of alternative investments, making data processing capability a key competitive edge among financial institutions.
Despite the rising interest in large models, Liu candidly noted that from an investment perspective, global investment applications of large language models in finance account for less than 5%, indicating that we are still in the early stages. Current technical models continue to evolve, with breakthroughs in multimodal, reasoning, and other areas, yet the application scenarios remain limited, primarily focusing on efficiency tools and improvements in customer service efficacy, while core areas like value creation, risk decision-making, and investment choices are still in experimental phases.
However, Liu emphasized that the development of large models presents equal opportunities for financial institutions of various sizes. In terms of foundational capabilities, large and small institutions are not significantly different; both rely on basic service providers. Recently, some domestic financial institutions have been at the forefront of launching native applications for large models, with the impact of these native applications on customer experience improvement and rollout expected to show initial results in the coming months.
He also cautioned that the widespread adoption of native large model applications faces numerous challenges, including compliance requirements, customer acceptance, and issues inherent to large models, such as hallucinations and inaccurate data computations. According to his assessment, it will take another 3-5 years for native applications of large models to reach maturity, and once mature, they will possess the capability to transform the industry's landscape.
Regarding the currently trending topic of computing power competition, Liu believes that the next developmental direction is that large model computing power platforms are likely to become the next-generation operating systems, with various industry applications migrating to these platforms. At that point, large models will serve as the technological foundation for industry chain development, and this transformation could reshape market dynamics, which is the fundamental reason behind the current market's high enthusiasm for computing-related fields.
Comments