Tencent executives explain AI strategy: More domestic chip supply will be available in the second half of the year, and C-end monetization is still in the early stages On May 14th, according to a report by Golden Finance, during a conference call with The Paper, Tencent executives discussed the topic of AI related expenditures. They stated that they have seen an increase in demand for AI related services from both internal products and external model users. Previously, they had indicated that capital expenditures would increase this year compared to last year, but now they are more certain and confident in this guidance. We expect a significant increase in capital expenditures, and more domestically produced AI chips will be delivered month by month in the second half of this year. ”Regarding the issue of returns on AI products, Tencent President Liu Chiping stated that model training is essentially an investment in the future and may not immediately generate returns. But over time, abilities will accumulate and help unlock many different business opportunities. Regarding the issue of C-end performance, Liu Chiping believes that in the Western market, the penetration rate of paid services is very high, and the consumption level is at a high level. The subscription price in Western markets is several times higher than that of equivalent services in China, whether it is music services or video services. The payment penetration rate in China is approximately in the single digits. When you apply it to the Chinese market, based on a subscription model, the scale will not be that large. When a service must be supported through payment, it is likely not a winner takes all business, as there will be multiple participants in the market, each with a certain market share and subscription volume. In addition, regarding e-commerce or advertising as a monetization method, it has been considered and is currently in its early stages. Even in the much higher ECPM US market, leading players have not yet launched very mature advertising models. ”(Dongxin Society)