The 2026 Sustainable Global Leaders Conference returns to the banks of Shanghai's Huangpu River from October 20-22. Nobel laureates and Turing Award winners will take the stage to discuss the future and trends of technology, and the internet-famous professors who "rescue" people from anxiety and involution — Macfarlane and Xiang Biao — will hold a live dialogue!
Limited seats available, first come first served. Let us journey together toward sustainability! [Click to register] Sina Tech News, October 19 evening — Recently, in an interview with CCTV Finance's "Dialogue" program, Xiwang Sunrise Co-CEO Wang Zhan stated that in the next decade, we will formally enter an era of human-machine symbiosis. The next battle in AI is not about who is smarter, but about who can deliver intelligence more stably and cheaply. We need to continuously reduce the unit Token cost so that more enterprises can afford it, use it reliably, and run it fast. Xiwang aims to be the "profit and loss optimizer" of the AI industry.
"In the future, each person may have 2 to 3 Agents, and the total number of Agents is expected to exceed the human population," Wang Zhan said. Token fees will be listed on everyone's and every company's bills just like water and electricity fees, and everyone needs to start accounting for Tokens, he added.
Wang Zhan introduced that in 2024, Xiwang discovered that the vast majority of business scenarios in which customers used the QiWang S2 were concentrated on the inference side. This led the company to further judge that the scale of inference computing power will soon surpass training. At the end of 2024, Xiwang wrote "All-in inference" into its corporate strategy, committed to reducing the unit Token cost. "Among domestic GPU companies, we are the first to eat the crab."
"Why are super nodes so important today? Large models have already reached the trillion-parameter level, and they will become even larger in the future. Users need powerful system-level computing power. Only by driving down the unit Token cost will a large number of intelligent applications truly take off across thousands of industries and achieve inclusive intelligence," Wang Zhan said.
Comments