On April 1, Lenovo Group's 2024/25 Global Pledge Conference was held in Beijing. Dr. Rui Yong, senior vice president and chief technology officer of Lenovo Group, said at the conference that it will continue to focus on “computing” and “artificial intelligence” to carry out cutting-edge technological innovation, because “computing” is Lenovo's dominant gene, and “artificial intelligence” represents the trend of the times.
In terms of “computing” technology strategy, Lenovo will focus on three major technologies, namely liquid cooling technology, heterogeneous computing, and integrated software and hardware design.
Dr. Rui Yong said that in the face of the strong computing power requirements of large AI models, liquid cooling is becoming more and more important. When it comes to liquid cooling technology, Lenovo has always been at the forefront of the industry. In the new fiscal year, Lenovo will continue to invest in R&D in this field and continue to lead the industry.
At the same time, Lenovo will also systematically create two new, differentiated key technologies with its own IP. One is heterogeneous computing, which is the unified management, optimization, and scheduling of chips of different types and manufacturers, so that after optimization and coordination between different chips, it is possible to provide the strongest and most efficient computing power.
The other is a collaborative design integrating software and hardware to improve the fault tolerance of the system and achieve rapid fault recovery. According to industry data, for a kilocalorie server cluster, there will be 10-15 system failures every month. Failure recovery is necessary for every failure. Therefore, rapid fault recovery technology is very important. Dr. Rui Yong said that once these three things are done, Lenovo will establish technical differentiation of its own IP in terms of “computing” infrastructure.
At the same time, future artificial intelligence will be a hybrid artificial intelligence framework that includes a large public model and a large private model. Lenovo is a proponent of hybrid artificial intelligence, and Lenovo's new “Edge Cloud Network Intelligence” IT architecture will well support hybrid AI.
Artificial intelligence transformation includes not only existing technology, but also more technical elements that are still under development and not even clearly defined. This is a racetrack for innovators. Rui Yong believes that future artificial intelligence will be a hybrid artificial intelligence framework including big public models, big private models, and big personal models. It can integrate the advantages of various large models, accurately meet diverse needs, and significantly broaden the breadth and depth of AI applications. And Lenovo Group's new “End-Edge-Cloud-Network-Smart” IT architecture will strongly support this framework.
In the past two years, AIGC has set off an AI boom and ushered in a new era of artificial intelligence development. The big model shows amazing power, and its limitations also require attention, such as “a book of serious nonsense,” synthetic videos that don't match the causal relationship, etc.
How to solve these problems with big models? Rui Yong said that the next step in the development of big models is intelligence.
Rui Yong further explained that the abilities and functions of an intelligent body include active perception, intention understanding, decomposition of complex tasks, and long-term and short-term memory mechanisms. Agents are based on big models and surpass big models.
Rui Yong believes that if web pages and applications are the main carriers in the Internet and mobile Internet era, respectively, smart devices will be the core carriers in our intelligent era.
“The next step in the big model is intelligence, and it is also Lenovo's next key direction,” Rui Yong said. Lenovo Group will continue to use the BLP (Build, Leverage, Partner) framework to create self-developed technology and innovative products, accumulate its own intellectual property rights, and make breakthroughs one by one.
Developing cutting-edge technology is inseparable from the support of talents. Over the past three years, Lenovo Group's R&D team has doubled in size, with a total of nearly 19,000 people. One out of every four people is engaged in R&D or innovation. Rui Yong said that Lenovo Group will cultivate and reserve core technical talents through the Technical Ladder technology sequence project.
Looking back in time, “computing”, one of the two major anchor technologies, has brought glory and glory to Lenovo over the past 40 years. Looking forward to the future, the two anchor technologies “computing” and “artificial intelligence” will work together to bring Lenovo's next 40 more brilliant years.