zhang yongwei of china electric vehicle 100: in the era of artificial intelligence, the automotive industry lacks computing infrastructure the most
2024-10-02
한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina
"the new technological route led by electrification has changed the development pattern of the global automobile industry, and chinese automobile companies have begun to rise rapidly. before electrification was completed, the wave of intelligence was coming again. with big computing power, big data, artificial intelligence, represented by large models, has begun to be integrated with automobiles, and artificial intelligence has become a new decisive factor in driving automobile revolution." zhang yongwei, vice chairman and secretary-general of the china electric vehicles association of 100, said at the global intelligent automobile industry conference on september 29. .
the industry generally believes that china's new energy vehicle industry has moved from the first half of electrification to the second half of intelligence. smart cockpits and smart driving have replaced engines and gearboxes and become the core configuration of new energy vehicles. at the same time, in addition to cockpits and intelligent driving, the impact of artificial intelligence on automobiles is gradually increasing, entering many fields and links such as automobile "research, production, supply, and sales".
"traditional intelligence has ushered in new automobile intelligence driven by artificial intelligence. the changes in the past have not yet been completed, and new changes are about to begin. this superimposed development has become the new normal for the development of the automobile industry." zhang yongwei said that the technological high ground of the automobile industry and the strategic competitive fulcrum of automobile companies are rapidly migrating to ai-driven intelligence. the competitiveness of the past can no longer support the next development of automobile companies.
in the era of ai competition, companies are competing not only on cognition and speed, but also on strength, because the threshold for development is getting higher and higher. the computing power level owned by tesla is 100 eflops, which is the sum of the computing power possessed by all car companies. if in this round of changes, artificial intelligence is not paid enough attention or cannot keep up, and cannot form core competitiveness in computing power, chips, and algorithms, such companies will lose their future.
in addition, data-driven cars have also become one of the consensuses in the industry, and data has become the core asset and element of the enterprise. at present, automobile companies have insufficient data mining capabilities and poor utilization of data value.
zhang yongwei said: "it is not enough to rely on the amount of data of a single car company to train software and systems. in the ai era, all competitiveness must be trained by data, and the problem of large-scale data must be solved. this requires the creation of our mechanism. promote data aggregation, enable everyone to input data into the platform in accordance with market-oriented principles, and use data in accordance with market-oriented principles to solve the current small-scale obstacle to data.”
in terms of several key elements of artificial intelligence such as data scale, computing power level and investment intensity, there is a big gap between chinese car companies and tesla. zhang yongwei suggested that companies can form a collaborative mechanism, such as jointly building a data platform, shared computing power, etc. according to public data, by the end of 2024, the total planned computing power resources of the three major operators will only be 53 eflops.
"in the era of artificial intelligence, what automobile companies lack is not production capacity, and the number of oems is not the main contradiction in industrial development. what the automobile industry lacks most is computing power infrastructure. to complete end-to-end intelligent driving research and development and training, the starting point of computing power is the power is 1 eflops. huge investment must be made in computing power, and continuous investment must be made to form a large-scale team around data, computing power, and algorithms. there is no computing power cluster with thousands or tens of thousands of cards, and there is no computing power cluster with thousands of cards. with tens of thousands of algorithm teams, it is difficult for companies to become competitive on the new track. "zhang yongwei said that in addition to building local computing power, we must also speed up solving the problems of immature domestic computing power, supporting tool chains, and weak ecological capabilities. , reduce the risk of computing power being "stuck" in hardware.
(this article comes from china business news)
