俄乌冲突以后，面对美西方全面制裁，俄罗斯仍然靠出口石油天然气，硬生生撑了3年多与整个西方打消耗战。然而随着特朗普威胁要对购买俄能源的国家 进行二级制裁。普京终于清醒了，他表示："如果我们继续只靠卖油气活着，那俄罗斯终将沦为一个末流国家。" 苏联时期，靠军火和油气撑起国家对外的贸易结构。到了俄罗斯，没怎么改动，继续这条路。只不过苏联时代还能靠军事扩张带来政治红利，而现在，军事 投入带来的则是经济压力和国际孤立。 过去，资源是优势，能换外汇，能养军队。可现在全球能源正在转型，欧美国家正在快速降低对化石能源的依赖，新能源、可再生能源大行其道。对俄罗斯 来说，欧洲对传统石化能源需求开始降低。 7月20日，俄罗斯国家电视台播出了一期《俄罗斯、克里姆林宫、普京》的节目。这档节目其实是今年1月录制的，但内容分量十足。节目里，主持人扎鲁宾 和普京展开了一场直面现实的对话。 普京罕见地敞开心扉，说："我们不能永远靠卖资源维持国家运转，这种经济结构在技术上没有前途。"这话一出口，无异于给国内一部分还沉浸在"油气金 矿梦"里的民众泼了一盆冷水。 事实上，俄乌冲突以来，西方已经对俄罗斯施加了超过三万项制裁。从能源、金融到科技、交通， ...

Core Viewpoint - The automotive industry is entering a new era of chip self-research and high computing power, driven by the need for advanced autonomous driving capabilities, particularly L3 level automation, as exemplified by companies like Xiaopeng and NIO [6][39][60]. Group 1: Chip Development and Competition - The competition for automotive computing power began around 2021, initiated by NVIDIA's Orin-X chip, which boasts a computing power of 254 TOPS, significantly surpassing Mobileye's Q5H and Tesla's HW3.0 [1][6]. - Companies like NIO have adopted multiple Orin-X chips, achieving over 1000 TOPS in computing power [3]. - The automotive computing power has fluctuated between 80 to 1000 TOPS over the past four years, but a new phase has emerged in 2023 with the introduction of self-developed chips [5][34]. Group 2: Self-Developed Chips and Industry Trends - Xiaopeng's self-developed 5nm chip, NX9031, is expected to reach 2000 TOPS with two chips in the ET9 model, while the Xiaopeng G7 features three self-developed Turing AI chips, achieving 2200 TOPS [6][39]. - The trend of automakers developing their own chips is gaining momentum, similar to Tesla's earlier journey, as companies seek to overcome the limitations of the "black box" model previously used with suppliers like Mobileye [9][30]. - The emergence of domestic chip companies like Horizon and Black Sesame Intelligence is diversifying the market, with many automakers now developing their own chips that can compete with NVIDIA's flagship products [35][38]. Group 3: The Shift to L3 Autonomous Driving - The automotive industry is approaching the L3 autonomous driving era, with Xiaopeng defining its G7 as the "world's first L3 level AI car" [39]. - The effective computing power required for L3 autonomous driving has been clarified by Xiaopeng at 2200 TOPS, indicating a significant leap from L2 systems [43][55]. - The transition to L3 involves not only technological advancements but also a shift in liability, as vehicles may be held accountable for accidents, increasing pressure on automakers to refine their technologies [56][58]. Group 4: Challenges in Chip Development - The journey of self-developing chips is fraught with challenges, including architectural issues and the risk of costly failures during the chip manufacturing process [62][64]. - Companies must also ensure that their chips meet stringent automotive safety standards, which can extend the validation period significantly [69]. - The need for large-scale production to recoup development costs is critical, with estimates suggesting that around 1 million units may be necessary for profitability [71].