Terafab 來真的了！ 馬斯克剛說 本週開動 現在已經在 瘋狂開招聘！ 他直接甩數據說 考慮良率後 Terafab每月 實際產能接近 16萬片晶圓 比很多人 猜的高60%！ 這是要一次解決 Optimus Robotaxi Dojo的晶片荒 更誇張的在後面 馬斯克說 靠AI加速設計 運氣好一點 2026年12月 就能完成 AI6 tape-out！ Tesla這是要把 晶片命脈抓在 自己手裡 這波AI 硬體狂飆 你跟得上嗎？. ...

Core Viewpoint - ByteDance is developing AI chips, aiming to receive chip samples by the end of March and plans to produce at least 100,000 AI inference chips this year, with a future production target of 350,000 chips [1] Group 1: Company Developments - ByteDance has not yet responded to inquiries regarding its self-developed chips [1] - The company has been actively investing in AI, with its core business heavily reliant on AI technologies, and has launched AI products such as Doubao [1] - Since 2022, ByteDance has initiated the self-development of cloud training and inference chips to build a complete AI technology stack [1] Group 2: Industry Context - The global tech competition is intensifying, with AI chips becoming a focal point for major tech companies, including Google's TPU, Amazon's Inferentia, and Tesla's Dojo, all aimed at reducing reliance on external suppliers [1] - The increasing demand for computing power in the industry drives ByteDance's decision to either self-develop AI chips or continue external procurement, with the goal of maintaining control over its development trajectory [1]

Investment Rating - The report maintains a neutral investment rating for the automotive and parts industry [5] Core Insights - Tesla is actively promoting the FSD subscription upgrade, which is expected to drive domestic automakers towards a software payment model. The launch of the Robotaxi service without a safety driver in Austin is anticipated to enhance FSD penetration rates, making software payments a significant revenue contributor for Tesla [8][12] - The restart of Tesla's Dojo project is expected to create new opportunities for domestic liquid cooling companies, supported by a $16.5 billion agreement with Samsung for AI6 chip production [13] - Continuous attention is recommended for companies in the autonomous driving and robotics supply chains, as partnerships and collaborations in these sectors are expected to boost demand for key components [14][15] Summary by Sections Investment Suggestions and Targets - The report suggests focusing on the liquid cooling supply chain, humanoid robotics chain, T chain, and autonomous driving companies, as they are expected to benefit from ongoing catalysts [15] - Key liquid cooling stocks include Invec (未评级), Silver Wheel (买入), Top Group (买入), Feilong (未评级), and Chuanhuan Technology (未评级) [16] - Key robotics stocks include New Spring (买入), Top Group (买入), Silver Wheel (买入), Daimai (买入), and Sanhua Intelligent Control (买入) [16] - Key autonomous driving stocks include Jingwei Hengrun-W (买入), Bertley (买入), and Desay SV (买入) [16] Sales Tracking - The report indicates that the wholesale sales of passenger cars in China for the week of January 12-18 were 359,000 units, a year-on-year decrease of 28% [17] - The retail sales for the same period were 351,000 units, down 22% year-on-year [17] Market Trends - The automotive sector showed overall strength, with the automotive index rising by 2.2%, outperforming the Shanghai and Shenzhen 300 index [28] - The commercial vehicle sector saw a significant increase of 7.38%, while the passenger vehicle sector experienced a slight decline of 0.67% [28]

Product Development - Dojo 2.0 将采用模块化设计，如同乐高积木，可以分解和构建 [1] - Introspect 功能即将推出 [1] - Torii, Sozo, 和 Katana 模块化程度进一步提高 [1]

Core Insights - The Dojo project, initially envisioned as a cornerstone for Tesla's AI ambitions, has been officially shut down after six years of development and hype, with the team disbanded by August 2025 [1][4][19] - Tesla's focus has shifted towards a new AI supercomputer project named Cortex, which aims to address real-world AI challenges and has already begun deployment [2][19] - The decision to terminate Dojo has sparked mixed reactions, with some viewing it as a necessary response to declining electric vehicle sales and others seeing it as a strategic pivot towards partnerships for chip development [3][4] Dojo Project Overview - Dojo was designed as a custom supercomputer to train Tesla's Full Self-Driving (FSD) neural networks, with the goal of transforming Tesla into an AI company rather than just an automotive manufacturer [1][5] - The project was expected to enhance Tesla's capabilities in autonomous driving, humanoid robots, and semiconductor independence [1][5] Transition to Cortex - Tesla has begun promoting Cortex, a new AI supercomputer that will support FSD and other AI training needs, while Dojo has not been mentioned in recent updates [2][19] - Cortex is being developed at Tesla's Austin headquarters and is intended to have vast storage for FSD and Optimus video training [2] Reactions to Dojo's Closure - The closure of Dojo has been interpreted by some as a response to Tesla's struggles in electric vehicle sales and the slow rollout of autonomous taxi services [3] - Others argue that the dissolution of Dojo does not signify failure but rather a strategic shift towards leveraging external partnerships for chip development [3] Key Personnel Changes - Following the announcement of Dojo's termination, approximately 20 employees left Tesla to establish a new AI chip and infrastructure company, DensityAI, including Dojo project lead Peter Bannon [4] Financial Implications - Tesla signed a $16.5 billion agreement with Samsung for the development of the next-generation AI6 chips, which are crucial for FSD and other AI applications [4] - Analysts suggest that the shift from Dojo to Cortex could impact Tesla's financial outlook, with potential new revenue streams from AI services [15] Future Directions - Tesla's long-term vision includes the possibility of competing directly with Nvidia in the AI space, although challenges remain in adapting existing AI software to work with Tesla's custom chips [15] - The company continues to invest in AI capabilities, with plans for a new supercomputer in Buffalo, albeit under a different name than Dojo [19]

Core Insights - Nvidia has invested $2 billion in Elon Musk's AI company xAI, which has reached a valuation of $20 billion in its latest funding round [2][5][6] Group 1: Investment Rationale - Nvidia aims to become the foundational "cornerstone" and "engine" of the AI world, requiring top-tier clients and partners to validate and promote its technology [2][3] - By investing in xAI, Nvidia secures a stable and significant demand for its GPUs, as xAI is developing large-scale AI models that require substantial computational resources [2][3] - A successful AI model from xAI could serve as a benchmark, showcasing Nvidia's hardware and software capabilities, thereby attracting more AI companies to its ecosystem [3][4] Group 2: Competitive Landscape - The investment helps Nvidia build a strong "Nvidia Alliance" to counter competition from AMD, Intel, and major cloud providers who are developing their own chips [3][4] - Access to valuable data from Musk's companies, such as Tesla and X platform, can enhance xAI's model training and optimize Nvidia's AI computing platforms [3][4] Group 3: Strategic Implications - Musk's AI projects are among the most computationally intensive, providing Nvidia with invaluable feedback to design more powerful chips for future AI workloads [4] - The investment signifies a bet on an AI-driven future across various sectors, positioning Nvidia as a key player in this evolving landscape [4][5] - A synergistic ecosystem could emerge if Tesla's autonomous driving, Optimus robot, X platform algorithms, and SpaceX's Starlink all utilize Nvidia's hardware and xAI's software [4][5]

Core Viewpoint - Tesla has officially terminated its AI supercomputer project "Dojo," which was initially seen as a pivotal step in its transformation from an electric vehicle manufacturer to an AI company. This decision reflects a significant shift in strategy and raises questions about Tesla's future direction in AI and autonomous driving [1][6]. Group 1: Project Termination - The Dojo project, aimed at training Tesla's autonomous driving neural networks with self-designed chips, has been disbanded as of August 2025, despite previous ambitions for commercialization by 2026 [1][3]. - The project was intended to create a system independent of Nvidia GPUs, promising faster computation and lower latency, but ultimately failed to deliver on its goals [3][5]. Group 2: Challenges Faced - Tesla struggled to link the outcomes of its autonomous driving efforts directly to Dojo, and the performance of its chips could not keep pace with Nvidia's advancements [6]. - The mainstream AI software ecosystem is primarily optimized for GPUs, which hindered Dojo's development and contributed to its eventual failure [6]. Group 3: Implications of Termination - The dissolution of Dojo highlights the high risks associated with pursuing complete technological independence through self-developed chips and infrastructure, revealing Tesla's limitations in resources and ecosystem [6][8]. - The loss of key personnel from the Dojo team, who have since founded a startup named "DensityAI," indicates a significant talent drain that can jeopardize future projects [8]. - The decision to end Dojo signals a strategic pivot for Tesla from self-reliance to leveraging partnerships, as evidenced by its new collaboration with Samsung for the development of the next-generation AI6 chip [8]. Group 4: Future Outlook - Despite the termination of Dojo, Tesla's ambitions in AI remain intact, with ongoing collaborations with Nvidia, AMD, and Samsung to expand its new supercomputer "Cortex," which is responsible for training the latest version of its Full Self-Driving (FSD) technology [8][9]. - The failure of Dojo serves as a case study of Tesla's bold attempts in the high-risk fields of autonomous driving and AI, raising the question of whether this setback is merely a conclusion or a necessary sacrifice for larger-scale transformation [9].

Core Viewpoint - Tesla aims to transition from being solely an automaker to an AI company, focusing on achieving full self-driving capabilities through advanced computing power and data processing [1][2]. Development of Dojo - Dojo was introduced as a custom-built supercomputer designed to train Tesla's Full Self-Driving (FSD) neural networks, which at the time required human oversight despite some automated capabilities [2][3]. - The timeline of Dojo's development includes its first mention in 2019, with Musk highlighting its potential to process vast amounts of video data for training AI [4][5][8]. - By 2021, Tesla officially announced Dojo, introducing its D1 chip and outlining plans for a supercomputer capable of significant AI training [9][10]. Progress and Challenges - Throughout 2022 and 2023, Tesla reported progress on Dojo, including the installation of its first cabinet and plans for a full Exapod cluster by early 2023 [10][12]. - Musk indicated that Dojo could significantly reduce training costs and potentially become a sellable service, similar to Amazon Web Services [11][12]. - However, by mid-2023, Tesla faced challenges with Nvidia hardware supply, prompting a renewed focus on Dojo to ensure adequate training capabilities [16]. Transition to Cortex - In 2024, Tesla began transitioning from Dojo to a new supercomputer called Cortex, which utilizes Nvidia GPUs and aims to enhance AI training for FSD [18][19]. - The Cortex supercomputer was reported to consist of approximately 50,000 H100 Nvidia GPUs, facilitating improvements in FSD performance [19][20]. - By early 2025, the Dojo project was officially shut down, with Tesla consolidating its resources towards the development of the AI6 chip, which is intended to serve multiple AI applications [22][23]. Future Directions - Tesla's future plans include scaling AI capabilities with the AI6 chip, which is designed for both inference and training, indicating a strategic shift in its AI development approach [22][23]. - The company aims to maintain a competitive edge in AI by focusing on integrated chip designs rather than dividing resources across different projects [23].

Project Overview - Tesla's Dojo is a custom-built supercomputer project initiated by Elon Musk [1] Future Development - The report discusses the original vision for Dojo, its current status, and future plans [1]

Core Insights - Tesla has decided to shut down its Dojo AI supercomputer project and disband the associated team, marking a significant shift in its AI strategy [2][10][44] - The decision comes after years of hype and promises from CEO Elon Musk regarding Dojo's potential to revolutionize Tesla's self-driving capabilities and AI initiatives [2][12][13] - The company is now pivoting towards partnerships for chip development, particularly focusing on its new AI6 chips from Samsung, which are intended to support various AI applications [11][31] Group 1: Dojo's Development and Shutdown - Dojo was designed as a custom-built supercomputer to train Tesla's Full Self-Driving (FSD) neural networks, aiming to achieve full autonomy and support the robotaxi initiative [3][4][18] - Despite initial ambitions, Tesla failed to effectively link its self-driving advancements to Dojo, leading to a lack of focus on the project in recent communications [5][8] - The shutdown of Dojo was announced shortly after Tesla signed a $16.5 billion deal for next-generation AI6 chips, indicating a strategic shift away from self-reliant hardware [11][12] Group 2: Implications for Tesla's AI Strategy - The closure of Dojo has sparked mixed reactions, with some viewing it as a failure of Musk's promises, while others see it as a necessary pivot towards a more sustainable AI strategy [8][9] - Analysts have noted that losing key talent from the Dojo team could hinder future AI projects, especially given the specialized nature of the technology [10] - Tesla's future AI efforts will now rely more on partnerships with established chip manufacturers like Nvidia and AMD, moving away from its previous goal of self-sufficiency in chip production [31][32] Group 3: Financial and Market Impact - The initial projections for Dojo included significant financial commitments, such as a $500 million investment for a supercomputer at the Buffalo gigafactory, which will now not be allocated to Dojo [39][44] - Analysts had previously estimated that Dojo could potentially add $500 billion to Tesla's market value by creating new revenue streams through AI and robotaxi services [35] - The shift in strategy may impact investor sentiment, as the ambitious goals set for Dojo were not met, leading to questions about Tesla's long-term AI vision [38][40]