技术解读 8小时前 196 阅读 0 评论

Tesla May Rely More on Nvidia as Dojo Supercomputer Team Said to be Shut Down

作者头像
AI中国

AI技术专栏作家 | 发布了 246 篇文章

TMTPOST -- Tesla Inc. may have to rely more on partners including Nvidia Corporation to launch chips for self-driving technology as its Dojo supercomputer team was said to be shut down.

Credit:Tesla

Tesla CEO Elon Musk has ordered to disband the Dojo team and its leader Peter Bannon is leaving , Bloomberg on Thursday reported, quoting people familiar with the matter. Around 20 Dojo team members recently have left to join Density AI, a startup co- founded by Ganesh Venkataramanan, the former head of Dojo, and other members are being reassigned to other data center and compute projects within Tesla, according to the report.

Tesla didn’t comment on the report. If it is accurate, the decision to shut down the Dojo team is inevitably a reversal of Tesla’s effort to develop in-house chips for autonomous driving. It is also the latest sign of a strategic shift. Tesla reportedly plans to increase its dependence on external technology partners, including Nvidia and Advanced Micro Devices Inc. (AMD) for compute, and Samsung Electronics Co. for chip manufacturing

The Dojo supercomputer and Tesla’s in-house chip called the D1 are a custom -built system designed to process vast amounts of data and video from Tesla electric vehicles (EVs) to train machine-learning models behind the automaker’s Autopilot and Full Self-Driving programs and Optimus humaniod robot. The D1 chip, used in training artificial intelligence (AI) much like Nvidia’s accelerators, is separate from the AI5 and AI6 inference chips, which run models in cars and robots.

Tesla has redirected its focus to AI-driven self-driving technology and robotics, with Musk pursuing an integration strategy across his business empire. In March, xAI acquired the social media platform X for $33 billion to bolster its chatbot training capabilities, while Tesla integrated the Grok chatbot into its vehicles.

The Information in April reported that Tesla’s planned Dojo supercomputer site has been beset by issues and Musk became “furious” about the delays when he found a facility missing most of its roof and ground floor because of delivery issues, bad weather, and other construction issues. Following his visit, Musk fired Amir Mirshahi, the construction project’s director of infrastructure.

Musk last month dismissed the possibility of a merger between Tesla and xAI, but announced plans for a shareholder vote to consider Tesla’s investment in the AI startup. He at an earnings call two weeks ago indicated at a strategic pivot suggesting future iterations of the company’s in-house technology could converge with that of its partners.

Muck last week confirmed Tesla has inked a $16.5 billion deal with Samsung Electronics as a supplier for semiconductors. He also unveiled the arrangement of Tesla’s supply chain in a post on X. “Samsung’s giant new Texas fab will be dedicated to making Tesla’s next-generation AI6 chip. The strategic importance of this is hard to overstate. Samsung currently makes AI4.TSMC will make AI5, which just finished design, initially in Taiwan and then Arizona,” he wrote.

作者头像

AI前线

专注人工智能前沿技术报道,深入解析AI发展趋势与应用场景

246篇文章 1.2M阅读 56.3k粉丝

评论 (128)

用户头像

AI爱好者

2小时前

这个更新太令人期待了!视频分析功能将极大扩展AI的应用场景,特别是在教育和内容创作领域。

用户头像

开发者小明

昨天

有没有人测试过新的API响应速度?我们正在开发一个实时视频分析应用,非常关注性能表现。

作者头像

AI前线 作者

12小时前

我们测试的平均响应时间在300ms左右,比上一代快了很多,适合实时应用场景。

用户头像

科技观察家

3天前

GPT-4的视频处理能力已经接近专业级水平,这可能会对内容审核、视频编辑等行业产生颠覆性影响。期待看到更多创新应用!