Elon Musk has long emphasized that Tesla is more than just an automaker—he envisions it as an AI-driven company on a mission to achieve full autonomy. A key part of that vision has been Dojo, a custom-built supercomputer designed to train Tesla’s Full Self-Driving (FSD) neural networks.
For years, Musk has promoted Dojo as a game-changer, capable of pushing Tesla past the threshold of “almost self-driving” to true autonomy. But as 2025 begins, another name has entered the conversation: Cortex, Tesla’s newest AI training supercomputer housed in its Texas Gigafactory. With this shift, Dojo’s role in Tesla’s AI strategy appears to be evolving, leaving many to wonder what’s next for the project.
Here’s a look back at Dojo’s journey, from its early mentions to where things stand today.
2019 – The First Mentions of Dojo
- April 22 – At Tesla’s Autonomy Day, Musk teases Dojo as a supercomputer for training AI, emphasizing its role in advancing FSD.
2020 – Musk Begins the Dojo Roadshow
- February 2 – Musk promotes Dojo’s potential, describing it as a powerful AI training system capable of processing massive amounts of video data.
- August 14 – Musk calls Dojo “a beast,” estimating its launch around August 2021.
- December 31 – Musk claims Dojo isn’t necessary for autonomy but will greatly enhance self-driving capabilities.
2021 – Tesla Officially Introduces Dojo
- August 19 – At Tesla’s first AI Day, the company formally announces Dojo and introduces the D1 chip, designed to work alongside Nvidia’s GPUs.
- October 12 – Tesla releases a Dojo Technology whitepaper, outlining a new floating-point arithmetic standard for deep learning.
2022 – Dojo Progress and First Installations
- September 30 – At Tesla’s second AI Day, the company unveils the first Dojo cabinet, with plans for a full Exapod cluster by Q1 2023.
>>>ZNS-09B1 Replacement Battery for Huabao Tong
2023 – Dojo Becomes a ‘Long-Shot Bet’
- April 19 – Musk tells investors that Dojo could dramatically cut training costs and potentially become a sellable service, similar to AWS.
- July 19 – Tesla announces it has started production of Dojo, with plans to invest over $1 billion into the project.
- September 6 – Musk confirms Tesla is data-limited, and both Nvidia and Dojo will be critical in scaling AI training.
2024 – Dojo Scaling Plans and the Rise of Cortex
- January 24 – Musk confirms Dojo is operational and will scale with future iterations (Dojo 1.5, 2, 3, etc.).
- January 26 – Tesla announces a $500 million investment to build a Dojo supercomputer in Buffalo but downplays its significance compared to Nvidia systems.
- May 20 – Musk reveals that Tesla’s Giga Texas extension will house a super-dense, water-cooled AI cluster.
- August 26 – Musk shifts focus to Cortex, calling it the “giant new AI training supercluster” at Tesla HQ in Austin.
2025 – Dojo Disappears from Tesla’s Narrative
- January 29 – In Tesla’s Q4 2024 earnings call, Dojo is not mentioned. Instead, Tesla highlights Cortex, stating it was instrumental in launching FSD V13.
>>>U3256125PHV-UTL Replacement Battery for Blackview Tab 11 SE
What’s Next for Dojo?
The lack of recent updates suggests that Tesla’s AI training focus has shifted from Dojo to Cortex, with the latter now playing a central role in Tesla’s AI infrastructure. While Dojo was once positioned as a groundbreaking, in-house alternative to traditional AI training solutions, Tesla’s increasing reliance on Nvidia hardware—combined with the rapid buildout of Cortex—raises questions about Dojo’s long-term significance.
It’s possible that Dojo will continue to operate alongside Cortex, supplementing Tesla’s AI training pipeline, especially as the company works toward next-generation Full Self-Driving (FSD) models. However, the lack of emphasis on Dojo in Tesla’s latest earnings reports and investor calls suggests that it may no longer be the game-changer Musk once envisioned.
Whether Dojo remains a critical AI asset, is gradually phased out, or evolves into a secondary component of Tesla’s broader AI strategy remains uncertain. But for now, Tesla’s AI future seems increasingly anchored to Nvidia-powered compute clusters, with Cortex taking center stage.