Elon Musk says Tesla will prioritise inference chips for real-time AI decisions
Tesla will streamline its AI chip programme to focus on developing inference chips – specialized processors designed to run AI models and make real-time decisions – CEO Elon Musk has confirmed.
His statement followed a Bloomberg report saying that the electric carmaker had disbanded its in-house Dojo supercomputer team, with team leader Peter Bannon leaving the company.
Dojo’s billion-dollar promise fades
The Dojo project had been a high-profile effort to build custom training chips capable of processing huge amounts of video and sensor data from Tesla’s vehicles to improve its self-driving software.
In a post on the X platform, Musk argued it no longer made sense for Tesla to “divide its resources” across two very different chip designs.
“The Tesla AI5, AI6 and subsequent chips will be excellent for inference and at least pretty good for training. All effort is focused on that,” he wrote, without explicitly naming Dojo.
Tesla has not responded to a request for comment. When first unveiled, Dojo was billed as a breakthrough in AI computing, with some analysts seeing it as a business opportunity beyond Tesla’s core automotive operations.
In 2023, Morgan Stanley analysts led by Adam Jonas valued Dojo at $500 billion, suggesting it could be as transformative for Tesla as Amazon Web Services was for Amazon.
They described the supercomputer as “the key accelerant at the intersection of hardware and software,” opening the door to lucrative AI services. Jonas did not immediately respond to queries on whether the shift away from Dojo would affect Tesla’s long-term valuation.
The Bloomberg report suggested the decision to wind down Dojo follows both strategic and staffing challenges. Around 20 engineers recently left to join DensityAI, a start-up formed by former Tesla employees.
Remaining staff from the Dojo group are reportedly being reassigned to other Tesla data centres and compute projects.
Tesla’s AI strategy consolidates
Musk’s announcement reflects a broader industry trend in which tech firms are streamlining chip design to cut latency, reduce power consumption and keep costs in check. Instead of maintaining multiple architectures for training and inference, many companies are consolidating around chips that can handle both tasks well enough – particularly if those chips can be deployed at scale.
Tesla’s next-generation AI5 chips are expected to be ready by the end of 2026, according to Musk. The company has also signed a $16.5 billion deal to source AI6 chips from Samsung Electronics, although no production timeline has been disclosed.
Musk has said the AI6 will power Tesla’s autonomous driving systems as well as its Optimus humanoid robots – and could have broader applications in other AI-powered services.
The shift in AI chip strategy comes during a turbulent period for Tesla. Over the past year, the firm’s share price has slid as electric vehicle sales slowed amid growing competition from rivals in China and the US. In Europe, consumer backlash against Musk’s outspoken political commentary has also weighed on the brand.
Tesla has undergone a significant restructuring, including several high-profile executive departures and thousands of job cuts. While the company continues to invest in vehicle production, its emphasis is increasingly on AI-driven software, self-driving capabilities and robotics – with Musk pushing for tighter integration between Tesla and his other ventures, including X and xAI.
By funnelling resources into inference chip development, Tesla is betting on faster, more energy-efficient AI processing at the edge – in vehicles, robots and other devices – rather than relying solely on vast data centre infrastructure.
This approach may allow Tesla to accelerate the deployment of its autonomous systems, but it also raises questions about the fate of the massive investments made in Dojo. If the in-house supercomputer was meant to be a cornerstone of Tesla’s AI future, the decision to wind it down signals a major shift in priorities.
For now, Musk remains bullish on the performance potential of the upcoming chips. Whether that optimism translates into the kind of cross-industry impact some analysts once envisioned for Dojo will depend on how quickly Tesla can bring its new hardware to market – and how well it can compete in a rapidly evolving AI landscape.
Get seen where it counts. Advertise in Cryptopolitan Research and reach crypto’s sharpest investors and builders.
Read More
Elon Musk says Tesla will prioritise inference chips for real-time AI decisions
Tesla will streamline its AI chip programme to focus on developing inference chips – specialized processors designed to run AI models and make real-time decisions – CEO Elon Musk has confirmed.
His statement followed a Bloomberg report saying that the electric carmaker had disbanded its in-house Dojo supercomputer team, with team leader Peter Bannon leaving the company.
Dojo’s billion-dollar promise fades
The Dojo project had been a high-profile effort to build custom training chips capable of processing huge amounts of video and sensor data from Tesla’s vehicles to improve its self-driving software.
In a post on the X platform, Musk argued it no longer made sense for Tesla to “divide its resources” across two very different chip designs.
“The Tesla AI5, AI6 and subsequent chips will be excellent for inference and at least pretty good for training. All effort is focused on that,” he wrote, without explicitly naming Dojo.
Tesla has not responded to a request for comment. When first unveiled, Dojo was billed as a breakthrough in AI computing, with some analysts seeing it as a business opportunity beyond Tesla’s core automotive operations.
In 2023, Morgan Stanley analysts led by Adam Jonas valued Dojo at $500 billion, suggesting it could be as transformative for Tesla as Amazon Web Services was for Amazon.
They described the supercomputer as “the key accelerant at the intersection of hardware and software,” opening the door to lucrative AI services. Jonas did not immediately respond to queries on whether the shift away from Dojo would affect Tesla’s long-term valuation.
The Bloomberg report suggested the decision to wind down Dojo follows both strategic and staffing challenges. Around 20 engineers recently left to join DensityAI, a start-up formed by former Tesla employees.
Remaining staff from the Dojo group are reportedly being reassigned to other Tesla data centres and compute projects.
Tesla’s AI strategy consolidates
Musk’s announcement reflects a broader industry trend in which tech firms are streamlining chip design to cut latency, reduce power consumption and keep costs in check. Instead of maintaining multiple architectures for training and inference, many companies are consolidating around chips that can handle both tasks well enough – particularly if those chips can be deployed at scale.
Tesla’s next-generation AI5 chips are expected to be ready by the end of 2026, according to Musk. The company has also signed a $16.5 billion deal to source AI6 chips from Samsung Electronics, although no production timeline has been disclosed.
Musk has said the AI6 will power Tesla’s autonomous driving systems as well as its Optimus humanoid robots – and could have broader applications in other AI-powered services.
The shift in AI chip strategy comes during a turbulent period for Tesla. Over the past year, the firm’s share price has slid as electric vehicle sales slowed amid growing competition from rivals in China and the US. In Europe, consumer backlash against Musk’s outspoken political commentary has also weighed on the brand.
Tesla has undergone a significant restructuring, including several high-profile executive departures and thousands of job cuts. While the company continues to invest in vehicle production, its emphasis is increasingly on AI-driven software, self-driving capabilities and robotics – with Musk pushing for tighter integration between Tesla and his other ventures, including X and xAI.
By funnelling resources into inference chip development, Tesla is betting on faster, more energy-efficient AI processing at the edge – in vehicles, robots and other devices – rather than relying solely on vast data centre infrastructure.
This approach may allow Tesla to accelerate the deployment of its autonomous systems, but it also raises questions about the fate of the massive investments made in Dojo. If the in-house supercomputer was meant to be a cornerstone of Tesla’s AI future, the decision to wind it down signals a major shift in priorities.
For now, Musk remains bullish on the performance potential of the upcoming chips. Whether that optimism translates into the kind of cross-industry impact some analysts once envisioned for Dojo will depend on how quickly Tesla can bring its new hardware to market – and how well it can compete in a rapidly evolving AI landscape.
Get seen where it counts. Advertise in Cryptopolitan Research and reach crypto’s sharpest investors and builders.
Read More