AI5 chip

Developer

Tesla

Predecessor

AI4 chip (also known as HW4)

Architecture

Modular, specialized, inference-optimized core with tileable 2x2 compute clusters

Optimized For

Models with fewer than 250 billion parameters

Primary Function

Inference chip

Power Consumption

Up to 800 watts (when processing complex environments)

Performance (TOPS)

2000-2500 TOPS (target for early 2026)

Key Characteristics

Lowest silicon cost and highest performance-to-power ratio (for its class)

Strategic Shift From

Dojo chip design project

Manufacturing Partner (AI5)

Taiwan Semiconductor (TSMC)

Projected Performance Increase

40-fold over AI4 chip

Manufacturing Partner (AI5/AI6)

Samsung Electronics

Elon Musk's Past Performance Claim

Up to 5000 TOPS (10 times more capable than HW4)

2024-06-13

2025-09-07

2025-01-01

2025-12-01

2026-01-01

2026-01-01

The AI5 and AI6 chips that Tesla is vigorously developing are expected to support the training of its artificial intelligence and autonomous driving. It is reported that the AI5 is a transitional or main chip for specific scenarios, produced by Taiwan Semiconductor; the AI6 will serve as the 'unified heart' of Tesla's future AI ecosystem and will be manufactured by Samsung Electronics. [...] ① Elon Musk stated that the Tesla AI5 chip design team has completed an outstanding design review, and the AI5 may be the best inference chip for models with fewer than 250 billion parameters; ② Previously, Tesla halted its chip design project Dojo; ③ The AI5 and AI6 chips, which Tesla is vigorously developing, are expected to support the training of its artificial intelligence and autonomous driving. [...] On September 7 local time, Tesla CEO Elon Musk posted on social media that he just completed an outstanding design review with the Tesla AI5 chip design team, stating that this chip will become an 'epic' product, and the subsequent AI6 chip is expected to be the most outstanding AI chip to date. Musk added that he believes the AI5 may be the best inference chip for models with fewer than 250 billion parameters, as 'it has the lowest silicon cost and the highest performance-to-power ratio; AI6

When I first dove into the design documents for Tesla’s AI5 inference engine, I felt a familiar thrill—the same excitement that drove me to architect power electronics for my electric vehicle startup. AI5 represents a pivotal shift away from the monolithic, high-throughput Dojo chip mindset toward a specialized, inference-optimized core that balances latency, power efficiency, and integration density. In this section, I’ll unpack the key architectural innovations I believe underpin AI5’s [...] Tesla’s decision to streamline its AI chip design around inference-centric architectures marks a significant strategic inflection. By reallocating resources from the ambitious Dojo training platform to the pragmatic development of AI5 and AI6 chips, Tesla aims to sharpen its competitive edge in both autonomous driving and robotics. While the shift raises valid concerns about training sovereignty and long-term IP defensibility, the immediate benefits in cost, time-to-market, and engineering [...] Dojo’s monolithic 7nm design—spanning multiple reticle boundaries—was a bold proof of concept. AI5, conversely, reflects a modular ethos. Each compute cluster is tileable in a 2×2 array, allowing Tesla to scale the chip’s size and TDP (thermal design power) based on vehicle variant or datacenter rack configuration.

Korea ML.co site reports that Tesla AI5 chip will have 2000~2500TOPS performance as early as 2026. (Tesla calls it HW4 and AI4 interchangeably from HW4). To this end, Tesla is working with Samsung Electronics and TSMC as contract manufacturers for AI5. Mass production is likely to begin next year. Tesla vehicles are expected to start being equipped with HW5 (AI5) in the second half of 2025, likely around December. ADVERTISEMENT Powered By 10 [...] OMG. Tesla MODEL Q Revealed. Evidence, Length, Weight, Price Share Next Stay Image 5Image 6 During a past earnings call, Elon claimed AI5 could be as much as 10 times more capable than HW4, which would imply 5,000 TOPS. There could be two AI5 chips for each Tesla car and one for each Teslabot. A combination of chips in packages would align the previous statement with the new specifications. The previous claim from 2024 could be adjusted based upon actual mass production specifications. [...] Artificial intelligence Superconductor Humanoid Robot Science About Contact About Privacy Policy Terms and Conditions Keep on reading Undo Undo Undo Undo Undo Undo Undo Undo Undo Undo Undo Undo Undo Undo

Musk has said next-generation AI5 chips would be produced at the end of 2026 and last month announced a $16.5 billion deal to source AI6 chips from Samsung Electronics (005930.KS), opens new tab, without providing a production timeline. He has said future AI inference chips, including AI6, would be deployed in self-driving vehicles and its Optimus humanoid robots, though he has noted the substantial computing power could enable broader AI applications. [...] "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 said, without directly mentioning Dojo. Morgan Stanley analysts led by Adam Jonas valued the Dojo supercomputer at $500 billion in 2023, saying it opened a new market for the automaker beyond cars sales, similar to how Amazon's cloud unit boosts profit for the ecommerce firm. Advertisement · Scroll to continue [...] Published Time: 2025-08-07T22:03:04.77Z Tesla to streamline its AI chip design work, Musk says | Reuters Skip to main content Exclusive news, data and analytics for financial market professionals Learn more about Refinitiv - Tesla will streamline its AI chip research to focus on its development of inference chips used to run AI models and make real-time decisions, CEO Elon Musk said, after a media report he had ordered the closure of the in-house Dojo supercomputer team.