Tesla has halved the memory on its AI5 chip and trimmed the AI6 by a third to unlock volume for the Optimus robot, with Elon Musk insisting the move will have a 'negligible effect' on performance.
The decision, confirmed by the Tesla chief executive on X late on Wednesday, comes as the company races to stand up supply chains capable of delivering thousands of humanoid units rather than prototypes.
Elon Musk Defends Tesla Memory Cuts for Optimus Production
The news came after Micron Technology warned that humanoid robots could eventually demand more than 200 gigabytes of memory and multiple terabytes of storage per unit, similar to autonomous vehicles.
Speaking on Micron's fiscal fourth‑quarter earnings call, chief executive Sanjay Mehrotra said physical AI could become a significant driver of memory and storage demand by the end of the decade, with several customers already sampling next‑generation products.
An X user seized on those figures, calculating that 10 billion robots at 200GB each would require 2 trillion gigabytes of RAM, far above the roughly 45 billion gigabytes of DRAM the world produces annually. It is that sort of arithmetic that helps explain why Tesla is willing to trade capacity for volume now.
We cut our RAM in half for the Tesla AI5 chip (now 72GB of LP5) and 1/3 for AI6 (now 144GB of LP6).
— Elon Musk (@elonmusk) October 1, 2026
This was the only way to get enough volume for Optimus production and greatly reduces cost.
As it turns out, we think this will have a negligible effect on Optimus performance,…
Musk said Tesla cut the AI5 chip's memory in half to 72GB of LPDDR5 and the AI6 chip's memory by a third to 144GB of LPDDR6, calling it 'the only way to get enough volume for Optimus production' while 'greatly' reducing cost.
He added that the cuts should have a 'negligible effect on Optimus performance', as memory bandwidth is a bigger limiting factor than total memory capacity, and bandwidth was held constant.
For those tracking the silicon, the AI5 and AI6 chips are intended to power both Tesla's full-self-driving compute stack and the Optimus humanoid robot.
Earlier in the year, Musk noted that both chips dedicate roughly half of their AI computation accelerators to SRAM, meaning effective memory bandwidth is an order of magnitude greater than DRAM bandwidth for calculations held in SRAM cache. That context matters when weighing whether 72GB or 144GB is actually a bottleneck.
Tesla Optimus Production Ramps as Elon Musk Targets 1 Million Units
Tesla has reportedly placed its first large‑scale component order for roughly 5,000 Optimus units and is auditing Chinese suppliers ahead of production.
Supply chain reports tied to recent audits in Ningbo put Tesla's near‑term goal at about 1,000 units a week by late September, rising to 2,000 to 2,500 units a week by year‑end, with a longer‑term aim of building 1 million Optimus units a year at scale.
The audits covered firms including Tuopu Group, Ningbo Joyson Electronic and Zhejiang Sanhua Intelligent Controls, which already supply parts to Tesla's electric vehicles. According to reports, the visits moved the three manufacturers from provisional status to certified mass production partners, with fresh orders attached.
Musk has charted out a future in which Optimus robots are commercially available and capable of household chores, caring for children, and caring for the elderly. Whether a robot with 72GB or 144GB of RAM can handle all that without leaning heavily on cloud assistance is the sort of question engineers will answer in the lab, not on X.
On Thursday, Tesla shares closed 0.2 per cent lower at $354.11 and gained 0.22 per cent to $355.60 in extended trading. Benzinga Edge rankings indicate Tesla's stock has a Momentum score in the 13th percentile and a Growth score in the 41st percentile.
Nothing is confirmed yet so everything should be taken with a grain of salt. IBTimes UK cannot independently verify these claims, so take everything lightly.
