Elon Musk said Tesla has cut memory on its next AI chips so Optimus can be produced in higher volume at lower cost. In a 1 October post on X, he said Tesla halved RAM on the AI5 chip to 72GB and reduced AI6 memory to 144GB.

He said memory bandwidth is unchanged and that the expected effect on robot performance is negligible. The comments treat total RAM as a cost and supply constraint, and bandwidth as the specification Tesla is not willing to give up.

That is a hardware allocation decision, not a new commercial forecast. Tesla has previously discussed Optimus work around Fremont. This post does not add unit targets, timelines, or deployment figures beyond those earlier remarks.

What is confirmed should be kept distinct from what is inferred.

  • Musk stated that AI5 RAM was halved to 72GB and AI6 RAM was reduced to 144GB to secure Optimus production volume and cut cost.
  • He said bandwidth is unchanged and that any performance impact should be negligible.
  • The post does not disclose chip output, robot bill of materials, factory capacity, or a change to previously reported Fremont ramp commentary.

For Canadian investors, Tesla is a widely held U.S. mega-cap growth name, typically accessed on Nasdaq as TSLA or through U.S.-heavy ETFs in taxable accounts, TFSAs, and RRSPs. Optimus sits in the long-duration part of the Tesla story: a possible second product engine after vehicles and energy storage, still unproven at commercial scale.

Chip memory is a scarce input. Cutting RAM to raise volume is consistent with trying to industrialize a humanoid platform rather than maximize on-device storage. If bandwidth truly holds, Tesla is arguing that inference and control will not be the binding constraint. That is a management claim. Independent performance data were not provided.

The investment-intelligence point is narrower. Supply and unit cost are the variables that would have to move before Optimus could matter in Tesla’s financials. A RAM cut aimed at volume is a production-enabler signal. It is not evidence that shipments, revenue, or margins have changed. Reading the post as a near-term earnings event would go beyond the source.

Canadians who already hold TSLA also face a USD listing, CAD/USD translation, and concentration risk if Tesla is a large slice of a growth sleeve. Those are portfolio-construction facts, not a recommendation. This article is not tax, legal, or investment advice.

The practical follow-up is the same as the gap in the post: any later Tesla or Musk comments on chip supply, and any quantified Optimus production metrics. Until those appear, the 1 October remarks are best read as commentary on how Tesla is rationing memory to prioritize robots, not as a change to the known volume picture.