Tesla Plans Upgrades for Older Models Amid HW3 Limitations for FSD

Source: Date:

During Tesla's Q1 2026 earnings call, CEO Elon Musk announced plans to upgrade the company's self-driving chip. Although the name for the new chip is yet to be finalized, it is expected to be called AI4.1 or AI4 Plus. This new chip will double the RAM from 16GB to 32GB per chip, and given the system's dual-chip design, total system memory will increase to 64GB.

Tesla admits HW3 isn't enough for FSD, will set up micro factories to retrofit older models

Musk acknowledged that vehicles equipped with the HW3 onboard computer will not be capable of supporting unsupervised Full Self-Driving (FSD), despite previous assurances to customers that the HW3 system, found in cars sold between 2019 and 2023, could meet FSD's computational requirements.

The development of the next-generation AI5 chip is still underway and production is not expected in the near term. Musk stated that AI4 and AI4 Plus will suffice for FSD, with AI5 likely being utilized for Optimus robots and data centers instead.

Interestingly, an upgraded AI4.5 chip is currently being sold in 2026 Model Y vehicles produced at the Fremont plant. This iteration employs a three-chip design, creating some confusion with the upgrade hierarchy: AI4 > AI4.5 > AI4.1/AI4 Plus, the latter set to launch in 2027.

What about those who have already purchased an HW3 vehicle and opted for the full FSD package, priced up to $15,000? Tesla has proposed a costly but feasible solution: offering HW3 owners a discounted trade-in for a newer vehicle with AI4 hardware, or retrofitting their current cars with the new chip and upgraded cameras. However, to execute this plan, Tesla will need to invest in micro factories across the US to facilitate these upgrades, representing a significant financial undertaking.

Crucially, only customers who purchased the FSD package will qualify for the upgrade, meaning those who solely bought an HW3-equipped Tesla will not benefit from this program.

Looking ahead, there are concerns regarding whether the AI4 hardware may face similar limitations as HW3, particularly as neural networks continue to evolve and their computational demands increase.

Via

Scroll to Top