AUSTIN, Texas — Tesla has announced plans to construct a network of urban micro-factories specifically designed to upgrade the self-driving computers and cameras in approximately 4 million vehicles currently running Hardware 3 — a direct response to CEO Elon Musk's Q1 2026 earnings call admission that HW3 lacks the capability to achieve unsupervised Full Self-Driving.
The announcement marks a major milestone in Tesla's commitment to delivering on its long-standing FSD hardware promise, and Musk framed it as both a technical solution and a business opportunity.
Why HW3 Cannot Do Unsupervised FSD
During the Q1 2026 earnings call on April 22, Musk was direct about the technical limitation: "Hardware 3 simply does not have the capability to achieve unsupervised FSD. It has only one-eighth the memory bandwidth of Hardware 4, and memory bandwidth is the chokepoint for AI inference."
This distinction — memory bandwidth, not just processing power — is at the heart of why a software update alone cannot bridge the gap. Unsupervised FSD requires large neural networks to run in real time, and HW3's bandwidth ceiling makes that impossible no matter how the software is optimized.
The Micro-Factory Solution
Rather than relying on existing service centers, which Musk acknowledged would be too slow and inefficient for the scale of upgrades needed, Tesla intends to establish small-footprint production facilities in major metropolitan areas.
The upgrade is more extensive than swapping a chip. It requires replacing the onboard AI computer and installing higher-resolution cameras throughout the vehicle, along with the associated wiring changes. Tesla's micro-factory model parallels the efficiency logic behind its Gigafactories: purpose-built production lines outperform general service bays when volume and consistency are the priorities.
Musk stated on the call: "We basically need many production lines to make the change. And I do think over time it's going to make sense for us to convert all HW3 cars to HW4, because that's what enables them to enter the Robotaxi fleet and have unsupervised FSD."
What Owners Can Expect
Tesla is offering two paths for HW3 owners who purchased the Full Self-Driving software package. The first is a discounted trade-in toward a new vehicle already equipped with AI4 hardware. The second is the hardware upgrade itself once the micro-factories come online.
As a bridge solution while the micro-factory program is being established, Tesla confirmed it will release FSD v14 Lite for HW3 vehicles in late June 2026. This stripped-down version will bring HW3 cars access to current-generation FSD features — including improvements from the MLIR rewrite in v14.3 — without requiring hardware changes. It is not a substitute for the full upgrade but gives owners a meaningful software improvement in the near term.
Context: HW2 Set the Precedent
This is not the first time Tesla has navigated a hardware gap of its own making. When the company concluded that Hardware 2 could not deliver full autonomy, it offered free HW3 upgrades to FSD purchasers. That precedent established both the expectation and the financial model that Tesla is now revisiting at a larger scale.
The key difference this time is the proposed delivery mechanism. Rather than handling upgrades piecemeal through service centers, the micro-factory model suggests Tesla is thinking about this as a fleet-level infrastructure challenge — one that integrates naturally with the company's Robotaxi expansion plans.
The Bigger Picture: Robotaxi Fleet Growth
Musk's comment about enabling HW3 vehicles to join the Robotaxi fleet is significant. Tesla's Cybercab production began at Giga Texas earlier this year, and the company has been expanding its unsupervised robotaxi service to new cities including Houston, Dallas, and seven more markets by mid-2026. Converting the HW3 fleet into Robotaxi-capable vehicles would dramatically expand the addressable vehicle pool for that service — potentially millions of cars in private hands becoming revenue-generating assets.
The micro-factory program is still in the planning stage, with specific locations and timelines yet to be announced. But Tesla's decision to address the HW3 gap head-on — rather than continuing to defer it — reflects growing confidence that the path to full autonomy is clear enough to justify the investment.
