Tesla’s automated driving roadmap is entering a new and more radical phase, with Elon Musk signaling a sweeping rework of how the company’s cars perceive the road, learn from drivers, and eventually operate without human oversight. The latest hints point to a larger, more capable Full Self-Driving model, a fresh pricing structure, and even changes to how drivers are monitored, all of which raise the stakes for regulators, investors, and owners.
Instead of incremental tweaks, Musk is now talking about a step change in capability that could redefine what “Full Self-Driving” means inside a Tesla. The promised overhaul blends new software, new hardware, and a new business model, and it is already testing how far the company can push safety expectations while still convincing drivers to trust the system.
Musk’s next FSD leap: bigger models and “insane” capabilities
Elon Musk has been previewing a new generation of Full Self-Driving that he describes in increasingly dramatic terms, including “insane capabilities” tied to a major software revision. In one recent outline of his plans, he framed the coming Version 14.3 as a fundamental upgrade to the neural networks that power Tesla’s driver assistance, promising a more fluid, more humanlike and more user friendly driving environment that goes well beyond the current behavior of the cars. That same vision casts the update as a turning point for how the system handles complex urban scenarios, with Musk positioning it as a qualitative jump rather than a routine over the air patch, a claim that aligns with his broader pattern of using software milestones to reset expectations for what the vehicles can do.
Alongside the software, Musk is also preparing a structural change in the underlying artificial intelligence, with Tesla set to launch a larger after the holidays that is explicitly described as bigger and more capable than the current system. In the same plan, the company ties this larger model to its long running robotaxi ambitions and even sets a three week window to remove robotaxi safety monitors, a move that would mark a dramatic shift in how much autonomy Tesla believes its software can safely handle. Taken together, the promise of Version 14.3 and the larger model suggests Musk is not just iterating on Full Self-Driving, he is trying to reframe it as a near complete driving stack that can eventually support commercial autonomous services.
From “Supervised” driving to robotaxis without monitors
The overhaul is also about branding and expectations, which is where the concept of “Full Self-Driving Supervised” comes in. Musk has been promoting this label as a way to acknowledge that the system still requires human oversight while still signaling that it is a step closer to full autonomy. Investors and owners have been told that the supervised system will benefit from everything Tesla has learned from its driver assistance deployments, with shareholders now hoping that this new supervised branding will coincide with faster, more visible improvements in how the cars behave. At the same time, Musk has urged people to manage their expectations, a tacit admission that even a supervised label does not eliminate the gap between marketing and real world performance.
The most striking part of the roadmap is the plan to remove robotaxi safety monitors in three weeks once the larger model is in place, effectively signaling that Tesla believes its software can handle commercial style driving without a dedicated human overseer. That target is embedded in the same description that lays out how Tesla will tease a crazy and explain when it is coming, tying the removal of monitors directly to the rollout of the upgraded neural networks. I see this as the clearest sign yet that Musk wants to pivot from a consumer driver assistance product to a platform that can support fleets of autonomous vehicles, even as regulators and safety advocates continue to scrutinize whether the technology is ready for that responsibility.
Hardware 5, AI5 and the deep tech behind the overhaul
Underneath the software hype, Tesla is also reshaping the hardware that will run these larger models. Musk has already announced that the company is working on Hardware 5, also called AI5, and he has described it as the next generation of the computing platform that powers Tesla Autopilot and Full Self-Driving. In his description, Musk announced Hardware 5 during a Tesla annual meeting and positioned AI5 as a more efficient and more powerful successor to the existing HW3 and HW4 chips, which already run complex neural networks inside cars like the Model 3 and Model Y. The explicit naming of AI5 underscores how central artificial intelligence has become to Tesla’s identity, with the hardware roadmap now marketed as aggressively as the vehicles themselves.
For drivers, the practical impact of AI5 will be whether it can run the larger FSD model with enough headroom to handle real time perception, planning and control without overheating or throttling performance. Musk’s description of AI5 includes specific power consumption figures for HW3 and HW4, and he has argued that the new platform will deliver more compute per watt, which is critical for an electric vehicle that must balance range with processing demands. By tying the “insane capabilities” of Version 14.3 to both a larger neural network and a dedicated AI5 platform, Tesla is effectively telling owners that the future of Full Self-Driving will depend as much on the silicon under the dashboard as on the code that arrives over the air.
Pricing shake up and the push to mass adoption
The overhaul is not limited to technology, it also extends to how Tesla charges for Full Self-Driving. Elon Musk has confirmed that Tesla is moving Full Self-Driving to a monthly structure, shifting away from a pure one time purchase model that locked in a high upfront fee. In a recent explanation, he described how Tesla is moving its Full Self-Driving software to a subscription option for buyers of its vehicles, giving them the choice between a recurring payment and a lump sum. I see this as a recognition that many owners are reluctant to pay thousands of dollars for a feature that is still labeled “Supervised,” and that a subscription can both lower the barrier to entry and create a steadier revenue stream for the company.
At the same time, Musk has been layering in new capabilities that are designed to make the package feel more indispensable. One of the most controversial is a major Full Self-Driving update that explicitly addresses texting and driving, with Tesla CEO Elon Musk announcing a change that integrates texting and driving capabilities into the system. In that description, Tesla CEO Elon Driving Update and presents it as a way to manage the reality that drivers often interact with their phones despite stringent rules against texting and driving. By promising that the car can shoulder more of the driving load while the human handles communication, Musk is betting that convenience will drive adoption, even as safety experts warn that any feature that normalizes distraction could backfire if the system fails at the wrong moment.
Safety, supervision and the gap between promise and reality
All of these changes land on top of a system that is still officially framed as driver assistance, not full autonomy, which is why the word “Supervised” is so important. Tesla has already showcased a Full Self-Driving Supervised demonstration that runs with zero interventions, highlighting how the car can navigate complex routes without the driver touching the controls. In one shared account, a user described how Tesla Full Self Driving FSD Supervised handled a drive without manual input, reinforcing Musk’s argument that the technology is rapidly approaching human level performance. Yet even in these demos, the driver is expected to remain ready to take over, a reminder that the system’s apparent competence can mask edge cases it still cannot handle.
More from Morning Overview