Tesla stock has seen a meteoric rise recently, nearly doubling to $1.4T market cap in just three months. The catalyst is Trump’s victory and Elon Musk’s instrumental role in it. Investors expect Elon’s influence to relax regulations, clearing the way for a robotaxi rollout—a potential game-changer for the automotive industry and transportation infrastructure as a whole.
Tesla is trading like an AI startup, and for good reason: its “Full Self-Driving” (FSD) system is at the heart of the valuation. But investors are making several assumptions that warrant a closer look.
# The Bull Case: Tesla’s data volume
Tesla has a massive data advantage. With over 4 million vehicles on the road, each equipped with a suite of cameras, Tesla’s fleet constantly collects real-world driving data. This data acts as a “shadow trainer” for its AI, gathering insights in every imaginable driving condition. Meanwhile, FSD subscribers are essentially paying to supervise the AI, providing Tesla with even more labeled data.
Behind the scenes, Tesla employs an army of data labelers to prioritize edge cases—rare and tricky driving scenarios. Tesla combines supervised, unsupervised, and reinforcement learning to continuously refine its AI. More importantly, Tesla’s massive data funnel gives it a clear edge in data volume, which usually means better machine learning.On top of this, Tesla’s vision-only approach—using cameras without lidar—makes its system cheaper and theoretically more scalable. The result? A system that could be deployed anywhere in the world, not just pre-mapped areas.
# The Bear Case: Quantity vs. Quality
More data is not always better, and it’s possible that Tesla might reach a plateau. Teslas are mostly driven in suburban and highway environments, where edge cases are relatively rare, while Waymo trains their AI in dense, chaotic urban areas. The result is a more diverse dataset that’s constantly evolving.
Waymo is owned by Alphabet (Google’s parent company) and has been operating fully autonomous L4 robotaxis in large cities since 2018. It uses both cameras and lidar. Combining 2D and 3D data in this way is something Tesla initially set out, but failed, to do. Elon has said that the lidar data was hard to make sense of, since it often conflicted with the vision data. But Waymo has somehow managed to make it work. This means that its data is much richer in quality. The urban-focused fleet also encounters far more edge cases than Tesla’s suburban-heavy dataset, giving its AI a potentially richer training environment.
The geofencing that Tesla bulls dismiss as a crutch is actually a strategic advantage for Waymo. By limiting its operations to pre-mapped areas, Waymo minimizes liability and achieves Level 4 autonomy with real-world deployments today. Tesla, by comparison, is still at Level 2, which means drivers must supervise and be ready to intervene. Moving from L2 to L4 isn’t just an incremental step—it’s an order-of-magnitude leap in complexity.
Tesla bulls overlook the difference between urban vs. suburban environments. Waymo trains its AI in the most chaotic environments—places like San Francisco and New York City. Tesla’s FSD, by contrast, collects much of its data in relatively predictable suburban and highway conditions. Cities provide more edge cases, such as jaywalking pedestrians, aggressive lane merges, and unexpected construction detours.
As the saying goes, “If you can make it here, you can make it anywhere.” If Waymo’s AI can survive the chaos of Manhattan or downtown LA, it’s far more likely to handle less complex environments like suburbs or rural highways. In contrast, Tesla’s suburban data might create a “same-shit-different-day” scenario, where the AI becomes great at average cases but struggles with rare, high-stakes scenarios.
# Regulatory relaxation is a double-edged sword
If Musk succeeds in relaxing regulations, Tesla might clear the path for robotaxi deployment. In theory, this could limit Tesla’s liability, especially if passengers are required to sign release forms. However, accidents involving unsuspecting third parties (e.g., pedestrians) remain a significant risk. Under current frameworks, manufacturers are liable for autonomous vehicles, meaning Tesla could face infinite exposure, even for a small number of accidents. Even if Musk changes the regulatory framework, courts ultimately determine liability, and his involvement could be seen as a conflict of interest at best, complicity at worst.
Tesla’s data suggests FSD already outperforms human drivers in safety metrics—which is very impressive. However, publishing accident rates also acknowledges that FSD causes accidents. They currently get away with it because, again, L2 puts the onus on the driver. But it's a different story if they deploy robotaxis as L3 or L4. In that case, any accident caused by FSD is the onus of Tesla, exposing them to unlimited liability.
# The scalability myth
Tesla bulls often criticize Waymo’s reliance on lidar as costly and unscalable, but this argument doesn’t hold weight for two reasons. First, lidar is becoming much cheaper, and with volume production, costs could approach Tesla’s camera-only system. Ironically, Elon’s first principles philosophy should be applied here: just because something was a certain way before doesn’t mean it has to be that way. Secondly, the market for taxis in general is urban. When was the last time you saw one in a random suburb or the countryside? Waymo only needs a few major cities to succeed, giving it a more focused path to profitability.
# Tesla vs. Waymo is really Tesla vs. Google
Waymo uses millions of AI-generated simulations to train its system. It’s AI teaching AI, similar to how AlphaGo and AlphaZero were trained. Tesla may also use simulations, but Waymo’s superior compute power (via Google’s custom TPUs) means it’s clearly dominant in this regard.
Tesla bulls often cite Elon’s dismissal of lidar and geofencing as evidence of Tesla’s superior approach. But they overlook Google’s hegemony in AI. They were in the game before anyone. They delivered AlphaGo, AlphaFold, and many other paradigm-shifting, world-changing products, backed by custom hardware and massive compute resources. Tesla’s Dojo may be promising, but it’s still a newbie.
# tl;dr
Tesla’s $1.4T valuation rests on its perceived lead in autonomous driving, which is to say its lead in AI. If Tesla dominates AI, the stock might still be undervalued. But the data quantity vs. data quality, liability risks, and Waymo’s technical advantages suggest Tesla may be actually falling behind. In which case, it's vastly overvalued.