Why a Tesla breakthrough in fully autonomous driving would be a short-lived victory for TSLA stock

The bullish case for TSLA stock, among many others, is solving fully autonomous driving. The dream is cars that operate on autopilot, like a taxi but without a driver. If it were cheap enough, it would even eliminate the need for many people to own a car.

This has certainly been Elon Musk’s dream for a decade. Unfortunately, he made so many promises that it became something of a meme.

“We will introduce our specially designed robotaxi or Cybercab in August,” Elon Musk said during the latest earnings conference call. I suspect it will just be a concept car with comfortable seats and no steering wheel.

But what if he actually kept his promises?

If so, I’m sure TSLA stock will skyrocket after the unveiling event, scheduled for August 8th.

I also think it would be a fleeting victory. TSLA’s main argument was that mountains of data collected from real cars would help them build an autonomous driving system via a sort of brute force method. I believe we have reached the limits of this approach.

Instead, Tesla turned to AI techniques.

The clues are in the latest earnings call, where Musk revealed this:

We also continue to expand our AI training capacity in the first quarter, more than doubling our training compute sequentially… We have installed and fielded, meaning they are actually working, 35,000 computers or H100 GPU. GPU is not a good word, they need a new word. I always feel like an asshole when I say GPU because I’m not. GPU stands for – G stands for graphics. About 35,000 H100S are active, and we anticipate there will probably be about 85,000 by the end of this year in training, just for training. We make sure to be as efficient as possible in our training. It’s not just about the number of H100s, but also how efficiently they are used.

This is a clear sign that Tesla is using a new technique to solve autonomous driving. I hope it works (although I doubt the lengthy regulatory approval process).

The problem is that if Tesla solves FSD with AI, they won’t be the last to do it. Everyone will be right behind. This technique is one that Tesla has just moved towards, so other automakers would only have to acquire 85,000 H100 chips and use the same techniques, which are hardly a secret.

There is no chance of being six months ahead in automotive technology. Yes, there would be an explosion in FSD sales at $99/month and prices might even increase again, but will you remain a loyal customer when Hyundai offers software equal to $19.99/month? If so, it will not maintain a more sustainable long-term advantage than anti-lock brakes or all-wheel drive.

Yes, they have more data, but with reinforcement learning, the size of the dataset is not that important, it’s more about quality. And other automakers also collect data. In an industry where the refresh cycle is +5 years, being first isn’t that valuable, and certainly not enough to justify a 10x higher multiple on TSLA stock.

So while I certainly wouldn’t fear any groundbreaking announcements, I doubt the enthusiasm will last a year. Moreover, as Morgan Stanley writes today, “China has already won the battle for global electric vehicle supremacy.”

However, there is one area where I think Tesla could create a lasting advantage that could send the stock significantly higher. I will write more about this in a future article.