Over the past year and change, the world has watched on as Tesla forges towards autonomous vehicles, with the release of FSDbeta to customer cars. This beta is designed to capture massive amounts of data to train their vehicles to accommodate the incredibly complex challenge of navigating the world.
One of the biggest differentiators with Tesla’s approach is that it in theory is learning how to navigate drivable space, using vision-only and is able to navigate roads that it has never driven on before, thanks to the learnings of billions of kilometers (or miles) of driving.
Given this context, and statements made publicly in the past, the expectations of many outside the US, who have paid for the FSD package (currently A$10,100), was that international markets would follow fairly closely behind the US, with the obvious need to train for any differences in road rules, signage, road markings etc.
Today that theory was basically shattered.
In a tweet by Elon Musk today, he says that they are now focusing on one region (the US) until FSD is out of beta. Only then would they consider expanding geographically to have FSD beta in other countries.
Obviously being a US-based company, Tesla would naturally focus on their own market first, however customers like myself, who have paid equally for the FSD package, are right to feel a little concerned by this timeline development.
With 12 months of FSDbeta progress, combined with assessments of how capable the system is today, we can attempt to build some projections about when FSD may come out of beta in the US.
I will preface this by saying that humans think in linear terms and we’ve been told that development improvements can happen exponentially, however that hasn’t been true so far, with each release offering incremental improvements.
It seems likely that we’re really looking at least another 12 months at least before FSD could even consider dropping the beta tag. Then we consider that Autosteer, Navigate on Autopilot, Traffic Light and Stop Sign Control and Summon are all still labelled with the (beta) tag. If Tesla takes this more gmail approach of leaving the beta tag on products, we could be realistically talking about multiple years.
In the best case scenario, let’s imagine international rollout of beta starts in a serious way sometime in 2023, from what Musk posted today, it seems unlikely they’d be able to add multiple countries at once, needing to then start the beta process of collecting data, training, iterating etc, Obviously they don’t need to start development from scratch each time, with many of the detections, routing issues surely locked away by that point, but still, we’re likely to see beta last at least a year in each market.
This begs the question, what is the order of rollout for each country and personally, I’m very keen on understanding where Australia lands on the list. Are we talking 2025 and beyond?
While some conflate the completion of FSD with the launch of robotaxis, personally I see those two things as very separate. Given the regulatory nightmare that is awaiting to approach level 4/5 autonomous vehicles across different markets, I expect this to be another year or more after Tesla is technically ready.
If the planets align and regulators see what’s coming from watching the development in the US, hopefully we can be ready with required tests to prove driverless vehicles are safe to launch and not suffer from additional months or years of delay.
Overall the I understand why Musk is taking this approach, but that doesn’t change the fact I’m disappointed this timeline is continuing to extend (particularly in international markets like ours) when we were really sold on the idea that the cars learn, then they can drive anywhere.
It is clear now that Musk’s statement at Autonomy Day back in 2019, that they’d have a million robotaxi’s on the road that year, was a mistake. As was selling FSD at full price to consumers outside the US that were realistically many years away from receiving the stated functionality. While some items on the FSD package are available (in beta), the big one, driving city streets, is still some time away.
I really wish Tesla could find a way to multi-task and simultaneously develop FSD beta in multiple countries. While I personally don’t have plans to enter my car into Tesla’s robotaxi fleet, I can understand people being frustrated if they purchased a Model 3 in 2019, with an expectation of using it as a robotaxi, and now facing the prospect of that car being 5+ years old, before you get a chance to try that out.