5D·

"We are years behind": Tesla AI boss admits to lagging behind in autonomous driving

Shortly before the launch of the new robotaxi service, a surprisingly honest statement provides a deep insight into the company's true capabilities.

By Hannah Klaiber

01.06.2025, 16:50

In a rare moment of candor, Ashok Elluswamy, Head of AI and Autopilot Software at Austin, Texas-based car manufacturer Tesla, has admitted to being significantly behind the competition in terms of technology. Speaking to a journalist, Elluswamy admitted that Tesla's technology for fully autonomous driving (FSD) lags behind that of Waymo, a subsidiary of Google parent Alphabet from Mountain View in the US state of California.

The head of AI said: "Technically, Waymo is already capable. We are perhaps a few years behind." This statement provides a remarkable categorization, as Waymo is already considered the "gold standard" of autonomous driving and operates commercial robotaxi services in several US cities. First reported by InsideEVs.

Vision vs Lidar: Which solution is better?

At the heart of the gap lies a fundamentally different technological philosophy. Tesla is known to rely on "Tesla Vision", an approach that relies solely on data from cameras, which, according to Elluswamy, only cost between one and ten US dollars each.

Waymo, on the other hand, equips its fleet with lidar and radar sensors in addition to cameras. These were long regarded as extremely expensive, but according to industry observers, the cost of lidar units has now fallen to less than 1,000 dollars, which makes their use in cheaper vehicles possible in principle.

Elluswamy defends Tesla's path as the "more valuable to the world", as a low-cost solution allows for faster scaling. However, he admits that it is "technically challenging" to solve the problem of autonomous driving with cameras alone. Corresponding tests are repeatedly posted on social media, for example on ex-Twitter X:

Business Insider magazine observed in a test just how challenging Tesla's alternative approach can be: According to it, a Tesla with FSD software activated had driven into a bicycle lane and run a red light, while a Waymo mastered a comparable route flawlessly.

A reality check for Musk's robotaxi plans

Elluswamy's admission puts another damper on Tesla boss Elon Musk's ambitious plans. Musk has been promising an imminent autonomy revolution for years and wants to launch his first robotaxi service in Austin as early as next month.

The realistic assessment of his own AI boss now raises the question of how efficient this service will really be in the beginning. In any case, it fits seamlessly into a series of postponed schedules. For example, a fully autonomous journey from coast to coast in the USA was already promised for 2016. So far, that has not happened.

https://t3n.de/news/wir-liegen-jahre-zurueck-tesla-ki-robotaxi-1689325/

t3n.de„Wir liegen Jahre zurück“: Tesla-KI-Chef gesteht Rückstand bei autonomem Fahren ein - t3n – digital pioneers

13 Comentários

PikaPika0105@PikaPika0105

I hope the man has saved enough for his retirement....😰

•

Marcel@Kytez

I drove both in California and can only confirm this. You feel safer in a Waymo than with a human driver. These things are amazing.

You can't relax as much in a Tesla FSD. The vehicles don't "see" enough, don't react as intelligently and even scrape kerbs or other edges in between.

In my opinion, the feeling of safety will be the biggest factor for long-term success. And Tesla still has a lot of work ahead of it. (You want to use the time in the car to do something other than remain "ready to intervene" and watch the traffic 😅)

•

PaulPK@PaulPK

These "first mover" gold standards are rarely sustainable. They are simply overrun by technological progress. Apple is perhaps an exception, although there are also better smartphones.

••

PikaPika0105@PikaPika0105

@PaulPK Apple is not the best technologically either, but the brand is very strong. Tesla used to be similar, but they are now technologically outdated and the brand has been destroyed by their CEO.

•

PaulPK@PaulPK

@PikaPika0105 Technologically outdated "only" in the e-vehicle sector. The pipeline is bigger. As far as FSD is concerned, they will soon be much cheaper and better - and roll out more broadly than Waymo. Brand is "artificially" damaged - in the end, "bad publicity is better than no publicity."

••

PikaPika0105@PikaPika0105

@PaulPK I don't agree with you. They are outdated in the e-vehicle sector, which is their only source of income and therefore essential for everything else. What is your assessment of FSD based on? It's just another legend and the whole post by @Tenbagger2024 was about them being behind there too. As for "bad publicity is better than no publicity", I'd say that's total nonsense. It doesn't work the same way as with clicks on social media. Your brand image is everything to you as a company, especially in such a brand-focused sector. Bad publicity is bad publicity, because bad publicity means bad business. Tesla will not be the end of Musk, but it looks more and more like Musk will be the end of Tesla.

•

PaulPK@PaulPK

@PikaPika0105 They are "only" lagging behind. That's all it says. The fact that Tesla's FSD (Full Self-Driving) could be better than Waymo in the long term is not due to today's performance, but to a fundamentally different technological approach and scaling advantage. Here are the most important reasons:

🔍 1. vision-first instead of sensor fusion
Tesla relies purely on cameras + neural networks (end-to-end AI), similar to human vision.

Waymo uses a classic "sensor stack" system: LiDAR, radar, cameras + rule-based software.

Tesla's advantage: the system is constantly learning, scales better and can also "see" and understand unfamiliar environments over time.

📈 2. scaling through real data
Tesla has millions of vehicles worldwide with FSD beta in use.

The cars collect realistic edge cases in everyday life (intersections, pedestrians, construction sites).

Waymo, on the other hand, only drives in a few well-mapped cities (e.g. Phoenix, San Francisco).

Tesla processes this data with a Dojo supercomputer to train its neural networks - a data-driven, scaling system.

🧠 3. end-to-end neural network
Tesla is developing a system that goes directly from camera recording to control output.

No intermediate step with fixed rules or modules - so no classic logic for "stop sign → brakes", but AI that "understands".

This increases the potential for human-like driving (even in chaotic situations).

🌍 4. geographical generalization
Waymo only works in geofenced zones - i.e. where detailed HD maps are available.

Tesla is planning a system that can be used worldwide and works everywhere without special mapping.

💸 5 Economy & mass market
Tesla's goal is to provide FSD for every car (with OTA updates).

Waymo is a fleet solution (robotaxi) with a high cost structure: expensive LiDAR systems, complex mapping, etc.

Tesla has a greater chance of delivering an affordable, globally usable system - a huge advantage in terms of investors and scaling.

••

PaulPK@PaulPK

@PikaPika0105 In Tesla's case, bad publicity does not equal bad business - it is mainly due to political commitment and driven by the left. This will fade faster than a fart in the wind. And if the FSD works out as planned + Optimus in mass production, Tesla will dominate markets again, this time probably unassailably.

••

PikaPika0105@PikaPika0105

@PaulPK It's all purely theoretical, none of it has been proven. It all sounds great, but in reality it's not always so good. And what about regulation? Safety will be a huge issue. Waymo only drives in well-mapped cities, but it does drive. A Tesla robotaxi has just crashed again and the cameras alone are a disadvantage in real-life conditions. The regulatory authority is also keeping a critical eye on them. How likely is approval in Europe, China or Japan? Rather low. Not only for security reasons, but also for strategic reasons. I therefore think the fantasies of world domination with regard to robotaxis are completely exaggerated. The shitstorm won't go away any time soon either, because these left-wing 50% of the population who are pushing this were previously the largest target group. It was the ecologists from California who drove Tesla and not the farmers from Tennessee. Optimus is another hype product. No sales, no application yet, no results. Once again, it's all fantasy and narrative. Not to mention the scandal that at their event, the things were simply remote-controlled by humans. I don't want to declare the robotics sector dead per se, but it's simply not worth the hype and there is no unique selling point that would justify the valuation. Until now, everything Tesla has done has been the best the world has ever seen, but I think this spectacle is coming to an end.

•

Tenbagger2024@Tenbagger2024

@PikaPika0105
The Chinese just laugh their heads off at such Tesla reports.
China is the place to be when it comes to autonomous driving

•

PikaPika0105@PikaPika0105

@Tenbagger2024 I think that all leading industrialized nations and regions will have their own systems and it would be too risky not to control something like this themselves. In the end, the battle is more likely to be who will prevail in the emerging countries that are not developing their own FSD technology (e.g. ASEAN countries, Latin America...). In my opinion, the networks of the corporations will be particularly relevant there. In other words, how long they have been on the ground, how good their contacts in politics are and how great the general trust in them is.

•

PaulPK@PaulPK

@Tenbagger2024 This is also where I see the greatest danger for Tesla. On the other hand, there was a time when many people laughed their heads off about Tesla :)

•

Charmin@Charmin

@PaulPK So. Rarely sustainable. I see. Is there anything well-founded about this steep thesis or is it just some Tesla disciple's wisdom?

Those who lead the way rarely stand still, but continue to develop. And with Google behind you, investing in the future is no problem at all.

Tesla has been working on FSD for almost ten years and can't get it right. I've said it umpteen times: there will only be a FSD vehicle with cameras n-i-e-m-a-l-s.

It doesn't matter for Tesla's share price - it will continue to rise until even the last one realizes that Tesla is still a mediocre electric car company at best.

•