Tesla Now Needs Its Own Maps for FSD. Musk Called Maps “a Huge Mistake”

Tesla is building its own maps. On the Q3 2026 earnings call, a retail shareholder asked why Tesla still relies on third-party maps, when speed limits, parking and typical routes “continue to be a problem” for Full Self-Driving. Tesla’s VP of AI, Ashok Elluswamy, agreed: “we shouldn’t be depending on external people’s maps.”

He said Tesla is building a new AI-based mapping and navigation system, fed by what the whole fleet learns, that will also learn each driver’s preferences and respond to voice commands. It will start with the beginning and end of trips in FSD v15 and, over the following months, take over navigation, parking and destination handling. Tesla gave no date.

In this article: Seven years ago, maps were “a huge mistake” · A 30 km/h sign is not a “serious signal” · Waymo did it smarter · Paying Elon for every promise · Maps from millions of cars? I will believe it when I see it · The one thing I like: talking to the car · A promise to hold the stock price · Unsupervised FSD in 2027? I am scared · FAQ

Seven years ago, maps were “a huge mistake”

At Autonomy Day in 2019, Elon Musk said Tesla had tried high-precision lane maps and realized it was “a huge mistake.” He called it a “really bad idea” that makes self-driving “extremely brittle.” In the same presentation he called lidar a “fool’s errand.” The message was clear: Waymo needs lidar and maps, Tesla will do it with cameras and AI alone, anywhere. To be fair, Tesla is not talking about centimeter-level HD maps now. But it is building its own map layer to fix what its cameras get wrong.

Now I turn to the Tesla fans. That is how Musk is now: he only claims that everything is OK, and then he starts doing something. That turns my stomach. How did it take so many years for them to understand that FSD is a threat on the road? How much had to happen before they said they will make their own maps?

And what turns me even more: why didn’t they admit that you should not rely on FSD as long as you don’t see our maps? Again it is only “we are working on it,” “it will be in 2027.” But we know that when Elon Musk’s companies talk about anything, it can easily take three or four years. He will find a way to defend it. It did not work because it snowed. Then it rained and the cameras could not read the centimeters well. And until he sees that Tesla is completely useless and losing its position to the competition, he will not start doing anything. He promised that cameras and AI are enough, and now it has caught up with him and turned against him. A drama queen, in my opinion, and I mean Elon Musk.

Year	What Tesla said about maps
2015	Building high-precision maps “100 times” more detailed than navigation, from customer cars
2019	Musk: high-precision maps were “a huge mistake,” a “really bad idea”; lidar is a “fool’s errand”
2026	“We shouldn’t be depending on external people’s maps.” Tesla builds its own AI maps for FSD v15

A 30 km/h sign is not a “serious signal”

Look at what Tesla wants to fix: speed limits, parking, routes. Last month, a Belgian road safety group tested FSD in Brussels and found that it sped in 55% of the 30 km/h zones it tested, at an average of 44 km/h, partly because the car showed a 50 km/h limit on its screen. In the US, NHTSA is weighing an investigation into claims that Tesla Robotaxis speed through active school zones, as I wrote in my article on the school zones.

And how people think today: they say it is a “serious signal” that the car cannot read signs. I am completely lost. A serious signal? That is a reason to take the cars off the road immediately. How can it not keep the speed limit? That is a threat on the road. How is it not tested, and why does nobody say: please pull those cars of yours, before there is a serious accident?

Waymo did it smarter

Electrek now calls Tesla’s approach the same thing Waymo does, only without lidar. Waymo is smart. When they really wanted to do it properly, they did it, even with that lidar on the roof. I agree it is not very pretty, but it is safe.

And as far as I know, when there is a problem, Waymo pulls its cars itself. Waymo has issued voluntary software recalls, including one over its cars passing stopped school buses. It does not just make noise that it is not true, that it is not what you think. Elon only wants to show that his cars are the best. But the further we go, the more he understands that the road is slowly getting shorter. The stock is falling, and people are starting not to believe him.

Paying Elon for every promise

People paid up to $15,000 for FSD with the promise that the car would drive itself everywhere. Should they get money back? The one who made the promise should answer that question. But he will surely stay silent. How else would he collect his pay package? Musk owns about 13% of Tesla, and his new pay package, worth up to $1 trillion, could take him to about 25%. One of its goals is exactly 10 million active FSD subscriptions.

Don’t you all find it strange that we have to pay Elon in advance for every promise, and then we just wait years while he keeps promising more? I don’t even know if it is legal. That really fascinates me, and I am not the only one asking. In December 2025, a California judge ruled that Tesla’s “Autopilot” name misled customers and broke state law. Tesla dropped the name, added that FSD needs supervision, and then sued the regulator anyway.

And this is not his first promise he collects money for. He collected the same way for a “flying” car: Roadster reservations started back in 2017, nine years ago, and we are still waiting. Now FSD for $15,000, promising the car will drive everywhere by itself. In Europe, Tesla has sold FSD for years, last for €7,500 and now €99 a month, while the full FSD (Supervised) was approved only in the Netherlands, in April 2026. Even there, the regulator said: “A vehicle with FSD Supervised is not self-driving.” And I am curious about that fleet. Does he have it everywhere in the world where Teslas are sold, or only in the US and China, where he sees the most money?

Maps from millions of cars? I will believe it when I see it

Some say maps from millions of Teslas could be a huge advantage, because no other company has so much data. Do you believe that, when you think about how many years we will wait for such a map? I am just curious how he can come out of this as the winner.

After so many years, I think if anyone is in a hurry, it will be Chinese cars like BYD, and BMW. They will do it faster and they will sell more. Tesla is expensive, with Elon’s promises. I don’t know why, but in the last two years, and especially in the last month, I don’t believe anything Tesla and Elon say. To me it looks like he only wants money in advance for his new ideas, which we will never get. That is how I see it.

The one thing I like: talking to the car

Tesla says the new system will learn your habits and respond to your voice, like “park at the second entrance.” That is a mega feature when I think about it. Controlling the car with my voice feels very futuristic, and to me it really looks like the biggest innovation Tesla can make.

But as I said, I don’t believe it. And even if it comes, I am afraid Tesla will need years more to fine-tune it.

A promise to hold the stock price

Why did Tesla admit this now, on an earnings call where profit fell 43%? I lean only toward the investors. When we look at the chart, Tesla is already losing a lot of its value, down about 30% this year. So this is just a promise to hold at least the price that is on the market now. But I think even investors don’t believe in Tesla that much anymore. I already called Tesla a factory of promises.

Unsupervised FSD in 2027? I am scared

Do I still believe unsupervised FSD will come in the second quarter of 2027? Anything can happen. But I am afraid of Teslas without supervision. I am really afraid of those cars when I hear that they decide the speed by themselves. Without supervision, that will be mega scary.

Maybe in the end it is only my fear, and maybe it will be OK. Or there is one more option: we won’t see that promise even in four years.

Also on Evolt News: Tesla driver says he was asleep on FSD when he hit a police car · Tesla and SpaceX merger

Photo: Ian Maddox, CC BY-SA 4.0, via Wikimedia Commons.

This article is opinion.

Tesla FSD maps: FAQ

Is Tesla building its own maps for FSD?

Yes. On its Q3 2026 earnings call, Tesla AI chief Ashok Elluswamy said Tesla is building its own AI-based maps and navigation from fleet data, starting with FSD v15. Tesla gave no release date.

What did Elon Musk say about maps?

At Tesla Autonomy Day in 2019, Musk called high-precision lane maps “a huge mistake” and a “really bad idea” that makes self-driving “extremely brittle.” He also called lidar a “fool’s errand.”

Are Tesla’s new maps HD maps like Waymo’s?

Tesla has not said it will build centimeter-level HD maps. Its new maps are built from what its camera-equipped fleet sees, while Waymo maps cities with lidar-equipped cars.

When will unsupervised FSD arrive?

Tesla now targets the second quarter of 2027 for unsupervised FSD, and only inside its Robotaxi service areas, not in all customer cars.

Is Tesla FSD available in Europe?

FSD (Supervised) was approved in the Netherlands in April 2026. EU-wide approval still needs a European Commission process and a vote by member states.

Sources: Electrek, DigitalToday, Teslarati, Heise, Notebookcheck, RDW.

Our Take

Tesla now says it needs its own maps for FSD, seven years after Elon Musk called maps “a huge mistake.” For me, Tesla only changes course when it is losing, and an FSD that cannot keep a 30 km/h limit belongs off the road.

Voice control could be Tesla’s best idea. But after years of paying in advance for promises, I will believe it when I see it.