Uber aims to launch Europe’s first robotaxi service with Pony AI and Verne
https://www.theverge.com/transportation/900656/uber-pony-ai-verne-robotaxi-europe

The companies are already testing autonomous vehicles in Croatia, where they will launch their first commercial service.

Bloomberg:
Pony AI reports 2025 revenue up 20% YoY to ~$90M, total annual losses down 72% YoY to ~$76.8M, and a $75.5M net income in Q4, its first profitable quarter ever — Pony AI Inc. delivered its first profitable quarter ever, bolstered by a windfall from an early investment, rather than its main robotaxi business.
https://www.bloomberg.com/news/articles/2026-03-26/pony-ai-swings-to-profit-aims-to-launch-robotaxis-in-20-cities

Mercedes-Benz EQE:n MBUX-järjestelmä käyttää 24 gigatavua keskusmuistia pelkästään käyttöliittymän ja viihdejärjestelmän pyörittämiseen. Samalla Micron Technology arvioi, että tulevat tason 4 robottiautot voivat tarvita yli 300 gigatavua RAM-muistia. Ero kertoo siitä, kuinka nopeasti auton elektroniikka on muuttumassa hajautetuista ohjainlaitteista kohti keskitettyä tekoälylaskentaa.

Nykyauton muistinkäyttöä tarkasteltaessa yksi konkreettinen esimerkki löytyy Mercedes-Benzin EQE-mallista. Sen MBUX-järjestelmässä keskusmuistia on 24 gigatavua, ja se on varattu käytännössä kokonaan käyttöliittymän, grafiikan, navigoinnin ja viihdejärjestelmien tarpeisiin. Toisin sanoen pelkkä kuljettajan ja matkustajien käyttämä digitaalinen käyttöliittymä toimii jo samalla muistitasolla kuin muutaman vuoden takaiset palvelimet.

Ajotoiminnot ovat tästä erillään. Nykyautoissa kuljettajaa avustavat järjestelmät on toteutettu hajautetusti useilla ohjainlaitteilla, joista jokaisella on oma prosessorinsa ja muistinsa. Kokonaisuudessaan muistia on tyypillisesti kymmeniä gigatavuja, mutta sitä ei käsitellä yhtenä keskitettynä resurssina. Tämä arkkitehtuuri on kuitenkin muuttumassa.

Micronin mukaan kehitys kohti tason 4 autonomisia ajoneuvoja kasvattaa muistitarpeen kertaluokkaa suuremmaksi. Yhtiön arvion mukaan muistimäärä voi nousta nykyisestä noin 16 gigatavusta yli 300 gigatavuun ajoneuvoa kohden.

Kasvu ei johdu yksittäisestä tekijästä, vaan useiden teknisten vaatimusten yhdistelmästä.

The ostensible point of robotaxis is that they’re supposed to be safer drivers than humans.

Tesla, apparently, didn’t get that memo.

According to updated filings from the National Highway Traffic Safety Administration examined by Electrek, the Elon Musk-owned automaker has reported another five crashes involving its capital-R “Robotaxis,” for a total of 14 documented collisions since the service began operating in Austin, Texas last June.

Based on the mileage data shared in Tesla’s Q4 2025 earnings, the EV blog estimates that the Robotaxi fleet accumulated roughly 800,000 miles by mid-January. Divide that by fourteen, and that works out to a crash once every 57,000 miles. Compare that to Tesla’s Vehicle Safety Report, which claims that the average American driver experiences a minor collision every 229,000 miles, and it means the Robotaxis are crashing at a rate four times higher than human motorists.

It looks even worse when you compare Tesla’s fleet to Waymo’s, a company that Musk constantly tries to pick fights with, insisting that his self-driving cars are superior. Across over 127 million fully driverless miles, Waymo averages an accident around every 98,000 miles.

Even this already sizable gap may slightly flatter Tesla. In reality, Waymo’s fleet comprises over 2,000 robotaxis, while Tesla has less than fifty. Waymo operates in several major US cities, while Tesla is limited to a small area of just one. And though it’s been vague about how much it relies on remote teleoperation, Waymo cars are considered fully driverless, meaning there’s no human driver or safety monitor inside the vehicle. Or for that matter, outside it: Tesla reportedly resorted to having a human-driven car tail its Robotaxis to fulfill a Musk boast that the service would start giving rides with no supervisor in the cabs.

In short, Waymo and Tesla aren’t in the same league.

The ostensible point of robotaxis is that they’re supposed to be safer drivers than humans.

Tesla, apparently, didn’t get that memo.

According to updated filings from the National Highway Traffic Safety Administration examined by Electrek, the Elon Musk-owned automaker has reported another five crashes involving its capital-R “Robotaxis,” for a total of 14 documented collisions since the service began operating in Austin, Texas last June.

Based on the mileage data shared in Tesla’s Q4 2025 earnings, the EV blog estimates that the Robotaxi fleet accumulated roughly 800,000 miles by mid-January. Divide that by fourteen, and that works out to a crash once every 57,000 miles. Compare that to Tesla’s Vehicle Safety Report, which claims that the average American driver experiences a minor collision every 229,000 miles, and it means the Robotaxis are crashing at a rate four times higher than human motorists.

It looks even worse when you compare Tesla’s fleet to Waymo’s, a company that Musk constantly tries to pick fights with, insisting that his self-driving cars are superior. Across over 127 million fully driverless miles, Waymo averages an accident around every 98,000 miles.

Even this already sizable gap may slightly flatter Tesla. In reality, Waymo’s fleet comprises over 2,000 robotaxis, while Tesla has less than fifty.

Waymo operates in several major US cities, while Tesla is limited to a small area of just one. And though it’s been vague about how much it relies on remote teleoperation, Waymo cars are considered fully driverless, meaning there’s no human driver or safety monitor inside the vehicle. Or for that matter, outside it: Tesla reportedly resorted to having a human-driven car tail its Robotaxis to fulfill a Musk boast that the service would start giving rides with no supervisor in the cabs.

In short, Waymo and Tesla aren’t in the same league.

The delay isn’t an isolated incident. The NHTSA previously launched an investigation into Tesla for repeatedly failing to report crashes in time, sometimes several months after the accidents occurred. It also heavily censors its reports by redacting crucial details like crash narrative under the guise of protecting “confidential business information.” The practice is legal, but shady.

Over the past few months, Tesla has been making a major pivot away from its highly lucrative electric vehicles toward humanoid robots, a pet obsession of CEO Elon Musk’s. That restructuring campaign seems to be ahead of schedule, with Tesla’s electric vehicle sales plunging to astonishing lows.

The EV maker hit a four-year low in the US last month, selling 17 percent fewer cars in January compared to the same period last year, indicating the company’s core business is falling apart as Musk focuses his attention elsewhere.