Five years. Ten years. By then, Musk insists, AI will be doing the driving for almost everyone. Ninety percent of all distance.
“It’ll be quite a niche thing to actually be driving your own car.”
He said it at a summit in Tel Aviv. Confidently. Like the weather report on a clear Tuesday.
Ignore the enthusiasm.
Look at the hardware.
Tesla just recalled 200,00+ cars in the US. Rearview cameras failing. Blind spots widening. Crash risks rising, as noted by the NHTSA. That doesn’t scream “reliable automation.” It screams “we shipped unfinished code.”
Waymo isn’t much cleaner.
They pulled nearly 4,000 robotaxis recently. Why? The cars tried to speed through floodwaters. Because the sensors said the road was clear. Water doesn’t care about lidar.
Why it won’t happen by 2028
Ali Kani, running the auto team at Nvidia, knows the math better than Musk knows Twitter. He called out “long tail scenarios.” Weird moments. Unforeseen chaos. Things the training data didn’t cover.
Remember last year in San Francisco?
A power outage. Stoplights died. Waymo’s fleet just… stopped. They froze. Stuck in the dark while humans rolled by. If a system can’t handle a blown fuse, how does it handle a drunk driver? A parade? A pile of leaves that looks like a pedestrian?
Don’t bet on it.
A 2025 World Economic Forum report kills the hype gently but firmly. No full driverless personal fleet before the mid-2035s. And even then? It won’t be everywhere. It’ll be patchy. Limited to specific cities.
Personal cars with Level 5 autonomy will be rare. Niche. Only about 4% of new vehicles sold globally. The rest will be robotaxis or big rigs on fixed routes.
Who gets there first
Europe is playing it safe.
Level 2 systems—steering helps, brakes help, but your eyes stay on the road—are universal there. Level 3 is approved for controlled spots, but regulators are twitchy. They watch every crash. London is running trials. But bureaucracy moves slower than a Model S launch.
The US and China? Faster. Dirtier.
Some American states are pushing robotaxis (Level 4) into dense traffic now. The car drives itself; no human needs to be ready to grab the wheel.
China might lead the pack. They have the manufacturing scale, the data, and a consumer base eager for novelty. But Level 5? Fully autonomous under any condition? Day, night, snow, fog, construction zones?
The International Energy Agency says it’s “not currently in sight.”
By 2035, expect maybe 3 million robotaxis worldwide. Scattered across 80 cities max. Not a sea of self-driving Sedans. Just pockets of automation in big tech hubs.
Musk sells dreams. Engineers build reality. And right now, reality has bad cameras and trouble with floods.
