Tesla's slickest move these past two years has been blurring the line between selling a car and selling a business, and leaving you to guess which one is actually on the table.

September 3, Austin — the Cybercab event.$Tesla(TSLA.US) 

Related leveraged ETFs: $XL2CSOPTSLA(07766.HK)（HK market） $TSLA 2X Long ETF(TSLL.US) $TSLA 2X Short ETF(TSDD.US) 

Most of the coverage,and abroad, is still stuck on the vehicle itself: how it looks, whether the interior is just one more screen, whether the wireless charging actually works.

But let's be honest — what the Cybercab looks like stopped being a mystery back at the "We, Robot" night in October 2024. Two seats, no steering wheel, no pedals, not even side mirrors. 

The spy shots are out, the prototypes have been torn down. The first unit came off the line at the Texas plant on February 17, production ramped in April, and there are now more than sixty of them parked around the campus for testing.

What actually matters on September 3 isn't the sheet metal. It's:

• How far along is FSD v15?
• When does Cybercab reach genuine scaled operation?
• And can Tesla turn its autonomy stack into a Robotaxi network of real size?

FSD v15 is probably the real headline

v15 is already running inside the Robotaxi fleet, with changes reportedly along seven fronts. The number that keeps coming up: roughly 10× the parameter count of the current version.

In the autonomy context, that figure is really a statement of belief — Tesla is betting that once the end-to-end approach crosses a certain scale threshold, the long tail starts falling away on its own.

The hard part of self-driving has never been going straight. It's the scenarios you can't enumerate. A detour around roadworks. A mattress tumbling off the truck ahead. A traffic officer waving you through while the light is still green. Someone clutching a cardboard box stepping out from between two parked cars in the rain.

Google's Waymo attacks this by mapping the physical world precisely with LiDAR, loading in priors via HD maps, keeping remote takeover as a backstop, and winning one city at a time.

Tesla's answer is: give me enough video and a big enough model, and it'll figure it out itself.

Which path is right? Honestly, nobody knows yet. But after September 3 we'll at least know how far Tesla's has gotten.

Autonomous driving platforms (Robotaxi)

ADAS chips / Domain controllers / Full-stack solutions

Chips / AI compute

Fractional shares: start from $1 and build a position flexibly 

Recurring investment: set up a DCA plan and build in steadily through the swings