特斯拉 FSD 中国落地：自动驾驶、数据主权，与一个慢了一拍的领跑者

On May 21, 2026,

Tesla officially announced

that its "FSD Supervised Edition" autonomous-driving system

has gone live in China.

This day,

for Tesla,

for China's electric-vehicle industry,

and even for US-China tech relations,

is a major moment.

Today,

starting from this day,

we'll talk about several bigger questions.

First,

what is FSD?

Is it really "Full Self-Driving"?

Second,

why has this matter

waited so many years?

The politics behind it,

the data rules,

and the mapping problem —

what exactly are they?

Third,

and the most interesting:

when Tesla finally arrives in China,

what it faces is a market

completely different from a few years ago.

Let's first talk about FSD itself.

FSD,

in English fully stands for "Full Self-Driving";

the literal Chinese is "完全自动驾驶."

It sounds

like the science-fiction-movie kind:

you sit in the back seat,

and the car drives you home by itself.

But the reality is:

the current FSD,

its official name is actually

"FSD Supervised Edition."

The two words "supervised"

point at the truth:

the car drives itself,

but a person must

sit in the driver's seat watching,

ready at any time to take over.

According to industry classification,

the kind of autonomous driving where people don't have to watch the road

is called L4 or L5.

All "self-driving" production cars on the market today,

including Tesla's FSD,

in essence

are still at the L2+ level.

That is to say,

legally

and in practice,

the responsibility for driving

still sits with the driver.

With that understood,

we can now look at

why this matter dragged on for so long.

Tesla's FSD

went fully live in the US in 2024,

and then progressively entered Canada,

Mexico,

and some European countries.

But in China,

it was stuck for a full two years and more.

The reason

is not a technical problem;

it's a rules problem.

China set up

two key barriers for autonomous driving.

The first

is data localization.

Autonomous driving requires collecting massive amounts of

road-condition,

road,

and even driver-behavior data.

By Chinese rules,

this data

must be stored within China's borders;

it cannot be transmitted directly abroad.

For this reason, Tesla,

back in 2021,

built in Shanghai

a dedicated China data center.

But cutting the entire

cross-border data pipeline

completely clean

and making it compliant

itself takes many years.

The second

is mapping.

Autonomous driving relies heavily

on high-precision maps.

And China requires:

high-precision maps

can only be made by Chinese companies

with "Grade-A surveying qualification."

For this, Tesla

partnered with Baidu,

using Baidu's high-precision maps

to power FSD.

These two barriers —

one is data sovereignty,

the other is surveying security —

are both red lines for Chinese regulators.

Tesla satisfied them one by one,

taking several years.

So this "launch"

isn't a sudden technical success;

it's compliance finally walking through.

There is also a backdrop we cannot ignore:

politics.

In early May,

US President Trump

made a state visit to China.

Elon Musk,

as one of the Trump administration's

important advisors,

went to Beijing along with him.

About a week after the visit ended,

FSD officially went live in China.

The timing of these two events

is very close.

Observers broadly believe:

FSD's final clearance

is not only an administrative approval victory —

it's also the US and China,

in technology and trade,

making subtle mutual concessions.

But

what's most interesting

isn't "Tesla finally arrived" —

it's

"what kind of China Tesla arrived in."

In 2024,

when FSD went live in the US,

China's smart-driving market

was still very early.

At that time,

Tesla was the unquestioned

"synonym for autonomous driving."

But in these past two years,

China's market has gone through

a quiet but very deep change.

Xpeng's XNGP,

Huawei's ADS,

Li Auto's AD Max,

BYD's "God's Eye,"

and the much younger Xiaomi —

almost every one of them

has rolled out their own

highway navigation,

city navigation,

and automatic-parking systems.

What's even more crucial,

their business models

are completely different from Tesla's.

China's smart-driving features

are either given away free,

or bundled into the car's price,

or charged only a small fee.

But Tesla's FSD,

priced in China at 64,000 yuan,

one-time purchase.

By comparison,

it looks expensive.

What's even more subtle is the pricing model itself.

Just this past February,

Tesla, globally,

eliminated FSD's "one-time purchase,"

switching entirely to subscription.

Car owners in other countries

can now only pay monthly.

But in China,

Tesla specifically kept

the "one-time purchase" option.

The logic behind it is clear:

the China market is too important

to miss.

And Chinese consumers

are generally cool to "long-term subscription";

they prefer to "spend once,

use forever."

For this, Tesla

made a

business-model concession

specifically for the China market.

This event,

its symbolic significance may be

even bigger than the financial impact.

So,

how do we evaluate this "launch"?

Different angles

yield very different conclusions.

From Tesla's perspective,

this is a major victory:

the gates of the China market opened,

the entry ticket for autonomous driving

in the world's hardest market.

From the perspective of Chinese homegrown automakers,

this is a "wolf is coming" —

but this wolf

is not as scary as a few years ago.

Because Chinese rivals

have already, on home turf,

planted their feet.

From the perspective of Chinese consumers,

this is one more choice,

but not a must-buy.

64,000 yuan

isn't cheap.

From the perspective of technological progress,

all of the world's "autonomous driving"

is still within the L2+ range.

The truly "person-not-required" —

that future —

is still very far away.

At this point,

I want to leave some more general reflections.

This event

is in fact a concentrated expression

of several trends.

The first trend:

global technology companies,

to enter large markets,

increasingly have to play

by local rules.

Where data is stored,

who makes the maps,

how the model is trained —

none of these are decisions

tech companies can make unilaterally.

"Data sovereignty"

is becoming a question

every country has to answer.

The second trend:

the China market

is moving from being a "recipient"

to a "participant,"

even a "front-runner."

Five years ago,

when China brought in a new technology,

it was often "they teach me."

Today,

when Tesla finally comes in,

local players

have already built

products that are no worse, and often cheaper.

This kind of role reversal

is happening in many fields.

The third trend,

perhaps the most important —

no matter what manufacturers call this thing,

"Full Self-Driving" or anything else,

the person who, in the end, sits in the car

and is responsible for the steering wheel

is still you yourself.

No matter how advanced the technology,

it won't carry the risk for you.

Marketing can be very bold;

the laws of law and of physics

remain very cold.

Finally,

let me leave you with two questions.

The first:

if one day

cars really can fully drive themselves,

would you be willing to let a computer

decide, completely, your driving safety?

Or

do you hope

the one who, in the end, "makes the decision"

is still a person?

The second,

a bit closer to home:

today's "self-driving"

is still far from fully automatic.

So,

for car makers

to continue using the word "automatic"

in their marketing —

is that appropriate?

These two questions

have no standard answer.

But they

are worth thinking about slowly.

Okay.

That's all for today.

See you in the next story.