They started with a car.

It made sense at the time.

One call. One session. One narrow purpose.

If you needed a little extra, you bolted something on.

A roof rack for recording.

A trailer hitch for transcription.

A bungee cord for translation.

Another strap for identity.

Another for policy.

Another for analytics.

Five suitcases was manageable.

You could still see out the windshield.

Then came AI.

Translations.

Captions.

Agents.

Multimodal data.

Accessibility requirements.

Identity pipelines.

Continuous policy.

Suddenly they weren’t carrying five suitcases.

They were carrying fifty.

At first, they did what they always did.

Reinforced the rack.

Widened the straps.

Added mirrors.

Added warnings.

But weight doesn’t fail gracefully.

It doesn’t degrade linearly.

It tips you over.

That’s when the problem finally became visible.

It wasn’t that the car was slow.

Or that the road was bad.

Or that the engine needed upgrading.

It was that this was never a car problem.

Once you stop thinking in cars, everything else snaps into place.

Trains were never about speed.

They were about order.

A train doesn’t care how many cars you attach.

Each one locks into the same coupling.

Each one rides the same rails.

Each one obeys the same signals.

You don’t renegotiate the locomotive every time you add a car.

You don’t build a new track for luggage.

You don’t ask each car to decide where it’s going.

You decide once at the front,

and everything behind it follows.

That’s why trains scale.

And it’s why the industry never built one.