The March of Nines

Fresh out of grad school, my first real job was engineering at Motorola, and the first thing they did was teach me quality. Motorola had invented a discipline called Six Sigma, and they were almost religious about it. The idea is easy to say and brutal to hit: build a process so good that you count the defects in parts per million. Six Sigma, the gold standard, meant no more than 3.4 defects for every million chances to make one. In an age when most factories were proud of 99%, Motorola was chasing 99.99966%.

That number is really just a count of nines. 99.99966% is about five and a half nines of reliability. Add one nine and you have cut your defects by ten. It is a ladder, and every rung costs ten times what the last one did.

I thought about that training this week, because on Tesla’s earnings call Elon reached for the same ladder and gave it a different name. He called it the “march of nines.”

Here is where Tesla stands on it. Ashok Elluswamy, their head of AI, said the unsupervised robotaxis have driven more than 380,000 miles with “zero notable incidents.” Run that against the ladder. Five nines of safety is one incident in 100,000 miles, and they are past it. It took about twenty cars, because a robotaxi runs almost around the clock and piles up miles no person could. So the driving is already in Motorola’s old neighborhood of quality.

Elon said he ultimately wants eight nines, 99.999999%, one incident in a hundred million miles. That is three hundred million clean miles to prove, and even growing “literally exponentially,” as Ashok put it, that rung is a couple of years out.

Here is my own read, straight from the Motorola training. Five nines is a great test. It is not yet a product. At Motorola, six sigma was the line where a process was finally trustworthy enough to ship and stop babysitting. For a two-ton robot that drives my kids around, I would want a nine past even that. I think seven nines, one incident in ten million miles, is where Tesla’s robotaxi stops being an impressive test and becomes a real product. It is the point where they can flip on unsupervised driving for the cars people already own and open Cybercab reservations to buyers.

So when does Tesla reach ten million clean miles? The call left another clue. Elon said the fleet is driving more than ten percent more unsupervised miles every week than the week before, and compounding like that moves faster than it sounds. Say last week was about 40,000 unsupervised miles. Grow it ten percent a week, add it to the 380,000 already on the board, and the fleet crosses ten million cumulative miles early next year

.

Now the part that turns a date into an opportunity. On the same call, an analyst asked whether Tesla would run its robotaxis through another network, an Uber or a Lyft, to keep the cars busy. Elon said no. Tesla will run its own. And then the tell. He said demand “will outstrip our ability to service the demand.” His own words. They will not be able to build cars fast enough.

Put those together. The point where a robotaxi becomes a real product is early next year, and the company building the cars just told you it cannot make enough of them. A company already running negative free cash flow, spending more than $25 billion, borrowing another $30 billion, is not going to personally finance every earning machine in America. That is not a moat. It is a supply gap with a date on it.

That gap gets filled by the people who own the machines. Someone buys the car, plugs it into a network, and collects while it works. And it is not only Tesla. It is the Cybercabs they will sell, the delivery vans, the fleets that need financing and insurance and a place to park and charge.

That is the bet I am making with Remount. Own the bots that do the work. Rideshare, delivery, entertainment. The safety is nearly proven, and the product bar is months away, not years. The one thing nobody has built is the fleet, because the company you would expect to build it just told you it cannot.

More below on what actually moved this week.

Jonas

---

This week’s evidence

Record spend, borrowed future. Tesla’s quarter looked profitable, but the profit leaned on a $1 billion paper gain on its SpaceX stake. Underneath, automotive margin excluding credits slid from 19.2% to 16.3%, free cash flow went negative, and capital spending more than doubled. The company guided past $25 billion of capex this year and said it is arranging debt to borrow up to $30 billion more. This is a company spending like wartime to build the autonomous future by itself.

https://www.cnbc.com/2026/07/22/tesla-tsla-q2-2026-earnings-report.html

Own the demand, short the supply. An analyst asked whether Tesla would run its robotaxis through other networks like Uber to keep them busy. Elon said no, Tesla will run its own rider network. Then the important part: demand “will outstrip our ability to service the demand.” A separate exchange pinned the fleet at dozens of cars, not hundreds, which Tesla explained by noting a robotaxi drives nearly continuously. So it is their own network, their own cars, and by their own account not enough of them.

https://electrek.co/2026/07/21/tesla-robotaxi-tampa-orlando-austin-fleet-stalls/

The march of nines. Elon named the single thing gating robotaxi growth, and it was not manufacturing. It was reliability, the “march of nines,” and he wants “99.999999%.” Eight nines. Ashok said the fleet has driven 380,000 miles with zero notable incidents, on early versions of Tesla’s V15 software, which lands it around five nines today. The distance between five nines and eight is the distance between a demo that works and a fleet that scales. Tesla just told you it is a march.

https://www.tesla.com/en_US/2026-q2-earnings

If the cars are nearly cleared and the company building them cannot make enough, the fleet has to come from somewhere other than one balance sheet. It comes from the people who own the machines. That is the whole idea behind Remount. The modeler at remount.co runs one earning machine’s numbers start to finish. Reply anytime. I read every one.

You’re getting this because you signed up at remount.co. This is a thought experiment, not financial advice and not an offer of securities. Reply anytime or write hello@remount.co.