Tesla just started building Optimus robots on a real production line — how do you tell if the ramp is actually on track?

Tesla says early production of its first production-quality Optimus robots has begun on a high-volume line at Fremont designed for 1 million units a year — a claim, not a result. Watch five concrete milestones instead of headlines: the weekly build rate, AI5 chip integration around April 2027, the Optimus Academy, first outside deliveries around the end of 2027, and real reliability data on hands and joints.

Short answer: Tesla says early production of its first production-quality Optimus robots has begun on a high-volume line at Fremont, in the space the Model S and X used to occupy, on a line designed for 1 million robots a year. Treat every number in that sentence as Tesla’s claim or plan, not a guarantee — Musk himself said reaching 20,000 robots a week will take at least a year, maybe longer. The honest way to follow this story is a short checklist of milestones, below.Why it matters: a demonstration robot proves a design; a production line proves a business.

The gap between the two is where most humanoid projects quietly die, and Tesla names the hard parts itself — hands and dexterity, reliability, and how long the robots last before wearing out. A hand with many small joints is the most failure-prone part of the machine, and it has to survive industrial work, not a stage.Five milestones that tell you the ramp is real:1. The weekly build rate.

Designed capacity is 1 million a year, but the line has to climb there. Any confirmed weekly figure — and whether it keeps rising quarter over quarter — matters more than the capacity number.2. AI5 chip integration, targeted around April 2027.

Tesla’s next AI chip is meant to be the robot’s brain at scale; a slip here slips everything behind it.3. The Optimus Academy. Tesla plans a facility where 10,000–20,000 robots learn tasks together.

If it opens and fills, Tesla is training a workforce, not building prototypes.4. First outside deliveries, expected around the end of 2027, industrial customers first. A paying stranger using Optimus for real work is the milestone no demo can fake.5.

Reliability and lifespan data. Watch for working hours per robot and repair rates, especially on hands — durability is the difference between a tool and a toy.Pro tip: The common mistake is judging a ramp by announcements. Judge it by boring evidence: build rates, delivery dates kept, and robots still working a year later.Ad’s take: I want this to work — a robot that does dangerous, repetitive work is a genuinely good idea.

But I will score it on the checklist, not the keynote.