Elon's Biggest Bet: Why Optimus Could Crash Labor From $46 to Under $2 an Hour

There is one number behind almost every price you pay. The shirt. The meal. The house. The Amazon box you did not need. That number is the cost of an hour of human labor. In the United States, the fully loaded figure sits near $46. Wages, health insurance, retirement, payroll taxes—the whole stack—for one person doing one job for one hour.

Warehouses pay it. Factories pay it. Restaurants, farms, and construction sites pay it. Sometimes more. A few weeks ago Tesla put a marker on the table that most people still have not priced in. The third-generation Optimus humanoid has a hand with roughly 50 actuators and about 22 degrees of freedom. Those little motors in the fingers and wrist are the hard part. Get them right and the robot can do thousands of distinct tasks that used to require a person.

The signal nobody can ignore

Elon Musk has been blunt about how hard this is. Tesla could not buy the actuators for any amount of money. They did not exist at the specs Optimus needs. The company designed every electric motor, gearbox, and control board from physics first principles. On difficulty, he put Optimus between the Model X and Starship—harder than Cybertruck or Gigafactory builds, not quite Starship. Right now the bottleneck is hardware, and the hands plus forearms carry most of the engineering pain.

Then came the real tell. On the Q4 earnings call in January 2026, Tesla said it is discontinuing the Model S and Model X—two cars that put the company on the map—and converting those Fremont lines to manufacture humanoid robots at millions of units a year. You do not sacrifice icon products unless you believe the next thing is so much larger that the cars stop being the main story. The pitch underneath that move is brutal: take average human labor from about $46 an hour down to under $2 for a humanoid at scale. I have covered Tesla since 2012. I have never heard a signal this loud.

How you get under two dollars an hour

If Optimus cannot hit that number, the rest is theory. If it can, the labor market rewrites itself. The math has four pieces.

First, the machine. Tesla's scale target is roughly $20,000 to $25,000 per unit. That is not science fiction. Unitree in China already ships full-size humanoids in a similar band and moved thousands of units last year. Second, utilization. A human works about 2,000 hours a year. A robot can run closer to 20 hours a day with time for charging and maintenance—call it 7,000 hours a year on a conservative schedule. Spread a $25,000 robot over three years and 21,000 hours and the hardware alone is about $1.20 per hour. These machines are electric with far fewer moving parts than a car, so a three-year life is a cautious assumption, not a stretch.

Third, energy. A few hundred watts while working. At commercial power rates that is maybe 10 to 15 cents an hour. Fourth, maintenance—joints, software, occasional repairs—another 20 to 30 cents when you smooth it out. Then stack everything you stop paying: health insurance, HR, training, legal, parking, turnover, recruitment, payroll taxes, productivity dips. Fully loaded, you land somewhere between $1.50 and $2 an hour. Against $46 for a human, that is roughly a 96% cut.

Where the economics hit first

A warehouse picker often costs $18 to $23 an hour before benefits. A robot at two bucks can pick, pack, and sort around the clock with no training pipeline and no turnover. Amazon alone employs more than 750,000 warehouse workers. Manufacturing wages in the U.S. average around $36 an hour. Welding, assembly, inspection—same tasks, different cost curve. Home health aides run near $30 an hour, and the country already has hundreds of thousands of unfilled elder-care roles. The shortage is not a lack of need. It is a lack of people willing to do the work at the wages on offer.

And the people problem is deeper than wages. The U.S. birth rate fell to about 1.6 children per woman—well below the 2.1 replacement rate. South Korea hit 0.72, the lowest among major economies, with a workforce that could shrink by half in a generation. China's working-age population peaked in 2011 and has contracted for more than a decade; total population has fallen three years running. Across the OECD, the combined working-age population peaked around mid-2025 and is now declining. The factory workers, warehouse staff, and caregivers of 2035 through 2055 were never born. You cannot manufacture a 25-year-old human on a production line. You can manufacture robots.

Why Tesla, not just "robots in general"

Figure AI has BMW running Figure 02 units in Spartanburg—tens of thousands of parts loaded across more than a thousand operational hours with sharp efficiency gains. Unitree shipped a large share of global humanoid volume last year. Boston Dynamics has electric Atlas in Hyundai plants. 1X is taking pre-orders on a home robot near $20,000. Plenty of teams are in the race.

Tesla's edge is the stack. It designs its own AI training chips. It writes its own software. It trained vision models on data from millions of cars over billions of miles. A robot on a factory floor and a car on a highway are cousins: see the world through cameras, decide what to do next. Tesla already builds millions of complex machines a year. Fremont lines are being converted for Optimus. Gigafactory Texas is aimed at a dedicated robot factory with a long-term target on the order of 10 million units a year. Competitors outsource chips, training clusters, or automotive-scale manufacturing. Every outsourced layer is a margin, a wait, and a negotiation. Tesla owns silicon, software, actuators, batteries, and the factory floor.

Calibrate the hype—and the damage

Honesty matters. On that same earnings call, Elon said no Optimus units are doing useful work in Tesla factories yet. They are in R&D, learning, collecting data. Gen 3 is the first design built for mass production, and the unveil is still later in 2026. If you expected laundry folding in the living room next quarter, reset that clock.

Jobs are the other hard edge. About 3.5 million Americans drive trucks. Close to 2 million work in warehouses. Millions more sit in manufacturing, food service, and retail. When robotic labor is cheaper and more reliable than human labor, pressure builds over five to ten years—not overnight, but too fast for anyone who pretends nothing changes. Demographics help in elder care and some factory roles where humans already will not show up. That cover story will not hold everywhere. Some people get displaced. Without real retraining and transition programs, the messy middle gets uglier. The weaver-to-factory shift of the 1800s is the closest rhyme, except software and robots move faster than power looms did.

Jevons, cloth, and what $2 labor unlocks

In 1810 a skilled weaver needed a full day to make one piece of cloth. By 1830 one power loom could match about ten hand weavers. The Luddites smashed machines because they saw the end of their craft. They were right that their way of life ended. They were wrong that work itself ended. Cloth got cheap. Demand exploded. Ready-made clothing, fashion, textile exports—new industries grew on the ruins of the old price. Within decades more people worked in textiles than before the loom.

Economists call the pattern Jevons paradox: when steam engines got more efficient, coal use rose because cheaper energy meant more factories, trains, and ships. Cheaper transistors put computers in everything. Cheaper storage created TikTok and YouTube. When something gets dramatically cheaper, usage does not rise a little. It jumps, and new demand appears that did not exist at the old price.

Map that onto labor. A $40 trillion-plus global labor market is the prize Tesla is chasing. If the cost of an hour of work-equivalent collapses by 95%, total work done does not shrink—it expands into jobs and services that were never viable at $46 an hour. Picture an autonomous van outfitted as a mobile salon or ice cream shop, with a humanoid delivering the service driveway to driveway. Transport cost falls through autonomy. Labor cost falls through robots. Niche ideas that looked ridiculous suddenly clear the hurdle. Multiply that by a thousand categories.

That is the bet. Kill two legendary cars. Convert the lines. Build the hand from scratch. Chase under $2 an hour. The transition will be messy for a lot of people. The opportunity on the other side is the largest product Tesla has ever pointed at—and one of the loudest industrial signals of this decade.

Check the video here.