The Jobs Aren't Coming Back.
The Robots Are.

The political promise behind automotive tariffs was direct and uncomplicated: bring the factories back, bring the jobs back. It was a message that resonated because it described something real — the hollowing out of American manufacturing employment over forty years — and pointed at a mechanism, trade policy, that felt proportionate to the wound.

The mechanism was wrong. Not because reshoring is impossible, but because the factory that left in 1998 is not the factory that comes back in 2026. The 1998 factory employed hundreds of people doing work that machines now do faster, cheaper, and without a UAW contract. When you force that factory back to American soil through tariff pressure, you don't recreate the 1998 employment base. You get a 2026 factory — and the 2026 factory's answer to American labor costs is not to hire American workers at American wages. It's to not hire them at all.

Tariffs didn't bring back American manufacturing jobs. They made the business case for eliminating them.

The Labor Assumption in the $30,004 Car

Earlier in this series, I built a component-level bill of materials for a five-passenger electric SUV in the Ford Edge class — 72 kWh LFP battery, 250-mile EPA range, full features, competitive in the segment BYD is dominating at $21,000 while generating a 21 percent gross margin.

The BOM landed at a $35,000 MSRP with an 18 percent gross margin, dropping to a $30,004 net purchase price after the $7,500 federal tax credit. Every line item was stated. Every assumption was declared. The one that matters most for this conversation is assembly labor.

That $1,800 figure was not naive. It reflected where competitive vehicle assembly actually happens — the Mustang Mach-E is built in Mexico. Stellantis builds there. The entire industry has been running its cost-sensitive programs through Mexican labor for decades, and USMCA compliance kept it inside the tariff fence.

The critical word in that last sentence is kept. Past tense.

The Tariff Regime Is Now Hostile to the BOM's Own Assumptions

The tariff architecture that was supposed to make American manufacturing competitive has created a new problem for the Mexico assembly assumption: Mexican content is now tariff-exposed in ways that the BOM's original cost structure did not fully account for. USMCA was designed to protect North American content. The current enforcement posture is testing the boundaries of what that protection covers and for how long.

This puts the $30,004 car in an uncomfortable position. The cheapest honest path to that price point runs through Mexican assembly at Mexican labor rates. The policy environment is actively hostile to that path. The alternative — American assembly at American labor rates — is more expensive by an order of magnitude unless something closes the gap.

That something is automation.

What the Math Looks Like When You Move Assembly Domestic

The UAW's 2023 contract established a wage trajectory that puts blended assembly labor in the $30-to-$35 per hour range when you include benefits and overhead — roughly three times the Mexican rate. Applied to 150 direct labor hours per vehicle, that takes the assembly labor line from $1,800 to somewhere between $4,500 and $5,250. On a vehicle with an $35,000 MSRP target and an 18 percent margin requirement, that delta — roughly $2,700 to $3,450 — has to come from somewhere.

The $3,000 gap between the two scenarios is not a rounding error. At an 18 percent margin target on a $35,000 vehicle, every dollar of cost increase requires roughly $1.22 in additional revenue to maintain the margin. A $3,000 labor cost increase requires either a $3,660 price increase — which puts the car at $38,660 before the credit that no longer exists — or a $3,000 reduction in cost somewhere else in the BOM.

There is nowhere left to cut in the BOM that doesn't compromise the vehicle. The battery is already sized to spec. The powertrain is already optimized. The infotainment is already stripped to the minimum the market will accept. The only remaining lever is labor — specifically, the number of labor hours required per vehicle. And the only mechanism that reduces labor hours without reducing output is automation.

The 18 Percent Margin Target Is the Business Case for the Robot

This is where the policy argument and the manufacturing engineering argument converge in a way that hasn't been made clearly enough.

Automation investment has always had a financial return. The question was always whether the payback period justified the capital outlay given current labor costs. At $12 per hour Mexican labor rates, a robotic system that replaces two workers has a payback period measured in years — long enough that many programs opted to keep the human labor and accept the ongoing cost. At $32 per hour American labor rates, that same robotic system pays back dramatically faster. The capital case for automation that was marginal at Mexican rates becomes straightforward at American rates.

But here's the mechanism that most of the automation conversation misses: it's not just the wage rate that changes the math. It's the margin target. A supplier or OEM that has committed to an 18 percent gross margin on a $35,000 vehicle has defined, implicitly, the maximum allowable cost structure for every line item in the BOM. When the labor cost line item exceeds what that margin target permits, there are two options: miss the margin target, or automate.

The margin target is the business case for the robot. It always was. The tariff regime just made it impossible to ignore.

What Automation Actually Buys in an Assembly Context

Not all of the 150 direct labor hours in a vehicle assembly are equal candidates for automation. Some positions have been automated for decades — body-in-white welding, paint application, powertrain installation. The frontier is in the labor-intensive processes that have historically resisted it: soft goods assembly, complex interior fabrication, quality inspection tasks that require the kind of judgment that human eyes provide cheaply and machine vision provides expensively.

The positions that are currently staffed by humans in a vehicle assembly plant and could be automated — where the technology exists and the process is in principle automatable — represent a meaningful fraction of those 150 hours. The reason they haven't been automated is not that automation is impossible. It's that the math hasn't justified the capital investment at prevailing labor rates and production volumes. That calculation is changing on both variables simultaneously: labor rates are going up and the production volumes required to justify the investment are coming down as automation technology gets cheaper.

A program that targets $35,000 MSRP with 18 percent margin at 100,000 units per year has a very different automation economics calculation than the same program at 30,000 units per year. At 100,000 units, the capital cost of an automated assembly cell amortizes across enough vehicles that the per-unit cost becomes competitive with — or better than — the human labor it replaces, even at Mexican rates. At American rates, the crossover point arrives earlier and the payback period shrinks further.

This is why scale matters so much in the affordable EV discussion. The $30,004 car is not just a price point. It is a volume commitment. And the volume commitment is what makes the automation investment rational — which is what makes the domestic assembly scenario viable — which is what makes the tariff policy coherent rather than self-defeating.

The Jobs the Tariff Advocates Were Imagining

The political coalition that pushed for automotive tariffs was not imagining a lights-out factory in Kansas City. They were imagining Lordstown at full employment — thousands of workers on the line, union cards, pension plans, the industrial economy of 1978 restored through the force of trade policy.

That factory is not available. The processes it ran have been automated, the institutional knowledge has dispersed, and the economics that made it viable — cheap domestic steel, limited global competition, a captive American consumer market — are gone. The tariff can force the factory back to American soil. It cannot force time to run backward.

What the tariff actually creates is an economic environment in which automation investment is more attractive than it has ever been — in which the payback period on robotic assembly systems shortens, in which the capital case that procurement teams used to reject now clears the hurdle rate, in which the supplier who has invested in automation capability is structurally advantaged over the supplier who has not. The beneficiary of the tariff regime, in manufacturing employment terms, is not the assembly line worker. It is the robotics engineer, the automation systems integrator, the controls programmer, the maintenance technician who keeps the line running when the robots need attention.

These are good jobs. They are skilled, well-compensated, and genuinely difficult to offshore. They are also dramatically fewer in number than the jobs the tariff advocates were promising. A lights-out assembly facility that produces 100,000 vehicles per year employs a fraction of the workforce that a comparable manual facility would require. The factory came back. The employment base did not.

This is not an argument against tariffs. It is an argument for honesty about what tariffs produce — and for policy frameworks that are designed around the manufacturing reality of 2026 rather than the political memory of 1978. The jobs that automation creates are worth fighting for. They require different training, different infrastructure investments, and different workforce development policies than the jobs the tariff coalition was imagining.

The $30,004 car is still achievable. The path to it runs through a factory that looks nothing like the one the tariff advocates were promising. The sooner policy is designed around that reality — investing in automation workforce development, accelerating the capital formation that makes domestic automated assembly viable, building the training infrastructure for the jobs that actually exist — the sooner the manufacturing economy that everyone wants becomes the manufacturing economy that everyone gets.

The robot doesn't get paid. But someone has to build it, program it, and keep it running. That's the job worth creating.