Why robots still can't make Nikes: Nike's decade-long automation failure explains re-industrialization limits

Summary

Nike's decade-long effort to automate shoe manufacturing in North America has largely failed, exposing a hard constraint on re-industrialization that tariff policy alone cannot overcome.

Starting in 2015, Nike poured millions into automating footwear production with Flex, the electronics manufacturer that had just built Apple's Mac Pro factory in Texas. Nike aimed to produce tens of millions of sneakers at a high-tech facility in Guadalajara, Mexico by 2023, staffed with thousands of workers instead of the armies of laborers required in Asia. Under Armour launched a parallel effort called Project Glory to make shoes in Baltimore. Adidas opened speed factories in Atlanta and Germany. The underlying thesis was that rising Chinese labor costs and advances in 3D printing and automation made North American production viable.

It didn't work.

The core problem was mechanical, not economic. Shoe uppers use soft, stretchy fabrics that expand and contract with temperature. Robot grippers that work flawlessly on rigid electronics components struggle with squishy materials. No two shoe soles are identical, making precision placement impossible at scale. Tasks that seemed trivial in theory—gluing a sole straight, placing a swoosh logo, lacing the shoe—became months-long engineering problems. At one point, Flex spent eight months automating swoosh placement, only for Nike to move to a new shoe line where the solution no longer worked.

The factories ballooned with human workers to compensate. Nike's Guadalajara facility swelled to 5,000 employees, roughly double the original projection and costlier than equivalent labor in Vietnam. Production never became as automated as planned. By 2017, Flex's investors balked at rising costs. By 2019, Flex and Nike wound down the project. Under Armour stopped mentioning Project Glory to investors. Adidas shipped its speed factory technology back to Asian suppliers.

Product design, not manufacturing capability, is the real constraint. Shoe automation demands standardized products manufactured in high volume with minimal variation. Electronics work because they use hard materials, precise tolerances, and repeatable assembly. Nike shoes do not. The company sells shoes in roughly 20 different sizes for men alone, with 23 distinct parts per shoe. Designer freedom means constant style changes: new colorways, materials, silhouettes. Building automation for one shoe model, only to have designers demand a new line weeks later, defeats the economics of capital investment.

Apple solved this through vertical integration and design constraint. The iPhone uses standardized chassis, limited color palettes, and integrated hardware-software optimization. Tesla strips product variation to the essentials: no fender flares, no bespoke trims, minimal color options. Both companies convinced consumers that standardization is a feature, not a limitation.

Nike never tried. Manufacturing had no seat at the design table. Designers dreamed up new shoes and manufacturers were expected to deliver them, whether or not the production method could adapt.

The timing problem is acute. Trump's tariff threat is meant to force manufacturing home now. But robot leasing markets don't exist. There is no on-demand robot supply that can scale for seasonal demand or rapid design pivots. Building dedicated automation for footwear requires 10 to 15 years of patient iteration. Tariffs impose quarterly earnings pressure.

Tariffs can raise the cost of imports, but they cannot accelerate the underlying capital-intensity and technical complexity of automating labor-intensive, design-variable manufacturing. Nike, Under Armour, and Adidas each tried with serious funding and outside expertise. All three returned to Vietnam, China, and Indonesia. The constraint is not capital or policy. It is the fundamental mismatch between what consumers demand (endless variety) and what machines can efficiently produce (standardized, high-volume output). Without a redesign of the product itself, re-industrialization in footwear remains economically implausible regardless of tariff levels.