When it comes to the recent or pending investment in the US, some are interested in the company names – TSMC, Samsung, Wistron, Pegatron.

I’m more curious about the technologies they might bring.

Reason is, all the capital investment in the world won’t matter if you don’t have the personnel to operate the factories.

Or maybe not.

Word comes down from Foxconn, the world’s largest electronics manufacturer, that it plans to deploy humanoid robots to build AI servers for Nvidia at its Houston plant.

This marks a big step – assuming the robots have feet, of course – in the automation of industrial manufacturing. (Interestingly, Foxconn will use Nvidia’s own Isaac GR00T N AI platform to train and operate the robots.)

I’ve seen AGVs in action in printed circuit board manufacturing since at least 2000, when I watched them moving laminate sheets around the spacious Isola plant in Düren, Germany. The best example I’ve seen to date: Universal Scientific’s Shanghai plant, where AGVs cart product throughout the multi-story building, loading and unloading material in a variety of assembly machines and testers.

Many companies have implemented cobots, too, and exhibitors at Productronica this month promise to have all sorts of flavors on display. Humanoid robots, however, are a different, uh, animal. This class of machinery has a form factor resembling a mid-size person, with arms, legs and the ability to walk upright.

Designers have been working on these bots for years, but attention ramped in 2021 when Elon Musk (who else?) announced plans to develop a general-purpose humanoid robot capable of performing a variety of tasks. The reported list price of about $30,000 would be attractive to many companies, not just those in areas where labor was hard to come by or prohibitively expensive, or both.

Telsa’s Optimus robot can serve popcorn. Can it make servers?

Moreover, the mobility and dexterity hyped by the designers suggest a certain mastery of handling complex physical tasks. This, coupled with easier programmability, makes the Foxconn project worth watching. Indeed, while Foxconn might be building lots in the hundreds of thousands of pieces, most domestic manufacturers are higher-mix, lower-volume operations, and that doesn’t seem likely to change in the near-term, even if a lower-cost bot were available. But if Foxconn’s Houston plant becomes the benchmark it foresees for AI-powered smart manufacturing, it will still represent a significant advancement in industrial production.

Even better: humanoid robots are an end-market where manufacturers can benefit not just in the factory but in the accounting suites too. Goldman Sachs Research forecasts that the total addressable market for humanoid robots will reach $38 billion by 2035 and will fill some 4% of the US manufacturing labor shortage in the next five years. What could be better than taking orders to build a product, then using that product to build even more orders?

Given the current landscape of low unemployment, a perceived lack of desire among younger generations in pursuing careers in manufacturing, and domestic policies that discourage immigration, humanoid robots can fill a burgeoning void in the workplace.

Whether they are worth hanging out with after hours, however, remains to be seen.

P.S. PCEA, on the other hand, is worth hanging out with! See us this month at PCB Carolina and Productronica (stand B3 202).

Mike Buetow is president of PCEA (pcea.net); mike@pcea.net.

• Next