America May Miss the Robotics Wave Too

After losing solar, batteries, and EVs, we are about to watch the same movie again with embodied AI

I. The moment felt familiar

I have seen this mood before.

A conference stage lights up. A company shows off a machine doing something that looked impossible five years ago. Investors grin. Politicians say the right patriotic things. A few writers announce a new age. Then the country goes back to arguing about apps while somebody else builds the factories.

That is why this week bothered me.

I was reading through a fresh round of reporting on the rise of “physical AI” at GTC, on humanoid systems entering real industrial settings, and on warnings that America already lost key industrial races and could lose this one too. And the old American disease came back into view. We are brilliant at invention, superb at demos, theatrical in our self-praise, and strangely sleepy about the dull work that decides who wins.

The dull work matters. It always has.

The country that controls the stack for robots, sensors, simulation, compute, supply chains, and factory deployment will not merely sell machines. It will shape labor, logistics, defense production, warehousing, elder care, construction, and eventually domestic life. That is not a niche market. That is the skeleton of the next industrial order. NVIDIA’s GTC 2026 devoted major attention to “Physical AI,” with platforms aimed at bringing AI into warehouses, roads, hospitals, and homes, while several recent Substack pieces describe humanoids and simulation systems moving from spectacle into rollout.

And yet America still talks about robotics as though it were a science fair for adults.

That is the part that makes me uneasy. We do this every time. We treat industrial mastery like a side quest and branding like the main campaign. Then, ten years later, we wake up and discover the factories are elsewhere, the margins are elsewhere, the engineers are elsewhere, and our grand national plan consists of asking whether we can subsidize our way back into a game we refused to take seriously when the board was still being set.

A nation can coast on old greatness for a while. It cannot coast forever. That bill arrives like a tax collector in hard shoes.

II. This is not about robots in a showroom

The word “robotics” still misleads people. They hear it and picture a metallic butler, a factory arm, or a silly humanoid falling over on a polished floor while reporters clap like seals at an aquarium. That picture is already too small.

The real story is that software is getting a body.

Recent writing on world models and simulation as a bottleneck-breaker, along with reporting on digital twins spreading into industry, hospitals, and infrastructure, points to a larger shift. The contest is no longer over who can make a clever chatbot. It is over who can build systems that perceive, predict, move, manipulate, and act in physical environments at scale. This includes the model, the edge hardware, the sensors, the synthetic training environments, the control software, the deployment pipeline, and the factory culture capable of using them well. World models were highlighted at NVIDIA GTC this week as a way to help robots simulate and respond to physical environments, and digital twins are already being used in sectors from health care to infrastructure.

That changes the stakes.

When AI leaves the screen, it stops being a parlor trick for office workers and starts becoming a lever on the physical economy. Warehouses change. Assembly lines change. Military logistics change. Energy systems change. Even immigration politics may change in time, because a country that can automate more physical labor has more room to decide who enters, who works, and under what terms. There is a reason serious people are suddenly paying attention.

And this is where America becomes strange. We have capital. We have research talent. We have real companies. We also have a bad habit of treating production as morally and culturally secondary, as though the clean hands belong to software while factories are for the grubby cousins. That attitude was foolish when manufacturing meant smokestacks. It is even more foolish now that manufacturing means code, sensors, vision systems, simulation, and machine control.

One recent piece on advanced manufacturing and the American failure to organize around new realities put the issue plainly. Modern manufacturing produces more with fewer workers, but the workers who remain are more skilled and better paid. That ought to have been good news. Instead, many Americans still talk as though the only choices are a return to 1957 or surrender. There is a middle road. It involves becoming frighteningly good at the new industrial arts.

That would require discipline. America prefers enthusiasm. Enthusiasm is cheaper, and it leaves more room for keynote speeches.

III. The labor question is where this turns political

This subject gets real the minute it touches work.

For years, the public argument around AI stayed half-abstract because a chatbot replacing a few tasks in marketing or coding still felt distant to many people. Robotics does not feel distant. A machine loading parts, moving inventory, assisting with assembly, or inspecting goods lands differently. People understand at once that this is about wages, leverage, status, and whether their son will have a trade that still belongs to human beings.

That is why I keep circling back to recent pieces on a labor market growing thin outside healthcare, on layoffs piling up as firms rework headcount around AI, and on humanoids already contributing inside a live factory environment. The old reassurance, that robots are always ten years away, is growing threadbare. Recent Substack reporting says BMW’s Figure pilot in Spartanburg contributed to production of more than 30,000 BMW X3 vehicles over a ten-month period, while separate labor-market analysis argues U.S. job growth has been weak outside healthcare and that AI-related deskilling risks are rising.

Now, I am not saying every warehouse will be packed with humanoids by Christmas. Plenty of hype will still collapse under gravity. Machines break. Integration is ugly. Costs matter. Human environments are messy. The robot future will arrive in patches, not with a trumpet blast from the clouds.

But that does not rescue the complacent view.

If even a modest portion of repetitive physical work becomes more machine-friendly over the next decade, labor politics will change. Employers will gain options. Workers without technical fluency will lose bargaining power. Regions that combine industrial land, reliable power, vocational retraining, and local manufacturing culture will pull ahead. Regions that cling to managerial fantasy, or worse, to nostalgia, will sag.

This is why the national conversation cannot stay at the level of spectacle. The key question is not whether a robot can dance, wave, or hand a man a screwdriver in a promotional clip. The key question is which country builds the plants, trains the technicians, owns the suppliers, secures the data, and writes the operating assumptions for the rest of the world.

When a society misses that question, it turns politics into vaudeville.

Then the vaudeville gets very expensive.

IV. China is not merely building machines. It is building position

Americans still love the bedtime story in which all serious rivals are either fraudulent or brittle. That story is soothing. It is also a fine way to become poor.

What concerned me most this week was the convergence of two lines of argument. One came from warnings that the United States already lost solar, batteries, and EVs and could repeat the pattern in robotics and AI. The other came from recent analysis of China’s lead in robotics for advanced manufacturing and broader accounts of the PRC’s long-term tech competition framework. These are not isolated commercial stories. They point to state capacity, industrial patience, and a willingness to treat production as strategic terrain rather than an embarrassing relic. Recent reporting argues China has surged ahead in robotics for advanced manufacturing, while separate analysis notes the PRC’s current five-year planning ties technology competition directly to national security and state direction.

America, by contrast, often behaves like a rich heir rummaging through the family silver while muttering about freedom.

There is another wrinkle here that people prefer not to discuss. Industrial robots and embodied systems are also sensors. As one recent warning stressed, these machines can function as mobile surveillance platforms, equipped with cameras, microphones, LiDAR, and network pathways that create security questions far beyond ordinary consumer devices. That means the robotics race is not merely about who sells the forklift with a brain. It is about who gains visibility inside warehouses, plants, labs, and infrastructure.

That should wake up even the laziest strategic thinker.

A robot in a factory is not like a toaster in a kitchen. It sees. It maps. It transmits. It can become part of a workflow, a dependency, and possibly an intelligence exposure. If America sleepwalks into foreign dependence in this layer of the industrial stack, then we will have managed a truly comic achievement. We will have imported the eyes and ears of the next manufacturing age and called it cost savings.

That would be a very modern kind of stupidity. Sleek, optimized, and purchased with a smiling procurement memo.

V. What I think America actually needs

I do not think the answer is panic. Panic is a cousin of vanity. It assumes we can recover by sheer emotion.

I think the answer is to recover respect for production.

That means treating robotics as a national capability question. It means more machine shops, more testing grounds, more vocational routes into advanced manufacturing, more secure supply chains, more grid capacity, more domestic tolerance for ugly industrial buildout, and far less sneering at the people who make physical systems work. It also means building local cultures around applied competence. Engineers, technicians, fabricators, controls specialists, supply-chain operators, and software people need to inhabit the same civic world more often. Countries win these races when skill becomes social, not when it remains trapped in a few glamorous firms.

The good news is that America still has real assets. Recent coverage of the broader physical AI push, the strategic importance of embodied systems, and the high-end U.S. bet on generalized robot intelligence suggests the country still has first-rate research, capital, and platform companies. American firms are still serious players in physical AI, and U.S.-backed humanoid companies are competing by trying to pair stronger general intelligence with robot hardware rather than merely rushing units out the door.

But assets are not enough.

You can have talent and still lose. Britain had talent. So did Venice. History is a museum full of brilliant people who discovered, a little late, that cleverness and command are not the same thing.

What I want is a country that remembers how to turn discovery into durable position. A country that respects the factory as much as the lab. A country that trains boys who like machines without telling them they are relics. A country that can look at a humanoid robot and ask boring, serious questions about suppliers, controls, data access, deployment cost, and workforce adaptation instead of behaving like a medieval crowd seeing a clock for the first time.

That country could still win this.

But it will have to stop mistaking commentary for capacity. That habit has already cost us enough.