On January 28, 2026, during the Chinese New Year Gala — the most-watched television broadcast on earth, with over a billion viewers — a troupe of humanoid robots performed a synchronised Kung Fu routine on live national television. The audience did not laugh. A year earlier, they might have. Not anymore.

What changed was not the spectacle. What changed was the context. By early 2026, the global humanoid robotics landscape had shifted so dramatically that the performance was not a novelty — it was a preview. Nearly ninety percent of all humanoid robots shipped in 2025 came from Chinese manufacturers. Companies like Unitree and AgiBot delivered over ten thousand units combined out of roughly thirteen thousand shipped globally, while their Western counterparts — Tesla's Optimus, Figure, Agility Robotics — managed between one hundred fifty and five hundred each.

Unitree's G1 robot entered the market at sixteen thousand dollars. That is less than an average car. Morgan Stanley doubled its forecast for Chinese humanoid production in 2026 to twenty-eight thousand units, and that number may already be conservative.

This is not a technology story

Let me be clear about what this essay is and is not. It is not a breathless account of technological progress. I am not here to marvel at robots doing backflips. I am here because I run a business that automates human work, and I see what is coming with a clarity that frightens me.

Every day, from my desk, I build AI workflows for small and medium-sized businesses. I connect their CRM to their invoicing system. I automate their content production. I build pipelines that turn raw data into structured reports. Each workflow I deploy eliminates hours of human labour — not in ten years, but now, this month, this week.

Until recently, there was a comforting narrative: AI replaces cognitive work, but physical work is safe. Lawyers and accountants should worry; plumbers and warehouse workers should not. The hands, we told ourselves, are harder to automate than the brain.

That narrative died in 2025.

The China playbook

What China is doing with humanoid robotics follows the exact pattern it used with electric vehicles: massive state investment, aggressive domestic deployment, rapid cost reduction through scale, and then — export. BYD and Geely are already deploying humanoid robots from UBTECH on their production lines. These robots are not prototypes. They are working units that can navigate factory floors, handle components, and — in UBTECH's latest demonstration — autonomously swap their own batteries when they run low.

The implications for Europe are staggering. European manufacturing has spent decades competing on quality, precision, and process efficiency. That advantage evaporates when your competitor has access to tireless workers that cost sixteen thousand dollars, operate twenty-four hours a day, and improve with every software update.

But this is not primarily a trade story. It is a social contract story.

The silent decoupling

Here is what I see happening, and what almost no political debate acknowledges: productivity is decoupling from human labour. Not gradually, not theoretically — measurably, now, in the businesses I work with every day.

A marketing agency that employed five content writers now uses one writer and an AI pipeline. The output has not decreased — it has tripled. A logistics firm that employed twelve dispatchers now uses four, plus an automated routing system. An accounting practice that processed client files with six junior staff now processes them with two, plus a workflow built in three days.

In every case, the business is more productive. In every case, fewer humans are needed. In every case, the humans who remain are doing higher-value work — for now.

We built the entire social contract on taxing human labour. What happens when there is dramatically less of it to tax?

The arithmetic, again

Belgium's social protection system costs approximately 120 billion euros per year (Healthcare excluded). It is funded overwhelmingly by social contributions — taxes levied on wages. Every employee who earns a salary, every employer who pays one, feeds the system that provides pensions, healthcare, unemployment insurance, and disability benefits.

This funding model has a clear assumption: that there will always be enough salaried workers to fund the system. That assumption is breaking. Not because people are lazy, not because of immigration, not because of any of the things the political debate focuses on — but because technology is making human labour less necessary, one workflow at a time.

When I automate sixty hours of monthly data entry for an accounting firm, I am not just saving that firm money. I am removing sixty hours of taxable labour from the social security funding base. When a factory deploys humanoid robots, it is not just increasing efficiency — it is removing taxable employment.

Multiply this across every sector, every country, every year. The trend line is not ambiguous.

The political debate is miles off

And here is what makes me write this series: the most passionate political debates of our time are about issues that will be irrelevant within a decade. We argue about retirement age as if the pension system will still exist in its current form. We argue about immigration as if the competition for manual jobs will still be between humans. We argue about education as if the skills we teach today will still be marketable in 2035.

These debates are not wrong in their own terms. They are wrong in their frame. They assume a world that is already disappearing — a world where human labour is the primary source of both value and tax revenue. The world that is emerging is one where machines generate an increasing share of both, and the systems we have built to distribute prosperity have no mechanism to capture it.

Augmentation, not apocalypse

I want to be careful here, because the narrative of technological unemployment has been wrong before. The Luddites were wrong. The automation panics of the 1960s were wrong. Every prediction that machines would make humans obsolete has, so far, been wrong.

But there is a difference this time, and it matters. Previous waves of automation replaced specific tasks while creating new categories of work. The loom replaced the hand-weaver but created the factory worker. The computer replaced the typist but created the programmer. There was always a new layer of human work that emerged.

This time, AI does not replace a task — it replaces a capability. It can write, analyse, translate, code, design, plan, and reason. Humanoid robots can walk, grasp, navigate, and manipulate. Together, they cover the full spectrum of human productive capacity — cognitive and physical. The question is not whether new jobs will emerge. Some will. The question is whether they will emerge fast enough and in sufficient quantity to replace what is being lost.

My bet — and it is the foundation of everything I build in my daily automations — is on augmentation: humans and machines working together, each doing what they do best. That is the transition model. But even augmentation means fewer humans doing more work, which means less taxable employment, which means the funding base of the social contract is eroding regardless.

The pension absurdity I started with is just the most visible symptom. The underlying condition is a political system debating in one century while the economy moves into another.

In the next essay, Iwill look at something most democratic criticism ignores: the people who votefor the extremes – not to judge them, but to understand what drives them there.

‍