Code and electrons will never replace atoms
At the inaugural Invest in Germany Summit in Berlin, Jensen Huang, NVIDIA’s CEO who had just seen his company claim the world’s largest market capitalization, sat down for a fireside chat with German Chancellor Friedrich Merz.
Here was a man who had built a multi-trillion-dollar fortune selling compute power and pure digital code, turning up in Europe’s preeminent industrial powerhouse to walk through the ledger of the physical world.
Many assume the rise of large language models means software will rewrite everything.
Jensen Huang’s take runs directly counter to that.
The past fifty years belonged to computing and coding.
The United States rose to global leadership in this field, yet made a fateful choice along the way: it gradually offshored most capacity in heavy industry, construction and foundational manufacturing processes.
Now the ground rules have been reset entirely.
For half a century, competition hinged on who could write more sophisticated code.
Not anymore.
Computers have started writing their own code.
State your requirement, and the model generates the code itself.
The pure programming edge accumulated by the prior generation is quickly being flattened by large AI models.
So where will the next great dividing line emerge?
Physical AI.
That is, artificial intelligence that intervenes directly in the physical world.
Robotics, automotive manufacturing, heavy machinery, precision machining.
Huang put it plainly: No matter how capable general artificial intelligence grows, it can never supplant specialized mechanical and electrical expertise built for niche industries.
Elegant software code alone cannot produce reliably operating machine tools and gears.
This gets to Germany’s trump card.
Germany is home to hidden champions entrenched in vertical sectors, holding decades of on-site industrial data and craft know-how.
Huang describes this data as a nation’s natural mineral resource.
It belongs to local firms, tied to specific production workflows.
Exporting proprietary industrial data to train overseas models, then buying those trained models back, makes no commercial sense.
Yet Germany faces a clear vulnerability: an aging population.
Its most skilled factory veterans are growing older year by year.
If that generational craft knowledge is lost, so goes its manufacturing heritage.
That is why AI’s most urgent value for manufacturing is not drafting better presentations.
Its job is to encode seasoned technicians’ process parameters, assembly intuition and hands-on tuning instincts into the digital brains of Physical AI — locking that expertise permanently into local production lines.
Huang even warned Chancellor Merz: This industrial transition carries extreme stakes. Assume this is the only window of opportunity, whether future chances exist or not.
This was no mere stage rhetoric.
NVIDIA has poured nearly a billion US dollars into Germany over the past year. That same week of the summit, it backed four local German startups focused squarely on embodied robotics and commercial space engineering — heavy engineering frontiers.
Chancellor Merz conceded in response: The decision to shift much heavy manufacturing offshore was, in hindsight, a major US misstep.
Outside a handful of cutting-edge segments, the domestic base of foundational manufacturing was hollowed out.
Germany must hold fast to its physical manufacturing capacity and never abandon that ecosystem.
On the fiercely debated topic of European AI regulation, Huang laid out a clear boundary: Security matters, but security does not come from slogans. It is rigorous engineering work.
To build sensible regulatory frameworks, local industries must first survive and scale.
You cannot regulate an industry that has never taken root.
While much of the world chases the hype of purely virtual models, the top supplier in the compute chain is sounding an alarm for the old industrial heartland:
The industrial foundation of tomorrow still rests on steel and concrete.
Code can evolve on its own, but the skilled hands that machine precision parts reliably remain an irreducible threshold no one can bypass.
If you enjoy what I do, consider supporting me! Every little bit means the world! ❤️
☕ Buy me a coffee