NVIDIA GTC Berlin 2027: Did the keynote content change at the last moment and nobody (AI included) take note?

NVIDIA GTC Keynotes are spectacles.

One man (NVIDIA CEO Jensen Huang). Speaking for upwards of two hours. A few video breaks thrown in. Hardware and cute robots on stage. Lots of applause from all the NVIDIA and Jensen fans in the audience. At GTC Berlin this week, I was one of the folks in the audience enjoying the spectacle alongside the many people tuning into watch parties.

Now, anything the world’s most valuable company does gets plenty of coverage. That means it isn’t too difficult to get a view of key takeaways from the keynote. You can check my view against everyone else, but I don’t think you’ll see much disagreement on the main messages.

• Europe, Europe, Europe. At the tail-end of a multi-country European trip, the main message was that NVIDIA loves Europe. It was literally spelled out multiple times and backed up with investment, customer and partnership citations along with a focus on the industrial economy.
• Open, Open, Open. NVIDIA’s support for open-source / open-weight models is no secret. It was always going to be a recurring theme in its GTC “opener.” (See what I did there?).
• NVIDIA and the AI Stack. Early in the keynote, Huang took time to ask, “what is AI?” His answer (energy + compute + factories + models + applications) was more than a philosophical detour. It was a framework he could return to but also use to explain that NVIDIA’s value lies well beyond the chips it’s best known for.
• Agentic Security. In talking up OpenShell (secure runtime and sandbox environment for agents), Huang said “this technology has arrived at exactly the right time.” I’m not sure all the folks who spent time this past year battling rogue agents would agree that the timing was perfect. Their travails, however, help to explain why the topic of AI agents and agentic security got so much airtime.
• HW Innovation: small, medium, large. It wouldn’t be a GTC Keynote without product showcases. RTX Spark-based laptops and workstations. A reminder of the power delivered by scaling via the DGX SuperPOD. Somewhere in between? The new NVIDIA DSX POD, pre-fab AI factory infra built on the Vera Rubin promising token production in a month vs. typical four-to-five month timeframes.
• The Physical AI Opportunity. It also wouldn’t be a GTC keynote without robots – big and small, humanoid and industrial. Add mobile RAN and medical imaging into the definition of Physical AI (which Huang did) and it’s hard to argue with the size of the opportunity.

Huang has talked about AI and GPU investment as a money maker (“revenue-generating assets”) on plenty of other occasions. Take this spring’s Morgan Stanley Technology, Media & Telecom Conference where he said, “I’m certain compute equals revenues.”

This wasn’t a major focus at the GTC Berlin keynote. You wouldn’t know that, however, if you just tuned in for the last 5 min.

In what was billed as a keynote recap, a robotic Eurovision-style song and dance (somewhat cringey, but pretty darn fun) focused a lot on AI economics; GPUs as “cash machines,” passive income and automated returns, AI factories as superior investments to traditional financial markets. If you look at the time spent on these themes in the “recap” it was about 20-25% of the end sequence. That’s more than any other theme and well-outsized compared to the actual keynote content itself.

The dissonance was stark and leads to two obvious questions. Why did the recap diverge from the actual content it was recapping and why does that matter?

I don’t have all the answers, but I’ve got some ideas.

• Who okays the final slides? Does Jensen Huang make the last edits on his keynote slides and decide what gets included (or not)? He seems like that sort of guy. Could those changes be too last-minute to get reflected in an AI-generated recap? I suspect so.
• What’s the Euro message? After stops in various European countries, the right keynote message at GTC Berlin was the European opportunity, European AI progress and NVIDIA’s role in supporting it all. It was all covered well in the keynote, and rightly so. Sure, the money of it all matters. Making AI economics the central keynote story would have been a distraction.
• Why didn’t anyone catch this earlier? Long before the keynote recap was built, someone should have seen this coming; the right message for a European GTC keynote did materialize overnight. If that call was made in the final moments, it explains why Jensen is the man who takes the stage.