Huang shrugged off protesters to open Nvidia's GTC Berlin event, highlighting DSX Pods and robotics.

GTC Europe opened with Jensen Huang promising to cut the time to first tokens at new datacenters from months to weeks, with a Vera Rubin-powered AI factory in a box.

The DSX Pod, essentially a containerized version of NVIDIA's DSX reference architecture, was the hardware highlight of the NVIDIA CEO and founder's keynote. The speech was largely about assuring Europe that it had a future in the AI era, and it namechecked key partners such as Schneider Electric and Siemens Electric while unveiling new technology.

Huang described the Pod as "bringing this full-stack platform into a factory-built integrated system." According to a video presented during the keynote, it "starts with Vera Rubin NVL72. Seven liquid-cooled compute racks connected with 17 kilometers of Spectrum X cables form a Vera Rubin DSX pod. Vera Rubin DSX pods are integrated, fully cabled, and factory tested, then delivered in transportable containers."

The design relies on dry cooling rather than traditional evaporative cooling, and on 800V DC for "efficient power delivery."

None of this is new in itself, as these elements are part of the standard DSX architecture. However, NVIDIA is targeting the deployment and supply chain challenges facing datacenter operators.

"The scale at which we are now producing computing is unheard of ever. Unimaginable," said Huang. "Well, even then, when you bring these systems onto your sites, there are still months and months of assembly, and you just heard 17,000 kilometers of wiring to connect just Spectrum X for the scale-out computing. It's a crazy number."

That is why buildouts such as TeraWulf's Lake Mariner site outside Buffalo, New York, run into shortages of electricians before they have to deal with shortages of GPUs and DRAM. Nearly half of the 1,800 trades on site are electricians.

Huang said, "Now suppose we were to do that all inside the supply chain, run everything up, build everything up, make sure that it runs completely, get it qualified, burned in, and by the time we ship it, it is fully functional. The amount of cabling and on-site assembly is going to be greatly reduced."

Just as importantly, he said, "By moving this into the supply chain the amount of idle computing, the amount of idle computing that's going to be sitting inside at your datacenters will reduce dramatically."

After all, he said, a 1 GW datacenter costs about $50 billion, and operators expect to generate revenues of $40 billion from that infrastructure. A three-month delay, he said, means missing out on $10 billion in revenue. Even Huang, who is worth an estimated $200 billion, concedes that "$10 billion worth of revenues will make anybody anxious."

"By moving this into the supply chain, the speed by which we can turn on, go from installation and arrival of computing to production of tokens, will go from, we think, something along the lines of four to five months down to about a month."

While the images of the Pods were in reassuring NVIDIA green, the presentation also displayed the logos of key partners, including Schneider Electric and Vertiv, as well as server vendors HPE, Lenovo and Dell.

NVIDIA said in a statement that it was working with them, and others, "to deliver DSX POD AI factories worldwide in early 2027."

More bad news for onsite contractors, arguably, came in the form of updates to NVIDIA's Isaac robotics platform, including a reference humanoid robot, Agile One SN. The robot has a 1.4-meter-tall humanoid body with five-finger hands for "tactile sensing for dexterous manipulation," multimodal sensing, 2,070 FP4 teraflops of onboard compute, and 128GB of unified memory.

"This reference robot will be available in the first quarter of next year," said Huang. "And it will come pre-assembled with computer inside, software integrated, and it should be able to perform quite well right off the bat."

On the supply chain front, the company announced plans with "a group of supply chain partners to expand global manufacturing capacity to meet growing demand for its accelerated computing and full-stack AI factory platform."

NVIDIA said it is working with Amkor Technology, IBIDEN, King Yuan Electronics (KYEC), Samsung Electro-Mechanics, Siliconware Precision Industries (SPIL) and Unimicron Technology Corporation.

The aim is to "expand critical manufacturing capabilities across substrates, advanced packaging, testing and optical engines, strengthening the resiliency and redundancy of NVIDIA's global supply chain."

Under the agreements, Samsung Electro-Mechanics is "joining NVIDIA's substrate supply chain, bringing immediate capacity from its existing operations in South Korea."

NVIDIA said it expects its total semiconductor testing capacity to "double from calendar 2026 to 2027, supported by buildouts in the US, Taiwan and Singapore."

The deal with SPIL means expanded manufacturing capacity in Taiwan for optical engines used in co-packaged optics. NVIDIA said, "The expansion targets an increase of 15x by mid-2027, with aims for another doubling by the end of 2028." How the shift of existing and new capacity into NVIDIA's ecosystem will affect other silicon suppliers and their customers remains to be seen.

Huang also took the opportunity to once again pronounce Moore's law dead. "In the course of that last 10 years, we've improved AI performance by 1 million times. If you would take a step back and think about what Moore's law would have done in 10 years' time, the first five years would have improved by a factor of 10. The second five years by another factor of 10, a 100x improvement in 10 years' time."

Separately, Huang cited a range of European companies using NVIDIA's DSX reference designs, which center on liquid cooling and dry coolers, to reduce water and power use and to stabilize the grid.

The datacenter industry is facing significant pushback across Europe over environmental and sovereignty concerns.

The early part of the speech was disrupted by activists, thought to be from Greenpeace, who shouted about "environmental destruction" and reportedly cited reports about the use of NVIDIA technology in Russia's drones in Ukraine.

Huang remained composed, telling security to "go easy… give her a chance," and saying, "I'll talk to you guys later… all important issues. But let me get through my keynote."