The DSX Pod represents a significant evolution in how high-performance computing is deployed. Huang described the system as a "factory-built integrated system" that essentially puts an entire AI factory in a box. The architecture begins with the Vera Rubin NVL72, a liquid-cooled configuration that serves as the heart of the pod. Each DSX Pod consists of seven liquid-cooled compute racks, which are intricately interconnected by 17 kilometers of high-speed Spectrum X cabling. By integrating these components, cabling them, and performing rigorous factory testing before they ever leave the manufacturing floor, NVIDIA aims to deliver transportable containers that are essentially "plug-and-play" at the datacenter site. This approach is intended to reduce the "time to token"—the duration between the arrival of hardware and the generation of AI outputs—from a typical window of four to five months down to just a few weeks.
Central to the design of the DSX Pod is a focus on thermal and power efficiency, which are critical concerns for European operators facing strict environmental regulations and high energy costs. The pods utilize a dry cooling system rather than traditional evaporative cooling, which can consume vast amounts of water. Furthermore, the architecture employs 800V DC power delivery. While these technical specifications are part of the broader DSX reference architecture, their inclusion in a pre-certified, containerized format addresses the specific deployment and supply chain challenges that have become the primary hurdles for the industry. Huang emphasized that the scale at which the world is now producing computing power is "unimaginable," but noted that the physical reality of assembling these machines on-site remains a "crazy" endeavor involving tens of thousands of kilometers of wiring.
The logistical strain of building massive AI facilities was a recurring theme in Huang’s keynote. He cited the example of TeraWulf’s Lake Mariner site in New York to illustrate a growing industry-wide problem: the shortage of skilled labor. At that site, nearly half of the 1,800 tradespeople required were electricians. Huang pointed out that even when GPUs and memory are available, the sheer complexity of on-site assembly can stall a project for months. By moving the assembly, qualification, and "burn-in" processes back into the supply chain, NVIDIA hopes to eliminate the need for extensive on-site wiring and troubleshooting. This "pre-assembled" model ensures that by the time the computing power reaches the datacenter, it is already fully functional and ready to be integrated into the grid.
The economic argument for this acceleration is perhaps even more compelling than the technical one. Huang laid out the stark financial stakes involved in modern datacenter construction. A 1GW datacenter today represents a capital investment of approximately $50 billion. Operators investing such sums generally expect to generate annual revenues in the neighborhood of $40 billion. In this high-stakes environment, a three-month delay in deployment is not merely a scheduling inconvenience; it represents a missed opportunity of roughly $10 billion in revenue. "Even for me," joked Huang, whose personal net worth is estimated at $200 billion, "$10 billion worth of revenues will make anybody anxious." The DSX Pod is positioned as a solution to this anxiety, ensuring that expensive hardware does not sit idle while waiting for on-site assembly to catch up.
While the DSX Pods were presented in NVIDIA’s signature green branding, Huang was quick to highlight that this is a collaborative effort involving a vast ecosystem of partners. The presentation featured the logos of infrastructure giants such as Schneider Electric and Vertiv, alongside major server vendors including Hewlett Packard Enterprise (HPE), Lenovo, and Dell. NVIDIA confirmed that it is working closely with these partners to deliver DSX Pod AI factories to customers worldwide starting in early 2027. This timeline suggests a coordinated push to standardize AI infrastructure on a global scale, providing a roadmap for enterprises and governments to build out their sovereign AI capabilities.
Beyond the datacenter infrastructure, Huang also looked toward the future of physical labor within these facilities and beyond. He introduced updates to the Isaac robotics platform, headlined by a new reference humanoid robot named the Agile One SN. Standing 1.4 meters tall, the robot features five-fingered hands equipped with tactile sensing for dexterous manipulation, allowing it to interact with the world in a human-like fashion. The onboard compute is equally impressive, featuring 2,070 FP4 teraflops of performance and 128GB of unified memory. Huang announced that this reference robot would be available in the first quarter of next year, arriving pre-assembled and software-integrated. The implication was clear: as AI factories become more automated and factory-built, the robots that maintain them or work within the broader supply chain will also be delivered as turnkey solutions.
To support this massive hardware push, NVIDIA is aggressively expanding its global manufacturing footprint. The company announced a series of agreements with a group of strategic supply chain partners intended to bolster the resiliency and redundancy of its operations. This expansion includes partnerships with Amkor Technology, IBIDEN, King Yuan Electronics (KYEC), Samsung Electro-Mechanics, Siliconware Precision Industries (SPIL), and Unimicron Technology Corporation. A key highlight of these agreements is Samsung Electro-Mechanics joining NVIDIA’s substrate supply chain, providing immediate capacity from its South Korean operations.
NVIDIA’s strategy also involves a massive scale-up in semiconductor testing and optical engine production. The company expects its total semiconductor testing capacity to double between 2026 and 2027, with new buildouts planned for the United States, Taiwan, and Singapore. Furthermore, the partnership with SPIL is aimed at expanding the manufacturing capacity for optical engines used in co-packaged optics—a critical component for the high-speed networking required in AI clusters. NVIDIA targets a 15-fold increase in this capacity by mid-2027, with plans to double it again by the end of 2028. This aggressive expansion is designed to ensure that the physical components of the AI era can keep pace with the exponential demand for compute.
Throughout the keynote, Huang took the opportunity to reiterate his belief that the traditional era of computing, defined by Moore’s Law, has come to an end. He argued that the old model of transistor scaling can no longer provide the performance gains required for modern AI workloads. Over the last decade, Huang noted, NVIDIA has improved AI performance by a factor of one million. By contrast, had the industry relied solely on the 10x-per-five-years cadence of Moore’s Law, performance would have only improved by 100x in that same period. This "million-fold" improvement is what has made the current generative AI revolution possible, and it serves as the justification for NVIDIA’s shift toward full-stack, accelerated computing architectures.
The European context of the event was particularly relevant given the region’s intense focus on sustainability and energy sovereignty. Huang highlighted several European companies already utilizing DSX reference designs to optimize their operations. By focusing on liquid cooling and advanced dry coolers, these companies are able to reduce their reliance on local water supplies and decrease their total power consumption. This efficiency is not just about cost; it is about stabilizing the energy grid and ensuring that the massive power requirements of AI do not come at the expense of local communities.
However, the path forward is not without friction. The early portion of Huang’s speech was briefly interrupted by environmental activists, reportedly from Greenpeace, who protested the environmental impact of large-scale computing. The protesters also raised concerns regarding reports of NVIDIA technology being used in Russian military drones in the ongoing conflict in Ukraine. Huang handled the disruption with a composed demeanor, asking security to "go easy" on the protesters and acknowledging the importance of the issues raised while asking for the opportunity to finish his presentation.
The incident underscored the complex position NVIDIA occupies as the primary architect of the AI age. As the company moves toward the 2027 rollout of its containerized AI factories, it is clear that the challenge will be as much about navigating social and environmental expectations as it is about engineering faster chips. By promising to cut deployment times and improve efficiency through the DSX Pod and its expanded supply chain, NVIDIA is betting that a more integrated, factory-led approach is the only way to meet the world’s insatiable hunger for AI while managing the immense physical and economic costs of building the future.
