Nvidia is treating 2026 as a reset moment for artificial intelligence, not a routine product cycle. Nvidia CEO Jensen Huang is arguing that everything from chips and data centers to software and national industrial policy has to be rethought if AI is going to move from clever text generators to a pervasive infrastructure. When he says Nvidia is rebuilding AI from the ground up, he is talking about a full stack redesign that reaches from the physics of robots to the politics of manufacturing.
The “new age of AI” and a changing stack
Jensen Huang has started describing the current moment as a new age of AI, and he is framing it as a structural break rather than an incremental upgrade. In his recent remarks, Nvidia CEO Jensen Huang has stressed that the entire computing stack is being reworked to meet surging global demand, from the way data centers are built to how developers write applications, and he has tied that shift directly to the rise of generative and agentic systems that behave less like tools and more like collaborators. That is the context in which he talks about rebuilding AI, not as a slogan but as a description of a multi‑layer engineering overhaul.
Huang’s language is unusually sweeping for a chip executive, and it is backed by a cadence of announcements that touch every layer of that stack. In Jan, Nvidia CEO Jensen Huang used a high profile stage to argue that the “entire stack is being changed,” presenting AI as a once‑in‑a‑generation platform shift that will reshape how companies design hardware, write code and deploy services, a message reinforced in a second segment of the same Jan keynote that highlighted how quickly expectations for AI systems are rising.
Rubin chips and the hardware foundation of “ground up” AI
If AI is being rebuilt, the most visible concrete is Nvidia’s new Rubin platform. Huang has positioned Rubin as the hardware base for this next era, with six new chips and an AI supercomputer designed to push performance and efficiency far beyond the current Hopper generation, and he has linked that roadmap to a belief that AI infrastructure will underpin everything from cloud services to industrial automation. The Rubin family is not just about faster training; it is pitched as the engine that will let companies run more capable models at scale and bring advanced inference into mainstream products.
That ambition is already moving from slides to factories. Nvidia CEO Huang has said that the next generation of chips is in full production, and he has claimed that the new Rubin architecture can cut the cost of running AI systems by 10 times, a leap that would change the economics of deploying large models in production and make it easier for smaller firms and public institutions to participate in the AI boom, as he outlined in a detailed CES address.
From data centers to “trillions of dollars” in AI infrastructure
Huang’s ground up rhetoric is also financial, not just technical. He has argued that AI infrastructure is becoming its own industry, with Nvidia, CEO Jensen Huang explicitly envisioning an AI infrastructure industry worth “Trillions of Dollars” as companies and governments build out data centers optimized for generative workloads, high bandwidth networking and accelerated storage. In his COMPUTEX keynote, Huang described how every major cloud and enterprise will need to retool their data centers around accelerated computing, treating GPUs and AI systems as the new default rather than specialized add‑ons.
That scale of investment is why he keeps returning to the idea that Nvidia is still at the beginning of a long buildout. In Sep, Nvidia, Jensen Huang Calls Its New Technology a “$4 Trillion Infrastructure Opportunity” and said They are Just at the start of this deployment wave, projecting that customers will keep pouring capital into AI data centers for the next five years as they modernize their fleets and chase productivity gains, a forecast he laid out while describing the $4 trillion opportunity.
Agentic and physical AI: rebuilding what AI actually does
Underneath the hardware, Huang is trying to redefine what AI systems are expected to do. He has embraced the idea of Agentic AI, where models are not static chatbots but dynamic systems that can decide on the right action at the right time, plan across multiple steps and interact with tools and environments. That shift from passive prediction to active agency is central to his claim that the AI stack must be rebuilt, because it demands new orchestration software, safety mechanisms and evaluation methods that can handle autonomous behavior rather than single prompts.
Huang has also started talking about “Physical AI,” describing Agents as “essentially digital robots” that can perceive, understand and plan, and he has argued that these systems will be crucial for semiconductor and electronics manufacturing as well as logistics and robotics. In one segment of his COMPUTEX remarks, he said the next wave of AI is physical and that Today’s models do not yet understand basic physics, They can reason about language but not reliably about forces and motion, a gap he wants to close with new simulation tools and Physical AI agents.
From “Super Bowl of AI” to CES: showcasing the rebuilt stack
Huang has used his biggest stages to stitch these ideas together into a narrative of reinvention. At GTC 2025, often described as the Super Bowl of AI, Huang focused his keynote on Nvidia’s advancements in AI and his prediction that the next generation of models will have the ability to reason, positioning Rubin and its software ecosystem as the platform for that leap in capability and for a new class of applications that behave more like colleagues than calculators. That same event, At GTC, was where he framed Nvidia’s role as architect of a new AI era rather than just a supplier of accelerators, a framing that has since become central to his public messaging.
By the time he reached CES in LAS VEGAS, Huang was using live demos to show what that rebuilt stack can do in the real world. One presentation video showed a human driver who did not touch the wheel a single time, letting Alpamayo do all the work on a public road, an illustration of how Nvidia’s automotive platform is moving from driver assistance to full autonomy and how its chips, software and simulation tools are converging into end‑to‑end systems, a point that was broken down in coverage of the Nvidia CES 2026 live keynote.
Europe, industry and the “next industrial revolution”
Huang’s argument that AI is being rebuilt from the ground up is also geopolitical. At the GTC Paris 2025 conference, Jensen Huang’s Keynote was framed as Setting the Stage for Europe’s AI Future, and he used that platform to urge European companies and policymakers to invest in accelerated computing, sovereign AI capabilities and the infrastructure needed to Enter the Agentic Era, presenting Nvidia as a partner for a continent that wants to compete with the United States and Asia in advanced computing. That message was tailored to European concerns about digital sovereignty and industrial competitiveness, but it fit neatly into his broader thesis that every region must retool for AI.
In a separate message, Jensen Huang Says NVIDIA Is Leading the Next Industrial Revolution, Powered by AI, and he has argued that accelerated computing will transform every major global industry, from automotive and healthcare to finance and manufacturing. By casting Nvidia, CEO Jensen Huang as a kind of industrial strategist, he is trying to convince governments and boardrooms that AI is not a niche technology but the backbone of a new production system, a claim he reinforced in a widely shared video about the next industrial revolution.
Rebuilding AI also means rebuilding manufacturing
Huang has been unusually blunt about the risks of relying on offshore manufacturing for the hardware that underpins AI. In Jan, Jensen Huang Calls for US AI Manufacturing Revival and said the United States must rebuild its capacity to produce advanced technology, arguing that prosperity should be created for a broad base of workers rather than just a highly‑educated elite and that domestic fabs and supply chains are essential for both economic resilience and national security. He has linked that call directly to the AI boom, warning that a country that outsources its chipmaking will struggle to capture the full value of the AI era.
He has also framed offshoring as a strategic mistake that needs to be corrected. In another Jan interview, he said “We’ve Done Our Country a Great Disservice” by Offshoring and argued that Nvidia, Jensen Huang Says the United States has to create prosperity for its middle class again by investing in advanced manufacturing, training and infrastructure, a critique that aligns with his push for onshore AI fabs and packaging plants and that he summarized in a pointed warning about offshoring.
The five‑layer “AI trade” and the software rebuild
Underneath the policy talk, Huang is trying to give investors and developers a mental model for how this rebuilt AI stack fits together. Nvidia, Corporation, CEO Jensen Huang has described the AI trade as a five‑layer cake, with chips, systems, cloud services, model and applications as the layers, and he has argued that Nvidia is positioned across that entire structure, from silicon to software. By presenting AI in this layered way, he is underscoring his claim that the company is not just a component vendor but a platform provider that can shape how each layer evolves.
That software story is increasingly about making AI systems more autonomous and integrated. The New Era, From Static Tools to Dynamic Systems One of the key concepts Huang has embraced is Agentic AI, where systems can choose the right action at the right time instead of waiting passively for user prompts, and he has pointed to orchestration frameworks, tool integration and feedback loops as the glue that turns large models into reliable agents, a shift that aligns with research on dynamic agentic systems and that demands new software abstractions on top of Nvidia’s hardware.
Autonomous driving and the consumer face of rebuilt AI
For most people, the rebuilt AI stack will show up first in cars, phones and everyday services rather than data centers. At CES in LAS VEGAS, Jan coverage noted that Sure, Nvidia, AMD and Intel all had important chip and AI platform announcements, but it was Nvidia’s automotive demos that made the stakes tangible, including hardware and software specifically designed for autonomous driving that can interpret sensor data, plan routes and control vehicles without human intervention. Those systems rely on the same Rubin‑era accelerators and agentic software that Huang talks about on stage, but they package it into something consumers can see and feel.
Huang has also used social platforms to hammer home how different the next wave of AI will feel. In one clip, NVIDIA, CEO, Jensen Huang says the next wave of AI is physical and emphasizes that Today’s models do not understand basic physics, They can generate convincing text and images but still struggle with the constraints of the real world, a gap that autonomous driving stacks, industrial robots and mixed reality devices will have to close, a point he made in a widely shared statement about physical AI.
“Beginning of the AI revolution” and the risks of uneven returns
Huang’s optimism about AI’s potential is tempered by a warning that the benefits will not be evenly distributed. In Sep, Nvidia CEO Jensen Huang said the world is “at the beginning of the AI revolution” and cautioned that the rapid deployment of AI could lead to uneven returns on investment, with some companies and countries racing ahead while others struggle to capture value. That framing reinforces his argument that rebuilding AI from the ground up is not just about faster chips but about designing systems, policies and business models that spread gains more broadly.
He has also tried to position Nvidia as a guide through that uneven landscape. In another Jan appearance, Nvidia CEO Jensen Huang looked ahead to the next generation of AI and described how Jan, Nvidia CEO Jensen Huang, Spe expects demand from across the world to keep growing, but he also acknowledged that there are “complete unknowns” in how AI will reshape labor markets and geopolitics, a note of caution that surfaced in a second segment of his Jan remarks and that hints at the social engineering challenge that comes with rebuilding such a powerful technology.
Why Huang’s “ground up” vision matters now
Huang’s insistence that AI is being rebuilt from the ground up is not just branding, it is a way of forcing investors, policymakers and engineers to think in systems rather than silos. When NVIDIA CEO Envisions AI Infrastructure Industry Worth Trillions of Dollars and talks about Physical AI agents that can perceive, understand and plan, he is arguing that the next decade of AI will be defined by how well societies integrate chips, data centers, software, robots and regulations into coherent architectures, a view he laid out in detail in his COMPUTEX data center pitch.
That is also why Nvidia is racing to ship Rubin hardware, expand its software stack and lobby for manufacturing revival at the same time. In LAS VEGAS, Jan, GLOBE, NEWSWIRE coverage of CES noted that NVIDIA used the show to kickstart the next generation of AI with Rubin, a six new chips, one incredible AI supercomp launch designed to accelerate mainstream AI adoption, and that combination of technical ambition and ecosystem building is what gives Huang’s ground up narrative its weight, as he showcased in the Rubin announcement.
More from Morning Overview