Focusing on the "Physical AI Era"! NVIDIA (NVDA.US) aims to reshape the communication ecosystem and collaborate with telecom giants to build AI-native 6G

Bloomberg News has learned that NVIDIA (NVDA.US), the world’s most valuable publicly traded company known as the “AI chip superpower,” is quietly working on a major initiative as a new round of geopolitical conflicts in the Middle East erupts and military strikes expand beyond Iran and Israel. The chip giant is fully supporting a significant effort to ensure that the upcoming 6G mobile intelligent network provides the most powerful platform for services, electronic devices, and a wide range of “physical AI” devices that can utilize cutting-edge artificial intelligence technology. Deep integration of 6G networks and AI is now a consensus across the industry. In the imminent “super era of physical AI,” NVIDIA’s AI+6G fusion technology is undoubtedly one of the most critical technological pillars of this era.

Undoubtedly, NVIDIA will play a central role in shaping the standards and architecture of 6G, collaborating with telecom giants like Nokia to build AI-native 6G platforms that emphasize AI capabilities integrated into wireless infrastructure. AI-native and software-defined networks will become hallmark features of 6G technology. This is not just a performance upgrade but a fundamental transformation of future communication systems. The “AI+6G” pathway is seen as a key underlying support for the future physical AI era, providing foundational network capabilities for new applications such as intelligent terminals, humanoid robots, and autonomous driving.

It is understood that NVIDIA is working in deep collaboration with a group of international telecom giants including Nokia (Nokia Oyj), SoftBank Group, and T-Mobile US Inc., aiming to build a sixth-generation network architecture based on new generations of computers and software. These technologies will be capable of using artificial intelligence to safely and efficiently direct radio traffic.

In a statement on Sunday, NVIDIA said, “This change is urgent and necessary because a large number of intelligent devices will connect to 6G networks in the future, and the demand for these high-performance devices is becoming more complex.” The statement was released just before the opening of the telecom industry conference in Barcelona. NVIDIA stated that current 5G networks are designed to connect people for voice and data retrieval but cannot support widespread use of AI micro-training/inference systems.

Ronnie Vasishta, NVIDIA’s head of telecom business and strategy, said: “Today’s networks are fundamentally unable to meet the needs of tomorrow’s use cases. Entering the AI era, everything will change. Network infrastructure will not only make smartphones smarter but will also provide intelligent services for all machines.” He added that telecom networks will need to achieve “thousands of times” efficiency improvements because there is not enough radio spectrum to support new uses.

The wave of “AI+6G” is coming, and NVIDIA is determined to be a leader

This chip manufacturer’s core infrastructure for AI training and inference—AI chips—is at the heart of the explosive growth of artificial intelligence. It is working to carve out a new market and clear potential barriers.

At the GTC conference in Washington in late October 2025, Jensen Huang, known as the “Godfather of AI,” announced a $1 billion equity investment in Nokia. The two sides will jointly develop the AI-RAN product line and AI-native 6G network platform; NVIDIA also released the Aerial RAN Computer (ARC/ARC Pro) computing platform, aiming to make “AI on RAN” a new computational layer for communications infrastructure. NVIDIA plans to build a new cloud computing platform on top of 6G, demonstrating the enormous potential of ultra-fast AI, which will significantly drive technologies like robotic vision intelligence and autonomous driving.

In the previous wireless era—5G—the absence of NVIDIA’s presence was mainly because 5G architecture did not require large-scale AI integration at the foundational level, as it was primarily led by traditional telecom vendors. The core goals of 5G were to increase bandwidth, reduce latency, and enhance connection density, mainly for voice, data, and video services. During standardization and deployment, traditional telecom equipment providers like Nokia, Ericsson, Huawei, etc., optimized networks based on conventional wireless PHY/MAC protocols. AI applications in 5G were mainly limited to edge enhancement and non-critical services, so platforms with large-scale AI computing capabilities like NVIDIA’s were not deeply involved in core network architecture.

In contrast, 6G planning from the outset is an “AI-native network,” which demands new computing capabilities. Unlike 5G, which focuses on ultra-high speed and ultra-low latency, 6G also requires the network itself to have intelligent perception, real-time scheduling, dynamic spectrum management, and automatic optimization capabilities. In the 6G architecture, AI must be deeply embedded from the radio access network (RAN) to the core network, achieving comprehensive intelligent optimization from terminals to edge and core. This architecture is not just about adding AI applications externally but involves deep integration of AI with communication protocols, requiring massive training and inference capabilities to handle complex resource scheduling, channel prediction, interference management, and more in real time. Standard-setting bodies and industry alliances emphasize AI as a core design element.

To enable AI-native capabilities, networks need powerful computing platforms to support large-scale AI inference and training, along with coordinated computing processes between the control plane and user plane. This is precisely NVIDIA’s strength: its GPUs and AI computing platforms can perform massive parallel computations and unify network and AI loads through “software-defined architecture” (Software‑Defined AI‑RAN), improving spectral efficiency and adapting to complex, dynamic wireless environments.

Traditional telecom hardware clusters (specialized ASICs, DSPs, etc.) are limited in large-scale AI computation. This is why NVIDIA and operators are working together to build open, AI-native 6G platforms. Industry collaborations and alliances clearly show that NVIDIA is involved with over 130 industry partners to promote AI-RAN innovation, positioning itself as a key technological hub in the AI-6G ecosystem.

NVIDIA has already provided the latest versions of core chips, computing components, and related software for high-performance network infrastructure and aims to expand this business significantly. Recently, the chip giant has been pushing cutting-edge AI technology into broader fields—such as robotics and autonomous vehicles, collectively called “physical AI”—to continue expanding demand and seek new growth points outside data center applications. Without wireless networks capable of supporting AI-level massive traffic, NVIDIA’s vision of a world filled with humanoid robots and autonomous vehicles in the “physical AI” real world could be delayed.

Approximately every decade, the telecom industry shifts to a new wireless technology generation, the next “G.” During the process of defining new hardware and software standards, telecom companies often lead the industry through alliances aligned with their product lines. However, this approach has inconsistent records and often results in deployment delays or incompatible networks due to competitive efforts.

NVIDIA believes that new devices and software must be fundamentally open. Instead of using closed, custom hardware devices, radio transmission and reception equipment should be controlled by updatable software and run on more general-purpose computing systems. Data traffic should be guided by AI software and increasingly large AI infrastructure capable of responding to rapidly changing patterns and priorities, which is currently unachievable, NVIDIA states.

In such an environment, the telecom industry will be more open to new vendors, including startups that could quickly reach a valuation of billions of dollars, according to Vasishta. He said, “This will be how a new telecom unicorn is born.” He added that over the past decade, few new companies have entered the industry.

Physical AI era, “AI+6G” is indispensable

According to NVIDIA CEO Jensen Huang, “Physical AI” emphasizes enabling robots/autonomous systems to perceive, reason, and act in the real world, and an era where “physical AI” assists human civilization is imminent. “Physical AI” focuses on enabling robots/autonomous systems to perceive, reason, and act in the real world, and these three capabilities are key tools to advance models from “just conversation” to “physical-world work.”

The integration of AI and 6G is not just an application overlay but a fundamental transformation of network architecture. Future 6G networks will be more than “faster connections”; they will become intelligent engines capable of analyzing terminal, environmental, and user data in real time, automatically tuning network resource allocation, and providing integrated intelligent services across endpoints, edges, and clouds. AI will play roles in spectrum sensing, resource allocation, network slicing optimization, and edge inference, most notably supporting large-scale AI workloads such as IoT, autonomous driving, robotics, and smart spaces. These use cases, which are more like add-ons in 5G, will become essential functions of the network itself in 6G. Some market opinions even suggest that without AI, true 6G cannot be realized.

A key feature of 6G planning is the AI-native network, which embeds AI capabilities into the fundamental communication architecture from RAN to core, enabling self-optimization, intelligent spectrum scheduling, and real-time response to billions of terminals’ complex demands. The vision of 6G is to achieve “smart connectivity of everything,” deeply integrating AI and communications, making the network a perceptive, predictive, and self-optimizing system. This evolution means the network will not only transmit data but also perform massive AI inference and real-time learning workloads—such as intelligent spectrum management, edge inference, and real-time signal optimization—requiring robust AI computing infrastructure and flexible software-defined platforms.

NVIDIA’s AI platforms (such as NVIDIA Aerial and AI‑RAN architecture) provide the high-performance, programmable, AI-accelerated infrastructure needed to build software-defined, open, and AI-native wireless networks, fulfilling the core requirements of future high-speed, intelligent, scalable communications.

In the so-called “physical AI era,” a vast number of terminals like intelligent robots, autonomous vehicles, and smart industrial devices will seamlessly connect and generate enormous AI workloads. This not only requires efficient connectivity but also edge AI inference, distributed machine learning, and network-aware intelligence.

To support these cutting-edge functions, networks need to have general AI computing capabilities and be deeply integrated with wireless protocol stacks. NVIDIA’s leadership in AI chips, accelerated computing architectures, and development tools like Aerial CUDA Accelerated RAN, Omniverse digital twin simulation platform, and AI radio frameworks provide the core foundation for building scalable, open, AI-native wireless networks. Therefore, although the future physical AI ecosystem will involve multiple participants, NVIDIA’s AI+6G fusion technology is undoubtedly one of the most critical technological pillars of this era.