Can Physical AI Finally Justify MEC and AI-RAN?

The telecoms industry has spent years searching for a use case that genuinely requires edge computing, ultra-low latency and a more intelligent RAN. Multi-access Edge Computing (MEC) has long been presented as a foundational element of future network architecture, yet many early deployments struggled to move beyond limited trials and niche applications. AI-RAN has followed a similar path. While the concept has generated significant industry interest, many operators have continued to question where the real commercial and technical value lies.

Recent demonstrations by SoftBank and Ericsson around network-enabled Physical AI suggest the industry may finally have found a use case that naturally brings these technologies together.

Unlike earlier MEC demonstrations focused on content delivery, video optimisation or traffic offload, Physical AI introduces a closed-loop interaction between robots, AI inference and the network itself. Rather than treating AI as an application that operates independently on top of the network, SoftBank is positioning AI as something that must be tightly integrated with connectivity, compute and radio infrastructure. This has important implications for MEC, distributed compute and the future evolution of telecoms infrastructure.

The concept behind Physical AI is relatively straightforward. Autonomous robots, drones and industrial systems need to perceive their environment, process information and make decisions in real time. Some AI inference can run locally on the device, but onboard compute is often constrained by size, thermal limits, cost and power consumption. More advanced AI models may require significantly more processing capability than can realistically be embedded into every endpoint.

SoftBank’s approach is to treat MEC as an extension of the robot’s compute environment. The network effectively becomes part of the control loop rather than simply acting as a transport layer between devices and the cloud.

This model requires a much tighter relationship between the RAN and application workloads than traditional mobile networks were originally designed to support. In conventional architectures, the RAN largely operates independently of application behaviour. SoftBank’s AI-RAN architecture, developed in collaboration with Ericsson, instead monitors radio conditions, compute availability and application requirements simultaneously. AI workloads can then move dynamically between the device and MEC infrastructure depending on latency, congestion and processing demands.

Ericsson’s role is particularly important because it demonstrates how differentiated connectivity becomes part of the AI execution framework itself. Network slicing, priority handling and RAN automation are integrated into the orchestration process so that the network can continuously adapt to changing AI workload requirements rather than relying on static QoS policies.

A robot performing real-time perception and navigation may initially process workloads locally. If additional compute power is required, inference tasks can be offloaded to nearby MEC resources. If network conditions deteriorate or latency increases, workloads may shift back onto the device. The network continuously orchestrates these decisions while attempting to maintain predictable, deterministic performance levels.

This is significantly different from the fixed slicing models often discussed during the early 5G era. Instead of static policies, the network dynamically adjusts connectivity and workload placement according to application behaviour and infrastructure conditions.

SoftBank’s broader strategy reinforces this direction through its Telco AI Cloud architecture, which combines GPU infrastructure, MEC platforms and RAN intelligence under a unified orchestration framework. The objective is to support AI workloads that span cloud, edge and endpoint devices across sectors such as industrial automation, logistics and construction.

A critical milestone in this development is the interworking between SoftBank’s AITRAS orchestrator and the Ericsson Intelligent Automation Platform (EIAP). This collaboration reflects a future where AI orchestration platforms and RAN automation systems operate through tightly integrated control and policy frameworks. By integrating these layers, the orchestrator can make informed decisions about where to place an AI workload based on a real-time view of the RAN’s capacity and health.

This shift highlights why many previous MEC deployments struggled to achieve large-scale commercial success. Early edge computing use cases rarely demanded strict latency guarantees or continuous workload mobility between device and edge. In many scenarios, centralised cloud platforms remained entirely sufficient.

Physical AI changes that. Real-time robotics and autonomous systems require predictable latency and continuous responsiveness. They also require the ability to distribute AI workloads dynamically across multiple compute domains without interrupting operations. Achieving this level of coordination demands a far deeper integration between MEC, AI orchestration and the RAN than most operators currently deploy.

If the network is expected to participate directly in AI execution and workload placement decisions, the RAN itself becomes part of the distributed AI infrastructure platform.

The implications for telecoms infrastructure are substantial. Operators aiming to support Physical AI applications may need to rethink network architecture around distributed compute rather than centralised capacity. MEC deployments would require tighter integration with radio infrastructure, orchestration systems and AI frameworks.

The industry has discussed this convergence for years through initiatives such as the AI-RAN Alliance and broader work around AI-native networking. What makes the SoftBank and Ericsson work particularly interesting is that it connects these ideas into a practical end-to-end implementation focused on real-world autonomous systems.

Physical AI may finally provide the commercial and technical justification the industry has been searching for to move MEC and AI RAN from research projects and isolated trials into mainstream network architecture. It hints at a future where operators are no longer simply connectivity providers, but providers of distributed AI infrastructure platforms capable of supporting real-time autonomous systems.

Related Posts: 

• Telecoms Infrastructure Blog: Understanding the Internet’s Hidden Infrastructure
• Telecoms Infrastructure Blog: SK Telecom's Vision for Future Telco Infrastructure in the AI Era
• Telecoms Infrastructure Blog: 'Story of the Edge' by Akamai
• The 3G4G Blog: Edge Computing Tutorial from Transforma Insights
• Telecoms Infrastructure Blog: AWS for Public and Private 5G Networks
• Telecoms Infrastructure Blog: SK Telecom’s 5G MEC Status and Plan
• Operator Watch Blog: How is KT Reducing the Latency of their 5G Network 

SK Telecom Builds AI Infrastructure Momentum with GPUaaS and the Haein Cluster

The growing demand for artificial intelligence computing is reshaping the role of telecommunications operators. As AI models become larger and more computationally intensive, the need for high performance infrastructure has moved into sharp focus. In response, SK Telecom is positioning itself not only as a connectivity provider but also as a key supplier of AI infrastructure through its GPU-as-a-Service offering.

At the centre of this strategy is the Haein GPU cluster, one of the largest AI computing platforms in South Korea. Built around more than 1,000 GPUs from NVIDIA based on the Blackwell architecture, the platform provides the computing power required for large scale AI training and inference workloads. The cluster represents a significant step forward from the earlier infrastructure based on NVIDIA H100 GPUs and forms part of SK Telecom’s wider sovereign AI infrastructure initiative.

The Haein cluster is hosted within the company’s Gasan AI Data Center in Seoul and is designed to deliver high performance computing capacity at national scale. The system supports intensive AI workloads including the training of large language models while also providing the flexibility required for enterprises and research organisations developing their own AI applications. The platform’s architecture allows large GPU resources to be combined into a single cluster while still being dynamically allocated to different users depending on demand.

A key component enabling this flexibility is SK Telecom’s proprietary virtualisation platform known as Petasus AI Cloud. The software layer allows the large GPU cluster to be partitioned and reconfigured dynamically, enabling customers to access the exact amount of computing power they require. This capability is essential for GPU-as-a-Service platforms where workloads can vary significantly, from small development environments to large scale model training that requires hundreds of GPUs operating simultaneously.

Alongside this, the company provides operational management through its AI Cloud Manager platform. This AIOps based environment supports the full lifecycle of AI services including development, training, deployment and operational monitoring. By combining infrastructure with operational tooling, SK Telecom aims to provide a more integrated AI computing platform rather than simply raw GPU capacity.

The Haein cluster also plays an important role in South Korea’s national AI strategy. The platform has been selected to support a programme led by the Ministry of Science and ICT that focuses on strengthening the country’s AI computing infrastructure and enabling the development of competitive national AI foundation models. Through this initiative, the cluster will contribute computing resources to projects developing sovereign AI capabilities tailored to the Korean language and domestic industries.

The name of the cluster itself reflects this national perspective. Haein takes inspiration from Haeinsa Temple, which houses the historic Tripitaka Koreana, a vast collection of Buddhist scriptures recognised as a UNESCO World Heritage archive. The naming reflects the ambition to create a modern repository of digital intelligence, supporting the development of AI knowledge and capabilities within the country.

Delivering infrastructure at this scale requires a broad ecosystem of partners. SK Telecom has worked with companies including Supermicro and Penguin Solutions to design and deploy the server infrastructure and integrated AI data centre solutions required for the cluster. These collaborations enable the rapid deployment of high density GPU servers and the supporting cooling, power and networking systems necessary to run large scale AI workloads.

The industry has already taken notice of the platform. The Haein GPU cluster was recognised at the MWC Barcelona 2026, where SK Telecom received the Best Cloud Solution award at the GSMA Global Mobile Awards. The recognition reflects the company’s continued progress in cloud and AI infrastructure development and marks the third consecutive year that its cloud related technologies have been acknowledged in this category.

For telecoms infrastructure professionals, SK Telecom’s GPU-as-a-Service strategy illustrates how operators are expanding beyond traditional connectivity services. By building large scale AI computing platforms inside their data centre footprint, operators can leverage existing strengths in infrastructure, power management and network integration to participate in the rapidly growing AI economy.

As AI adoption accelerates across industries, the demand for scalable computing infrastructure will continue to grow. With platforms such as the Haein cluster and its GPUaaS offering, SK Telecom is positioning its network and data centre assets as part of the core infrastructure supporting the next generation of AI innovation.

Related Posts: 

• Telecoms Infrastructure Blog: SK Telecom's Vision for Future Telco Infrastructure in the AI Era
• Operator Watch Blog: SK Telecom on 'Applying AI to Transform Customer Experience'
• Free 6G Training - SK Telecom 6G White Paper: View on Future AI Telco Infrastructure
• Free 6G Training: SK Telecom Releases 6G White Paper Looking at Lessons Learned in 5G and Way Forward with 6G
• Free 6G Training: SK Telecom Claims ‘Bare Metal-Based Cloud-Native 5G Core’ as Foundation for 6G Evolution
• Telecoms Infrastructure Blog: SK Telecom’s 5G MEC Status and Plan 

Samsung Brings AI Across Every Layer of the Network to Power Next Generation Telecom Infrastructure

Artificial intelligence is rapidly becoming a defining capability in modern telecom networks. As operators continue to expand 5G and prepare for the transition to 6G, the scale and complexity of networks are increasing significantly. In this environment, automation, efficiency and adaptability are becoming essential. Samsung is positioning artificial intelligence as a core technology that can operate across every layer of the network to help operators manage this complexity while unlocking new capabilities.

Much of the industry conversation around AI integration in telecom networks has recently focused on the concept of AI-RAN. Within the AI-RAN Alliance, this is commonly described through three dimensions: AI for RAN, AI on RAN and AI with RAN. These categories describe how artificial intelligence can enhance radio network performance, support new edge-based services and enable the coexistence of AI workloads and network functions on shared infrastructure.

Samsung is actively involved in this industry effort, but its strategy extends beyond the radio layer alone. The company is promoting a broader approach to AI-powered networks that combines end-to-end software-based architecture with distributed computing capabilities. In this model, artificial intelligence is not limited to a specific part of the network. Instead, it is embedded across the entire infrastructure, from the radio access network to the core network and operational management systems.

A key element of this approach is Samsung’s focus on software-based and virtualised network architectures. Virtualised RAN deployments running on commercial off-the-shelf servers provide a flexible platform where both network workloads and AI functions can operate together. This allows operators to introduce AI capabilities without needing to completely redesign their infrastructure.

Through its network automation platform, Samsung is applying AI to a wide range of operational tasks. These include predicting traffic patterns, identifying anomalies in network performance, optimising radio parameters and balancing loads across spectrum bands. By analysing large volumes of operational data, AI systems can automatically adjust network behaviour to maintain performance and improve efficiency.

Energy optimisation is another area where AI-driven techniques are being applied. As mobile networks expand and traffic patterns fluctuate throughout the day, intelligent algorithms can determine when certain network features can be adjusted or scaled down to reduce power consumption without affecting user experience. These types of capabilities are becoming increasingly important as operators focus on both operational efficiency and sustainability.

Samsung is also exploring how artificial intelligence can improve radio performance directly within the protocol stack. Machine learning techniques can enhance channel estimation at the physical layer, allowing the network to reconstruct radio signals more accurately even in challenging environments. At higher layers, AI can support link adaptation by identifying optimal modulation and coding schemes for each user based on real time radio conditions. Even connection management processes can benefit from AI driven optimisation, improving both device battery efficiency and network resource utilisation.

Beyond improving the network itself, Samsung is also examining how telecom infrastructure can support AI workloads. Modern base stations and edge compute platforms contain significant computing resources. When network traffic demand is low, some of this capacity can remain unused. By running AI inference tasks on the same infrastructure, operators can make better use of these resources while supporting new services.

Edge based AI applications are particularly relevant in industrial environments. Real time video analytics, safety monitoring and automated quality inspection are examples of workloads that benefit from processing close to the data source. Running these applications on infrastructure that already supports radio functions reduces latency and avoids sending large volumes of data to central cloud platforms.

Samsung describes this convergence between communications infrastructure and computing capabilities as a shift towards networks functioning as distributed data centres. In this model, the network becomes both a connectivity platform and a processing environment capable of supporting AI driven applications. The concept combines two complementary perspectives: building networks that support AI workloads and using AI to improve how networks operate.

This architectural shift also has implications for the hardware layer of telecom infrastructure. Traditional mobile network equipment has relied heavily on specialised system-on-chip designs. However, the rapid development cycle of general purpose processors and accelerators is encouraging a more flexible approach. Samsung’s virtualised infrastructure strategy allows operators to deploy workloads on a mix of CPUs and GPUs, drawing on technologies from companies such as Intel, NVIDIA and Arm Ltd.. This enables operators to scale AI capabilities across different parts of the network depending on where computing power is needed.

As telecom networks evolve towards cloud native and software driven architectures, the role of artificial intelligence will continue to expand. By embedding AI across radio, core and operational layers, Samsung is highlighting how networks can move beyond traditional connectivity and become intelligent platforms capable of continuous optimisation.

With 5G Advanced deployments underway and early discussions around 6G gathering momentum, the integration of AI into telecom infrastructure is likely to accelerate. Samsung’s strategy suggests that the future network will not simply transport data, but will increasingly analyse, optimise and process it within the network itself, transforming the way operators design and operate their infrastructure.

Related Posts: 

• Private Networks Technology Blog: Samsung’s Private 5G Networks Supporting Digital Transformation Across Industries
• Private Networks Technology Blog: Hyundai and Samsung Bring RedCap Private 5G to Smart Manufacturing
• Private Networks Technology Blog: Private Networks and Network-In-a-Box (NIB) Solutions from MWC 2025
• Private Networks Technology Blog: Enterprise Services with Private 5G - Key Enabler in Expanding Opportunities for Vertical Industries
• Private Networks Technology Blog: Samsung and NAVER Cloud launch Korea’s first private 5G network in the construction sector
• Free 6G Training: Samsung’s Research Vision for Energy Saving in 6G Networks
• Free 6G Training: Samsung’s Vision for 6G and the Rise of Native AI in Wireless Systems
• Telecoms Infrastructure Blog: How Samsung is Leveraging vRAN to Match Traditional RAN (T-RAN) Performance
• Telecoms Infrastructure Blog: Samsung's 3GPP-Compliant PS-LTE Network
• Connectivity Technology Blog: Samsung to showcase benefits of 5G in Industry 4.0 with IBM, IMDA & M1 in Singapore 

Meta's Project Waterworth: The Next Evolution in Subsea Connectivity

Meta has unveiled its most ambitious subsea cable project to date — Project Waterworth, previously referred to as "W", because of it's shape. The multi-billion-dollar initiative is set to become the longest subsea cable in the world, spanning over 50,000 km and connecting five major continents, including the U.S., India, Brazil, and South Africa. With 24 fibre pairs delivering the highest capacity technology available, Project Waterworth will redefine global digital infrastructure and enhance connectivity for billions of users.

Subsea cables form the backbone of the internet, carrying more than 95% of intercontinental traffic and enabling global communication, financial transactions, and AI-driven innovations. With this latest venture, Meta aims to open three new oceanic corridors, ensuring high-speed, reliable connectivity that will power the next wave of AI advancements worldwide. By leveraging cutting-edge routing techniques, enhanced burial methods in high-risk areas, and deep-sea deployments up to 7,000 metres, Project Waterworth is designed for maximum resilience and security.

India at the Centre of Meta’s Connectivity Vision

India is central to Meta’s strategy, with its platforms—Facebook, Instagram, and WhatsApp—serving over a billion users in the country. With AI adoption accelerating, demand for data centre capacity and seamless connectivity is at an all-time high. Project Waterworth is expected to play a pivotal role in supporting India’s digital economy by providing the necessary infrastructure to handle AI workloads, cloud services, and high-speed internet demands.

The project also underscores Meta’s shift in subsea cable strategy. Unlike its earlier 2Africa initiative, which followed a consortium approach, Project Waterworth appears to be a fully owned and controlled system. This mirrors Google's model of securing dedicated infrastructure for strategic markets rather than relying on shared capacity. While this approach ensures end-to-end control and security, it diverges from the collaborative model that has been highly successful in previous large-scale subsea cable projects.

Bypassing Global Chokepoints

One of the key aspects of Project Waterworth is its avoidance of politically sensitive and high-risk regions. Meta has reportedly designed the cable to steer clear of the Red Sea, the South China Sea, Egypt, and the Malacca Strait—areas that have become significant geopolitical bottlenecks for global internet traffic. By taking a direct route between the U.S. and India with strategic stops in South Africa and potentially Australia, Project Waterworth aims to ensure long-term security and avoid the risks associated with conflict zones and regulatory challenges in transit countries.

However, this bypassing of traditional routes does come with a trade-off: increased latency. Despite this, Meta appears to prioritise long-term security and reliability over marginal improvements in data transmission speeds. The project will also likely face regulatory hurdles, particularly in India, where obtaining permits for marine surveys and installations is notoriously complex and time-consuming.

The Battle for AI Connectivity Dominance

Meta’s decision to fully own Project Waterworth could have wider implications for the subsea cable industry. If Meta excludes partners, it may push competitors like Google to develop their own dedicated infrastructure to serve India’s growing digital ecosystem. Given the scale of investment—potentially exceeding $10 billion over the next decade—this move signals a new era of tech giants building independent, AI-optimised connectivity solutions.

While Project Waterworth marks a significant leap forward in global connectivity, the challenge will be balancing rapid deployment with regulatory constraints. If successful, it will not only strengthen Meta’s position as a digital infrastructure leader but also cement India’s role as a global AI powerhouse in the decades to come.

Related Posts:

• Telecoms Infrastructure Blog - Data Centers At Meta: Heterogeneous Integration Driven By AI/ML And Network Applications
• Telecoms Infrastructure Blog: The Role of Border Gateway Protocol (BGP) in Making Sure Internet Works
• Telecoms Infrastructure Blog: Facebook's Fiber Deploying Robot
• The 3G4G Blog: AI and Analytics Based Network Designing & Planning 

SK Telecom's Vision for Future Telco Infrastructure in the AI Era

Last month, SK Telecom released a 6G white paper that explores the evolution of wireless and wired infrastructure through the convergence of AI and telecommunications. The white paper highlights how Telco Edge AI infrastructure can redefine the value of network systems by enabling real-time data processing alongside AI-driven services. You can find my detailed blog post on the white paper here.

Dr. Takki Yu, Vice President of the Infra Tech Office at SK Telecom, leads R&D efforts across end-to-end mobile communication technologies. His work spans Radio Access, Core, Transport, Devices, Location, and Network AI. Dr. Yu’s primary focus is on advancing mobile communications, including 5G and Beyond 5G (6G) systems, as well as innovations in network virtualization, cloud, location-based quantum security, and AI integration. Notably, he played a key role in the successful commercialization of the world’s first 5G network in Korea and continues to lead the charge in developing Beyond 5G and 6G technologies.

At the Brooklyn 6G Summit (B6GS), Dr. Takki Yu delivered a keynote presentation titled "The Path to AI Telecommunications Infrastructure Evolution as Future Architecture." In his talk, he shared SK Telecom's vision for the future of telco infrastructure, reflecting on the expectations of the 6G era and the transformative shift towards AI-driven telecommunications infrastructure.

Related Posts: 

• Free 6G Training - SK Telecom 6G White Paper: View on Future AI Telco Infrastructure
• Telecoms Infrastructure Blog: SK Telecom’s 5G MEC Status and Plan
• Operator Watch Blog: SK Telecom on 'Applying AI to Transform Customer Experience'
• Free 6G Training: A Discussion on SK Telecom's First 6G Whitepaper from Wireless Future Podcast
• Free 6G Training: SK Telecom Releases 6G White Paper Looking at Lessons Learned in 5G and Way Forward with 6G
• Free 6G Training: SK Telecom Claims ‘Bare Metal-Based Cloud-Native 5G Core’ as Foundation for 6G Evolution 

Data Centers At Meta: Heterogeneous Integration Driven By AI/ML And Network Applications

Last year, tech giants including Intel, Meta, Arm, Google Cloud, AMD, Qualcomm, TSMC and ASE formed the Chiplet consortium. A news article in Fierce Electronics said:

Several giant tech companies have joined hands to promote an open standard for chip components called chiplets and how they are crammed together in system-on-chip (SoC) designs deemed critical to a variety of future handheld and high-performance computers that power AI applications and much more.

The open standard, called Universal Chiplet Interconnect Express (UCIe), has been developed by Intel and clearly benefits Intel’s integrated device manufacturer (IDM) strategy as it builds new chip fabs in Arizona and Ohio and elsewhere outside the U.S.  Intel has been a prominent voice in the push to expand chip manufacturing outside of Taiwan and the rest of Asia where it is heavily focused today.

Intel has donated its UCIe standard to founding members in a new consortium that includes Intel along with Advanced Semiconductor Engineering, Taiwan Semiconductor Manufacturing Co., AMD, Arm, Google Cloud, Meta, Microsoft, Samsung and Qualcomm. The founders have already ratified UCIe 1.0 which covers the die-to-die physical layer, die-to-die protocols and software stacks which leverage the existing PCI Express (PCIe) and Compute Express Link (CXL) industry standards.

Ravi Agarwal, a technical sourcing manager at the Facebook/Meta Infrastructure group is responsible for driving advanced packaging architectures and foundry for both networking and AI/ML compute applications to meet Facebook’s future workloads. He is driving Chiplet Business Workstream in Open Domain-Specific Architecture (ODSA) Sub-Project within the Open Compute Project (OCP), working with ecosystem partners to enable a Chiplet marketplace. 

In a talk delivered for the IEEE Electronics Packaging Society (EPS) SFBA, he focused on heterogeneous integration for Artificial Intelligence, Machine Learning and network applications at Meta Infrastructure, and discussed implications for packaging and system-level considerations. In the talk he also shared some of the advanced packaging (chiplet) initiatives in which Meta is participating to develop an open ecosystem.

The talk is embedded below:

While the slides of this talk is not available, you can see slides of another talk he delivered here.

Related Posts: 

• Telecoms Infrastructure Blog: The Role of Border Gateway Protocol (BGP) in Making Sure Internet Works
• Telecoms Infrastructure Blog: Facebook's Fiber Deploying Robot
• Telecoms Infrastructure Blog: 'Story of the Edge' by Akamai