AI
SoftBank’s Transformer AI model boosts 5G AI-RAN uplink throughput by 30%, compared to a baseline model without AI
Softbank has developed its own Transformer-based AI model that can be used for wireless signal processing. SoftBank used its Transformer model to improve uplink channel interpolation which is a signal processing technique where the network essentially makes an educated guess as to the characteristics and current state of a signal’s channel. Enabling this type of intelligence in a network contributes to faster, more stable communication, according to SoftBank. The Japanese wireless network operator successfully increased uplink throughput by approximately 20% compared to a conventional signal processing method (the baseline method). In the latest demonstration, the new Transformer-based architecture was run on GPUs and tested in a live Over-the-Air (OTA) wireless environment. In addition to confirming real-time operation, the results showed further throughput gains and achieved ultra-low latency.
Editor’s note: A Transformer model is a type of neural network architecture that emerged in 2017. It excels at interpreting streams of sequential data associated with large language models (LLMs). Transformer models have also achieved elite performance in other fields of artificial intelligence (AI), including computer vision, speech recognition and time series forecasting. Transformer models are lightweight, efficient, and versatile – capable of natural language processing (NLP), image recognition and wireless signal processing as per this Softbank demo.
Significant throughput improvement:
- Uplink channel interpolation using the new architecture improved uplink throughput by approximately 8% compared to the conventional CNN model. Compared to the baseline method without AI, this represents an approximately 30% increase in throughput, proving that the continuous evolution of AI models leads to enhanced communication quality in real-world environments.
Higher AI performance with ultra-low latency:
- While real-time 5G communication requires processing in under 1 millisecond, this demonstration with the Transformer achieved an average processing time of approximately 338 microseconds, an ultra-low latency that is about 26% faster than the convolution neural network (CNN) [1.] based approach. Generally, AI model processing speeds decrease as performance increases. This achievement overcomes the technically difficult challenge of simultaneously achieving higher AI performance and lower latency. Editor’s note: Perhaps this can overcome the performance limitations in ITU-R M.2150 for URRLC in the RAN, which is based on an uncompleted 3GPP Release 16 specification.
Note 1. CNN-based approaches to achieving low latency focus on optimizing model architecture, computation, and hardware to accelerate inference, especially in real-time applications. Rather than relying on a single technique, the best results are often achieved through a combination of methods.
Using the new architecture, SoftBank conducted a simulation of “Sounding Reference Signal (SRS) prediction,” a process required for base stations to assign optimal radio waves (beams) to terminals. Previous research using a simpler Multilayer Perceptron (MLP) AI model for SRS prediction confirmed a maximum downlink throughput improvement of about 13% for a terminal moving at 80 km/h.*2
In the new simulation with the Transformer-based architecture, the downlink throughput for a terminal moving at 80 km/h improved by up to approximately 29%, and by up to approximately 31% for a terminal moving at 40 km/h. This confirms that enhancing the AI model more than doubled the throughput improvement rate (see Figure 1). This is a crucial achievement that will lead to a dramatic improvement in communication speeds, directly impacting the user experience.
The most significant technical challenge for the practical application of “AI for RAN” is to further improve communication quality using high-performance AI models while operating under the real-time processing constraint of less than one millisecond. SoftBank addressed this by developing a lightweight and highly efficient Transformer-based architecture that focuses only on essential processes, achieving both low latency and maximum AI performance. The important features are:
(1) Grasps overall wireless signal correlations
By leveraging the “Self-Attention” mechanism, a key feature of Transformers, the architecture can grasp wide-ranging correlations in wireless signals across frequency and time (e.g., complex signal patterns caused by radio wave reflection and interference). This allows it to maintain high AI performance while remaining lightweight. Convolution focuses on a part of the input, while Self-Attention captures the relationships of the entire input (see Figure 2).
(2) Preserves physical information of wireless signals
While it is common to normalize input data to stabilize learning in AI models, the architecture features a proprietary design that uses the raw amplitude of wireless signals without normalization. This ensures that crucial physical information indicating communication quality is not lost, significantly improving the performance of tasks like channel estimation.
(3) Versatility for various tasks
The architecture has a versatile, unified design. By making only minor changes to its output layer, it can be adapted to handle a variety of different tasks, including channel interpolation/estimation, SRS prediction, and signal demodulation. This reduces the time and cost associated with developing separate AI models for each task.
The demonstration results show that high-performance AI models like Transformer and the GPUs that run them are indispensable for achieving the high communication performance required in the 5G-Advanced and 6G eras. Furthermore, an AI-RAN that controls the RAN on GPUs allows for continuous performance upgrades through software updates as more advanced AI models emerge, even after the hardware has been deployed. This will enable telecommunication carriers to improve the efficiency of their capital expenditures and maximize value.
Moving forward, SoftBank will accelerate the commercialization of the technologies validated in this demonstration. By further improving communication quality and advancing networks with AI-RAN, SoftBank will contribute to innovation in future communication infrastructure. The Japan based conglomerate strongly endorsed AI RAN at MWC 2025.
References:
https://www.softbank.jp/en/corp/news/press/sbkk/2025/20250821_02/
https://www.telecoms.com/5g-6g/softbank-claims-its-ai-ran-tech-boosts-throughput-by-30-
https://www.telecoms.com/ai/softbank-makes-mwc-25-all-about-ai-ran
https://www.ibm.com/think/topics/transformer-model
https://www.itu.int/rec/R-REC-M.2150/en
Softbank developing autonomous AI agents; an AI model that can predict and capture human cognition
Dell’Oro Group: RAN Market Grows Outside of China in 2Q 2025
Dell’Oro: AI RAN to account for 1/3 of RAN market by 2029; AI RAN Alliance membership increases but few telcos have joined
Dell’Oro: RAN revenue growth in 1Q2025; AI RAN is a conundrum
Nvidia AI-RAN survey results; AI inferencing as a reinvention of edge computing?
OpenAI announces new open weight, open source GPT models which Orange will deploy
Deutsche Telekom and Google Cloud partner on “RAN Guardian” AI agent
RtBrick survey: Telco leaders warn AI and streaming traffic to “crack networks” by 2030
Respondents to a RtBrick survey of 200 senior telecom decision makers in the U.S., UK, and Australia finds that network operator leaders are failing to make key decisions and lack the motivation to change. The report exposes urgent warnings from telco engineers that their networks are on a five-year collision course with AI and streaming traffic. It finds that 93% of respondents report a lack of support from leadership to deploy disaggregated network equipment. Key findings:
- Risk-averse leadership and a lack of skills are the top factors that are choking progress.
- Majority are stuck in early planning, while AT&T, Deutsche Telekom, and Comcast lead large-scale disaggregation rollouts.
- Operators anticipate higher broadband prices but fear customer backlash if service quality can’t match the price.
- Organizations require more support from leadership to deploy disaggregation (93%).
- Complexity around operational transformation (42%), such as redesigning architectures and workflows.
- Critical shortage of specialist skills/staff (38%) to manage disaggregated systems.
The survey finds that almost nine in ten operators (87%) expect customers to demand higher broadband speeds by 2030, while roughly the same (79%) state their customers expect costs to increase, suggesting they will pay more for it. Yet half of all leaders (49%) admit they lack complete confidence in delivering services at a viable cost. Eighty-four percent say customer expectations for faster, cheaper broadband are already outpacing their networks, while 81% concede their current architectures are not well-suited to handling the future increases in bandwidth demand, suggesting they may struggle with the next wave of AI and streaming traffic.
“Senior leaders, engineers, and support staff inside operators have made their feelings clear: the bottleneck isn’t capacity, it’s decision-making,” said Pravin S Bhandarkar, CEO and Founder of RtBrick. “Disaggregated networks are no longer an experiment. They’re the foundation for the agility, scalability, and transparency operators need to thrive in an AI-driven, streaming-heavy future,” he added noting the intent to deploy disaggregation as per this figure:
However, execution continues to trail ambition. Only one in twenty leaders has confirmed they’re “in deployment” today, while 49% remain stuck in early-stage “exploration”, and 38% are still “in planning”. Meanwhile, big-name operators such as AT&T, Deutsche Telekom, and Comcast are charging ahead and already actively deploying disaggregation at scale, demonstrating faster rollouts, greater operational control, and true vendor flexibility. Here’s a snapshot of those activities:
- AT&T has deployed an open, disaggregated routing network in their core, powered by DriveNets Network Cloud software on white-box bare metal switches and routers from Taiwanese ODMs, according to Israel based DriveNets. DriveNets utilizes a Distributed Disaggregated Chassis (DDC) architecture, where a cluster of bare metal switches act as a single routing entity. That architecture has enabled AT&T to accelerate 5G and fiber rollouts and improve network scalability and performance. It has made 1.6Tb/s transport a reality on AT&T’s live network.
- Deutsche Telekom has deployed a disaggregated broadband network using routing software from RtBrick running on bare-metal switch hardware to provide high-speed internet connectivity. They’re also actively promoting Open BNG solutions as part of this initiative.
- Comcast uses network cloud software from DriveNets and white-box hardware to disaggregate their core network, aiming to increase efficiency and enable new services through a self-healing and consumable network. This also includes the use of disaggregated, pluggable optics from multiple vendors.
Nearly every leader surveyed also claims their organization is “using” or “planning to use” AI in network operations, including for planning, optimization, and fault resolution. However, nine in ten (93%) say they cannot unlock AI’s full value without richer, real-time network data. This requires more open, modular, software-driven architecture, enabled by network disaggregation.
“Telco leaders see AI as a powerful asset that can enhance network performance,” said Zara Squarey, Research Manager at Vanson Bourne. “However, the data shows that without support from leadership, specialized expertise, and modern architectures that open up real-time data, disaggregation deployments may risk further delays.”
When asked what benefits they expect disaggregation to deliver, operators focused on outcomes that could deliver the following benefits:
- 54% increased operational automation
- 54% enhanced supply chain resilience
- 51% improved energy efficiency
- 48% lower purchase and operational costs
- 33% reduced vendor lock-in
Transformation priorities align with those goals, with automation and agility (57%) ranked first, followed by vendor flexibility (55%), supply chain security (51%), cost efficiency (46%) and energy usage and sustainability (47%).
About the research:
The ‘State of Disaggregation’ research was independently conducted by Vanson Bourne in June 2025 and commissioned by RtBrick to identify the primary drivers and barriers to disaggregated network rollouts. The findings are based on responses from 200 telecom decision makers across the U.S., UK, and Australia, representing operations, engineering, and design/Research and Development at organizations with 100 to 5,000 or more employees.
References:
https://www.rtbrick.com/state-of-disaggregation-report-2
https://drivenets.com/blog/disaggregation-is-driving-the-future-of-atts-ip-transport-today/
Disaggregation of network equipment – advantages and issues to consider
NTT Data and Google Cloud partner to offer industry-specific cloud and AI solutions
NTT Data and Google Cloud plan to combine their expertise in AI and the cloud to offer customized solutions to accelerate enterprise transformation across sectors including banking, insurance, manufacturing, retail, healthcare, life sciences and the public sector.. The partnership will include agentic AI solutions, security, sovereign cloud and developer tools. This collaboration combines NTT DATA’s deep industry expertise in AI, cloud-native modernization and data engineering with Google Cloud’s advanced analytics, AI and cloud technologies to deliver tailored, scalable enterprise solutions.
With a focus on co-innovation, the partnership will drive industry-specific cloud and AI solutions, leveraging NTT DATA’s proven frameworks and best practices along with Google Cloud’s capabilities to deliver customized solutions backed by deep implementation expertise. Significant joint go-to-market investments will support seamless adoption across key markets.
According to Gartner®, worldwide end-user spending on public cloud services is forecast to reach $723 billion in 2025, up from $595.7 billion in 2024.1 The use of AI deployments in IT and business operations is accelerating the reliance on modern cloud infrastructure, highlighting the critical importance of this strategic global partnership.
“This collaboration with Google Cloud represents a significant milestone in our mission to drive innovation and digital transformation across industries,” said Marv Mouchawar, Head of Global Innovation, NTT DATA. “By combining NTT DATA’s deep expertise in AI, cloud-native modernization and enterprise solutions with Google Cloud’s advanced technologies, we are helping businesses accelerate their AI-powered cloud adoption globally and unlock new opportunities for growth.”
“Our partnership with NTT DATA will help enterprises use agentic AI to enhance business processes and solve complex industry challenges,” said Kevin Ichhpurani, President, Global Partner Ecosystem at Google Cloud. “By combining Google Cloud’s AI with NTT DATA’s implementation expertise, we will enable customers to deploy intelligent agents that modernize operations and deliver significant value for their organizations.”
Photo Credit: Phil Harvey/Alamy Stock Photo
In financial services, this collaboration will support regulatory compliance and reporting through NTT DATA solutions like Regla, which leverage Google Cloud’s scalable AI infrastructure. In hospitality, NTT DATA’s Virtual Travel Concierge enhances customer experience and drives sales with 24×7 multilingual support, real-time itinerary planning and intelligent travel recommendations. It uses the capabilities of Google’s Gemini models to drive personalization across more than 3 million monthly conversations.
Key focus areas include:
- Industry-specific agentic AI solutions: NTT DATA will build new industry solutions that transform analytics, decision-making and client experiences using Google Agentspace, Google’s Gemini models, secure data clean rooms and modernized data platforms.
- AI-driven cloud modernization: Accelerating enterprise modernization with Google Distributed Cloud for secure, scalable modernization built and managed on NTT DATA’s global infrastructure, from data centers to edge to cloud.
- Next-generation application and security modernization: Strengthening enterprise agility and resilience through mainframe modernization, DevOps, observability, API management, cybersecurity frameworks and SAP on Google Cloud.
- Sovereign cloud innovation: Delivering secure, compliant solutions through Google Distributed Cloud in both air-gapped and connected deployments. Air-gapped environments operate offline for maximum data isolation. Connected deployments enable secure integration with cloud services. These scenarios meet data sovereignty and regulatory demands in sectors such as finance, government and healthcare without compromising innovation.
- Google Distributed Cloud sandbox environment: Google Distributed Cloud sandbox environment is a digital playground where developers can build, test and deploy industry-specific and sovereign cloud deployments. This sandbox will help teams upskill through hands-on training and accelerate time to market with Google Distributed Cloud technologies through preconfigured, ready-to-deploy templates.
NTT DATA will support these innovations through a full-stack suite of services including advisory, building, implementation and ongoing hosting and managed services.
By combining NTT DATA’s proven blueprints and delivery expertise with Google Cloud’s technology, the partnership will accelerate the development of repeatable, scalable solutions for enterprise transformation. At the heart of this innovation strategy is Takumi, NTT DATA’s GenAI framework that guides clients from ideation to enterprise-wide deployment. Takumi integrates seamlessly with Google Cloud’s AI stack, enabling rapid prototyping and operationalization of GenAI use cases.
This initiative expands NTT DATA’s Smart AI Agent Ecosystem, which unites strategic technology partnerships, specialized assets and an AI-ready talent engine to help clients deploy and manage responsible, business-driven AI at scale.
Accelerating global delivery with a dedicated Google Cloud Business Group:
To achieve excellence, NTT DATA has established a dedicated global Google Cloud Business Group comprising thousands of engineers, architects and advisory consultants. This global team at NTT DATA will work in close collaboration with Google Cloud teams to help clients adopt and scale AI-powered cloud technologies.
NTT DATA is also investing in advanced training and certification programs ensuring teams across sales, pre-sales and delivery are equipped to sell, secure, migrate and implement AI-powered cloud solutions. The company aims to certify 5,000 engineers in Google Cloud technology, further reinforcing its role as a leader in cloud transformation on a global scale.
Additionally, both companies are co-investing in global sales and go-to-market campaigns to accelerate client adoption across priority industries. By aligning technical, sales and marketing expertise, the companies aim to scale transformative solutions efficiently across global markets.
This global partnership builds on NTT DATA and Google Cloud’s 2024 co-innovation agreement in APAC. In addition it further strengthens NTT DATA’s acquisition of Niveus Solutions, a leading Google Cloud specialist recognized with three 2025 Google Cloud Awards – “Google Cloud Country Partner of the Year – India”, “Google Cloud Databases Partner of the Year – APAC” and “Google Cloud Country Partner of the Year – Chile,” further validating NTT DATA’s commitment to cloud excellence and innovation.
“We’re excited to see the strengthened partnership between NTT DATA and Google Cloud, which continues to deliver measurable impact. Their combined expertise has been instrumental in migrating more than 380 workloads to Google Cloud to align with our cloud-first strategy,” said José Luis González Santana, Head of IT Infrastructure, Carrefour. “By running SAP HANA on Google Cloud, we have consolidated 100 legacy applications to create a powerful, modernized e-commerce platform across 200 hypermarkets. This transformation has given us the agility we need during peak times like Black Friday and enabled us to launch new services faster than ever. Together, NTT DATA and Google Cloud are helping us deliver more connected, seamless experiences for our customers,”
About NTT DATA:
NTT DATA is a $30+ billion trusted global innovator of business and technology services. We serve 75% of the Fortune Global 100 and are committed to helping clients innovate, optimize and transform for long-term success. As a Global Top Employer, we have experts in more than 50 countries and a robust partner ecosystem of established and start-up companies. Our services include business and technology consulting, data and artificial intelligence, industry solutions, as well as the development, implementation and management of applications, infrastructure and connectivity. We are also one of the leading providers of digital and AI infrastructure in the world. NTT DATA is part of NTT Group, which invests over $3.6 billion each year in R&D to help organizations and society move confidently and sustainably into the digital future.
Resources:
https://www.nttdata.com/global/en/news/press-release/2025/august/081300
Google Cloud targets telco network functions, while AWS and Azure are in holding patterns
Deutsche Telekom and Google Cloud partner on “RAN Guardian” AI agent
Ericsson and Google Cloud expand partnership with Cloud RAN solution
NTT & Yomiuri: ‘Social Order Could Collapse’ in AI Era
AI Meets Telecom: Automating RF Plumbing Diagrams at Scale
By Chode Balaji with Ajay Lotan Thakur
In radio frequency (RF) circuit design, an RF plumbing diagram is a visual representation of how components such as antennas, amplifiers, filters, and cables are physically interconnected to manage RF signal flow across a network node. Unlike logical or schematic diagrams, these diagrams emphasize signal routing, cable paths, and component connectivity, ensuring spectrum compliance and accurate transmission behavior.
In this article, I introduce an AI-powered automation platform designed to generate RF plumbing diagrams for complex telecom deployments, which dramatically reduces manual effort and engineering errors. The system has been field-tested within a major telecom provider’s RF design workflow, showing measurable reduction in design time and increased compliance, where it has cut design time from hours to minutes while standardizing outputs across markets. We discuss the architecture of the platform, its real-world use cases, and the broader implications for network scalability and compliance in next-generation RF deployments.
Introduction
In RF circuit and system design, an RF plumbing diagram is a critical visual blueprint that shows how physical components—antennas, cables, combiners, duplexers, and power sources—are interconnected to manage signal flow across a network node. Unlike logical network schematics, these diagrams emphasize actual deployment wiring, routing, and interconnection details across multiple frequency bands and sectors.
As networks become increasingly dense and distributed, especially with 5G and Open RAN architectures, RF plumbing diagrams have grown in both complexity and importance. Yet across the industry, they are still predominantly created using manual methods—introducing inconsistency, delay, and high operational cost [1].
Challenges with Manual RF Documentation
Creating RF plumbing diagrams manually demands deep subject matter expertise, detailed knowledge of hardware interconnections, and alignment with region-specific compliance standards. Each diagram can take hours to complete, and even minor errors—such as incorrect port mappings or misaligned frequency bands—can result in service degradation, failed field validations, or regulatory non-compliance. In some cases, incorrect diagrams have delayed spectrum audits or triggered failed E911 checks, which are critical in public safety contexts.
Compliance requirements often vary by country due to differences in spectrum licensing, environmental limits, and emergency services integration. For example, the U.S. mandates specific RF configuration standards for E911 systems [3], while European operators must align with ETSI guidelines [4].
According to industry discussions on automation in telecom operations [1], reducing manual overhead and standardizing documentation workflows is a key goal for next-generation network teams.
System Overview – AI Powered Diagram Generation
To streamline this process, we developed CERTA RFDS—a system that automates RF plumbing diagram generation using input configuration data. CERTA ingests band and sector mappings, node configurations, and passive element definitions, then applies business logic to render a complete, standards-aligned diagram.
The system is built as a cloud-native microservice and can be integrated into OSS workflows or CI/CD pipelines used by RF planning teams. Its modular engine outputs standardized SVG/PDFs and maintains design versioning aligned with audit requirements.
This system aligns with automation trends seen in AI-native telecom operations [1] and can scale to support edge-native deployments as part of broader infrastructure-as-code workflows.
Deployment and Public Availability
The CERTA RFDS system has been internally validated within major telecom design teams and is now available publicly for industry adoption. It has demonstrated consistent savings in engineering time—reducing diagram effort from 2–4 hours to under 5 minutes per node—while improving compliance through template consistency. These results and the underlying platform were presented at the IEEE International Conference on Emerging and Advanced Information Systems (EEAIS) [5]. (Note – Paper is presented at EEAIS 2025; publication pending)
Output Showcase and Engineering Impact
Below is a sample RF plumbing diagram generated by the CERTA platform for a complex LTE and UMTS multi-sector node. The system automatically determines feed paths, port mappings, and labeling conventions based on configuration metadata.
As 5G networks continue to roll out globally, RF plumbing diagrams are becoming even more complex due to increased densification, the use of small cells, and the incorporation of mmWave technologies. The AI-driven automation framework we developed is fully adaptable to 5G architecture. It supports configuration planning for high-frequency spectrum bands, MIMO antenna arrangements, and ensures that E911 and regulatory compliance standards are maintained even in ultra-dense urban deployments. This makes the system a valuable asset in accelerating the design and validation processes for next generation 5G infrastructure.
Figure 1. AI-Generated RF Plumbing Diagram from CERTA RFDS: Illustrating dual-feed, multi-sector layout for LTE and UMTS deployment.
Benefits include:
- 90%+ time savings per node
- Consistency across regions and engineering teams
- Simplified field validation and compliance review
Future Scope
CERTA RFDS is being extended to support:
- GIS visualization of RF components with geo-tagged layouts
- Integration with planning systems for real-time topology generation
- LLM-based auto-summary of node-level changes for audit documentation
Conclusion
RF plumbing diagrams are fundamental to reliable telecom deployment and compliance. By shifting from manual workflows to intelligent automation, systems like CERTA RFDS enable engineers and operators to scale with confidence, consistency, and speed—meeting the challenges of modern wireless networks.
Abbreviation
- CERTA RFDS – Cognitive Engineering for Rapid RFDS Transformation & Automation
- RFDS – Radio Frequency Data Sheet
- GIS – Geographic Information System
- LLM – Large Language Model
- OSS – Operations Support System
- MIMO – Multiple Input Multiple Output
- RF – Radio Frequency
Reference
[1] ZTE’s Vision for AI-Native Infrastructure and AI-Powered Operations
[5] IEEE EEAIS 2025 Conference, “CERTA RFDS: Automating RF Plumbing Diagrams at Scale,”
About Author
Balaji Chode is an AI Solutions Architect at UBTUS, where he leads telecom automation initiatives including the design and deployment of CERTA RFDS. He has contributed to large-scale design and automation platforms across telecom and public safety, authored multiple peer-reviewed articles, and filed several patents.
“The author acknowledges the use of AI-assisted tools for language refinement and formatting”
OpenAI announces new open weight, open source GPT models which Orange will deploy
Overview:
OpenAI today introduced two new open-weight, open-source GPT models (gpt-oss-120b and gpt-oss-20b) designed to deliver top-tier performance at a lower cost. Available under the flexible Apache 2.0 license, these models outperform similarly sized open models on reasoning tasks, demonstrate strong tool use capabilities, and are optimized for efficient deployment on consumer hardware. They were trained using a mix of reinforcement learning and techniques informed by OpenAI’s most advanced internal models, including o3 and other frontier systems.
These two new AI models require much less compute power to run, with the gpt-oss20B version able to run on just 16 GB of memory. The smaller memory size and less compute power enables OpenAI’s models to run in a wider variety of environments, including at the network edge. The open weights mean those using the models can tweak the training parameters and customize them for specific tasks.
OpenAI has been working with early partner companies, including AI Sweden, Orange, and Snowflake to learn about real-world applications of our open models, from hosting these models on-premises for data security to fine-tuning them on specialized datasets. We’re excited to provide these best-in-class open models to empower everyone—from individual developers to large enterprises to governments—to run and customize AI on their own infrastructure. Coupled with the models available in our API, developers can choose the performance, cost, and latency they need to power AI workflows.
In lockstep with OpenAI, France’s Orange today announced plans to deploy the new OpenAI models in its regional cloud data centers as well as small on-premises servers and edge sites to meet demand for sovereign AI solutions. Orange’s deep AI engineering talent enables it to customize and distill the OpenAI models for specific tasks, effectively creating smaller sub-models for particular use-cases, while ensuring the protection of all sensitive data used in these customized models. This process facilitates innovative use-cases in network operations and will enable Orange to build on its existing suite of ‘Live Intelligence’ AI solutions for enterprises, as well as utilizing it for its own operational needs to improve efficiency, and drive cost savings.
Using AI to improve the quality and resilience of its networks, for example by enabling Orange to more easily explore and diagnose complex network issues with the help of AI. This can be achieved with trusted AI models that operate entirely within Orange sovereign data centers where Orange has complete control over the use of sensitive network data. This ability to create customized, secure, and sovereign AI models for network use cases is a key enabler in Orange’s mission to achieve higher levels of automation across all of its networks.
Steve Jarrett, Orange’s Chief AI Officer, noted the decision to use state-of-the-art open-weight models will allow it to drive “new use cases to address sensitive enterprise needs, help manage our networks, enable innovating customer care solutions including African regional languages, and much more.”
Performance of the new OpenAI models:
gpt-oss-120b outperforms OpenAI o3‑mini and matches or exceeds OpenAI o4-mini on competition coding (Codeforces), general problem solving (MMLU and HLE) and tool calling (TauBench). It furthermore does even better than o4-mini on health-related queries (HealthBench) and competition mathematics (AIME 2024 & 2025). gpt-oss-20b matches or exceeds OpenAI o3‑mini on these same evals, despite its small size, even outperforming it on competition mathematics and health.
Sovereign AI Market Forecasts:
Open-weight and open-source AI models play a significant role in enabling and shaping the development of Sovereign AI, which refers to a nation’s or organization’s ability to control its own AI technologies, data, and infrastructure to meet its specific needs and regulations.
Sovereign AI refers to a nation’s ability to control and manage its own AI development and deployment, including data, infrastructure, and talent. It’s about ensuring a country’s strategic autonomy in the realm of artificial intelligence, enabling them to leverage AI for their own economic, social, and security interests, while adhering to their own values and regulations.
Bank of America’s financial analysts recently forecast the sovereign AI market segment could become a “$50 billion a year opportunity, accounting for 10%–15% of the global $450–$500 billion AI infrastructure market.”
BofA analysts said, “Sovereign AI nicely complements commercial cloud investments with a focus on training and inference of LLMs in local culture, language and needs,” and could mitigate challenges such as “limited power availability for data centers in US” and trade restrictions with China.
References:
https://openai.com/index/introducing-gpt-oss/
https://newsroom.orange.com/orange-and-openai-collaborate-on-trusted-responsible-and-inclusive-ai/
https://finance.yahoo.com/news/nvidia-amd-targets-raised-bofa-162314196.html
Open AI raises $8.3B and is valued at $300B; AI speculative mania rivals Dot-com bubble
OpenAI partners with G42 to build giant data center for Stargate UAE project
Reuters & Bloomberg: OpenAI to design “inference AI” chip with Broadcom and TSMC
Nvidia’s networking solutions give it an edge over competitive AI chip makers
Nvidia’s networking equipment and module sales accounted for $12.9 billion of its $115.1 billion in data center revenue in its prior fiscal year. Composed of its NVLink, InfiniBand, and Ethernet solutions, Nvidia’s networking products (from its Mellanox acquisition) are what allow its GPU chips to communicate with each other, let servers talk to each other inside massive data centers, and ultimately ensure end users can connect to it all to run AI applications.
“The most important part in building a supercomputer is the infrastructure. The most important part is how you connect those computing engines together to form that larger unit of computing,” explained Gilad Shainer, senior vice president of networking at Nvidia.
In Q1-2025, networking made up $4.9 billion of Nvidia’s $39.1 billion in data center revenue. And it’ll continue to grow as customers continue to build out their AI capacity, whether that’s at research universities or massive data centers.
“It is the most underappreciated part of Nvidia’s business, by orders of magnitude,” Deepwater Asset Management managing partner Gene Munster told Yahoo Finance. “Basically, networking doesn’t get the attention because it’s 11% of revenue. But it’s growing like a rocket ship. “[Nvidia is a] very different business without networking,” Munster explained. “The output that the people who are buying all the Nvidia chips [are] desiring wouldn’t happen if it wasn’t for their networking.”
Nvidia senior vice president of networking Kevin Deierling says the company has to work across three different types of networks:
- NVLink technology connects GPUs to each other within a server or multiple servers inside of a tall, cabinet-like server rack, allowing them to communicate and boost overall performance.
- InfiniBand connects multiple server nodes across data centers to form what is essentially a massive AI computer.
- Ethernet connectivity for front-end network for storage and system management.
Note: Industry groups also have their own competing networking technologies including UALink, which is meant to go head-to-head with NVLink, explained Forrester analyst Alvin Nguyen.
“Those three networks are all required to build a giant AI-scale, or even a moderately sized enterprise-scale, AI computer,” Deierling explained. Low latency is key as longer transit times for data going to/from GPUs slows the entire operation, delaying other processes and impacting the overall efficiency of an entire data center.
Nvidia CEO Jensen Huang presents a Grace Blackwell NVLink72 as he delivers a keynote address at the Consumer Electronics Show (CES) in Las Vegas, Nevada on January 6, 2025. Photo by PATRICK T. FALLON/AFP via Getty Images
As companies continue to develop larger AI models and autonomous and semi-autonomous agentic AI capabilities that can perform tasks for users, making sure those GPUs work in lockstep with each other becomes increasingly important.
The AI industry is in the midst of a broad reordering around the idea of inferencing, which requires more powerful data center systems to run AI models. “I think there’s still a misperception that inferencing is trivial and easy,” Deierling said.
“It turns out that it’s starting to look more and more like training as we get to [an] agentic workflow. So all of these networks are important. Having them together, tightly coupled to the CPU, the GPU, and the DPU [data processing unit], all of that is vitally important to make inferencing a good experience.”
Competitor AI chip makers, like AMD are looking to grab more market share from Nvidia, and cloud giants like Amazon, Google, and Microsoft continue to design and develop their own AI chips. However, none of them have the low latency, high speed connectivity solutions provided by Nvidia (again, think Mellanox).
References:
https://www.nvidia.com/en-us/networking/
Networking chips and modules for AI data centers: Infiniband, Ultra Ethernet, Optical Connections
Nvidia enters Data Center Ethernet market with its Spectrum-X networking platform
Superclusters of Nvidia GPU/AI chips combined with end-to-end network platforms to create next generation data centers
Does AI change the business case for cloud networking?
The case for and against AI-RAN technology using Nvidia or AMD GPUs
Telecom sessions at Nvidia’s 2025 AI developers GTC: March 17–21 in San Jose, CA
Open AI raises $8.3B and is valued at $300B; AI speculative mania rivals Dot-com bubble
According to the Financial Times (FT), OpenAI (the inventor of Chat GPT) has raised another $8.3 billion in a massively over-subscribed funding round, including $2.8 billion from Dragoneer Investment Group, a San Francisco-based technology-focused fund. Leading VCs that also participated in the funding round included Founders Fund, Sequoia Capital, Andreessen Horowitz, Coatue Management, Altimeter Capital, D1 Capital Partners, Tiger Global and Thrive Capital, according to the people with knowledge of the deal.
The oversubscribed funding round came months ahead of schedule. OpenAI initially raised $2.5 billion from VC firms in March when it announced its intention to raise $40 billion in a round spearheaded by SoftBank. The Chat GPT maker is now valued at $300 billion.
OpenAI’s annual recurring revenue has surged to $12bn, according to a person with knowledge of OpenAI’s finances, and the group is set to release its latest model, GPT-5, this month.
OpenAI is in the midst of complex negotiations with Microsoft that will determine its corporate structure. Rewriting the terms of the pair’s current contract, which runs until 2030, is seen as a prerequisite to OpenAI simplifying its structure and eventually going public. The two companies have yet to agree on key issues such as how long Microsoft will have access to OpenAI’s intellectual property. Another sticking point is the future of an “AGI clause”, which allows OpenAI’s board to declare that the company has achieved a breakthrough in capability called “artificial general intelligence,” which would then end Microsoft’s access to new models.
An additional risk is the increasing competition from rivals such as Anthropic — which is itself in talks for a multibillion-dollar fundraising — and is also in a continuing legal battle with Elon Musk. The FT also reported that Amazon is set to increase its already massive investment in Anthropic.
OpenAI CEO Sam Altman. The funding forms part of a round announced in March that values the ChatGPT maker at $300bn © Yuichi Yamazaki/AFP via Getty Images
………………………………………………………………………………………………………………………………………………………………………………………………………………………………………………………………………………………..
While AI is the transformative technology of this generation, comparisons are increasingly being made with the Dot-com bubble. 1999 saw such a speculative frenzy for anything with a ‘.com’ at the end that valuations and stock markets reached unrealistic and clearly unsustainable levels. When that speculative bubble burst, the global economy fell into an extended recession in 2001-2002. As a result, analysts are now questioning the wisdom of the current AI speculative bubble and fearing dire consequences when it eventually bursts. Just as with the Dot-com bubble, AI revenues are nowhere near justifying AI company valuations, especially for private AI companies that are losing tons of money (see Open AI losses detailed below).
Torsten Slok, Partner and Chief Economist at Apollo Global Management via ZERO HEDGE on X: “The difference between the IT bubble in the 1990s and the AI bubble today is that the top 10 companies in the S&P 500 today (including Nvidia, Microsoft, Amazon, Google, and Meta) are more overvalued than the IT companies were in the 1990s.”
AI private companies may take a lot longer to reach the lofty profit projections institutional investors have assumed. Their reliance on projected future profits over current fundamentals is a dire warning sign to this author. OpenAI, for example, faces significant losses and aggressive revenue targets to become profitable. OpenAI reported an estimated loss of $5 billion in 2024, despite generating $3.7 billion in revenue. The company is projected to lose $14 billion in 2026 while total projected losses from 2023 to 2028 are expected to reach $44 billion.
Other AI bubble data points (publicly traded stocks):
- The proportion of the S&P 500 represented by the 10 largest companies is significantly higher now (almost 40%) compared to 25% in 1999. This indicates a more concentrated market driven by a few large technology companies deeply involved in AI development and adoption.
- Investment in AI infrastructure has reportedly exceeded the spending on telecom and internet infrastructure during the dot-com boom and continues to grow, suggesting a potentially larger scale of investment in AI relative to the prior period.
- Some indices tracking AI stocks have demonstrated exceptionally high gains in a short period, potentially surpassing the rates of the dot-com era, suggesting a faster build-up in valuations.
- The leading hyperscalers, such as Amazon, Microsoft, Google, and Meta, are investing vast sums in AI infrastructure to capitalize on the burgeoning AI market. Forecasts suggest these companies will collectively spend $381 billion in 2025 on AI-ready infrastructure, a significant increase from an estimated $270 billion in 2024.
Check out this YouTube video: “How AI Became the New Dot-Com Bubble”
References:
https://www.ft.com/content/76dd6aed-f60e-487b-be1b-e3ec92168c11
https://www.telecoms.com/ai/openai-funding-frenzy-inflates-the-ai-bubble-even-further
https://x.com/zerohedge/status/1945450061334216905
AI spending is surging; companies accelerate AI adoption, but job cuts loom large
Will billions of dollars big tech is spending on Gen AI data centers produce a decent ROI?
Canalys & Gartner: AI investments drive growth in cloud infrastructure spending
AI Echo Chamber: “Upstream AI” companies huge spending fuels profit growth for “Downstream AI” firms
Huawei launches CloudMatrix 384 AI System to rival Nvidia’s most advanced AI system
Gen AI eroding critical thinking skills; AI threatens more telecom job losses
Dell’Oro: AI RAN to account for 1/3 of RAN market by 2029; AI RAN Alliance membership increases but few telcos have joined
China gaining on U.S. in AI technology arms race- silicon, models and research
Introduction:
According to the Wall Street Journal, the U.S. maintains its early lead in AI technology with Silicon Valley home to the most popular AI models and the most powerful AI chips (from Santa Clara based Nvidia and AMD). However, China has shown a willingness to spend whatever it takes to take the lead in AI models and silicon.
The rising popularity of DeepSeek, the Chinese AI startup, has buoyed Beijing’s hopes that it can become more self-sufficient. Huawei has published several papers this year detailing how its researchers used its homegrown AI chips to build large language models without relying on American technology.
“China is obviously making progress in hardening its AI and computing ecosystem,” said Michael Frank, founder of think tank Seldon Strategies.
AI Silicon:
Morgan Stanley analysts forecast that China will have 82% of AI chips from domestic makers by 2027, up from 34% in 2024. China’s government has played an important role, funding new chip initiatives and other projects. In July, the local government in Shenzhen, where Huawei is based, said it was raising around $700 million to invest in strengthening an “independent and controllable” semiconductor supply chain.
During a meeting with President Xi Jinping in February, Huawei Chief Executive Officer Ren Zhengfei told Xi about “Project Spare Tire,” an effort by Huawei and 2,000 other enterprises to help China’s semiconductor sector achieve a self-sufficiency rate of 70% by 2028, according to people familiar with the meeting.
……………………………………………………………………………………………………………………………………………
AI Models:
Prodded by Beijing, Chinese financial institutions, state-owned companies and government agencies have rushed to deploy Chinese-made AI models, including DeepSeek [1.] and Alibaba’s Qwen. That has fueled demand for homegrown AI technologies and fostered domestic supply chains.
Note 1. DeepSeek’s V3 large language model matched many performance benchmarks of rival AI programs developed in the U.S. at a fraction of the cost. DeepSeek’s open-weight models have been integrated into many hospitals in China for various medical applications.
In recent weeks, a flurry of Chinese companies have flooded the market with open-source AI models, many of which are claiming to surpass DeepSeek’s performance in certain use cases. Open source models are freely accessible for modification and deployment.
The Chinese government is actively supporting AI development through funding and policy initiatives, including promoting the use of Chinese-made AI models in various sectors.
Meanwhile, OpenAI’s CEO Sam Altman said his company had pushed back the release of its open-source AI model indefinitely for further safety testing.
AI Research:
China has taken a commanding lead in the exploding field of artificial intelligence (AI) research, despite U.S. restrictions on exporting key computing chips to its rival, finds a new report.
The analysis of the proprietary Dimensions database, released yesterday, finds that the number of AI-related research papers has grown from less than 8500 published in 2000 to more than 57,000 in 2024. In 2000, China-based scholars produced just 671 AI papers, but in 2024 their 23,695 AI-related publications topped the combined output of the United States (6378), the United Kingdom (2747), and the European Union (10,055).
“U.S. influence in AI research is declining, with China now dominating,” Daniel Hook, CEO of Digital Science, which owns the Dimensions database, writes in the report DeepSeek and the New Geopolitics of AI: China’s ascent to research pre-eminence in AI.
In 2024, China’s researchers filed 35,423 AI-related patent applications, more than 13 times the 2678 patents filed in total by the U.S., the U.K., Canada, Japan, and South Korea.
References:
https://www.wsj.com/tech/ai/how-china-is-girding-for-an-ai-battle-with-the-u-s-5b23af51
Huawei launches CloudMatrix 384 AI System to rival Nvidia’s most advanced AI system
U.S. export controls on Nvidia H20 AI chips enables Huawei’s 910C GPU to be favored by AI tech giants in China
Goldman Sachs: Big 3 China telecom operators are the biggest beneficiaries of China’s AI boom via DeepSeek models; China Mobile’s ‘AI+NETWORK’ strategy
Gen AI eroding critical thinking skills; AI threatens more telecom job losses
Softbank developing autonomous AI agents; an AI model that can predict and capture human cognition
AI spending is surging; companies accelerate AI adoption, but job cuts loom large
Big Tech and VCs invest hundreds of billions in AI while salaries of AI experts reach the stratosphere
ZTE’s AI infrastructure and AI-powered terminals revealed at MWC Shanghai
Deloitte and TM Forum : How AI could revitalize the ailing telecom industry?
Huawei launches CloudMatrix 384 AI System to rival Nvidia’s most advanced AI system
On Saturday, Huawei Technologies displayed an advanced AI computing system in China, as the Chinese technology giant seeks to capture market share in the country’s growing artificial intelligence sector. Huawei’s CloudMatrix 384 system made its first public debut at the World Artificial Intelligence Conference (WAIC), a three-day event in Shanghai where companies showcase their latest AI innovations, drawing a large crowd to the company’s booth.
The Huawei CloudMatrix 384 is a high-density AI computing system featuring 384 Huawei Ascend 910C chips, designed to rival Nvidia’s GB200 NVL72 (more below). The AI system employs a “supernode” architecture with high-speed internal chip interconnects. The system is built with optical links for low-latency, high-bandwidth communication. Huawei has also integrated the CloudMatrix 384 into its cloud platform. The system has drawn close attention from the global AI community since Huawei first announced it in April.
The CloudMatrix 384 resides on the super-node Ascend platform and uses high-speed bus interconnection capability, resulting in low latency linkage between 384 Ascend NPUs. Huawei says that “compared to traditional AI clusters that often stack servers, storage, network technology, and other resources, Huawei CloudMatrix has a super-organized setup. As a result, it also reduces the chance of facing failures at times of large-scale training.
Huawei staff at its WAIC booth declined to comment when asked to introduce the CloudMatrix 384 system. A spokesperson for Huawei did not respond to questions. However, Huawei says that “early reports revealed that the CloudMatrix 384 can offer 300 PFLOPs of dense BF16 computing. That’s double of Nvidia GB200 NVL72 AI tech system. It also excels in terms of memory capacity (3.6x) and bandwidth (2.1x).” Indeed, industry analysts view the CloudMatrix 384 as a direct competitor to Nvidia’s GB200 NVL72, the U.S. GPU chipmaker’s most advanced system-level product currently available in the market.
One industry expert has said the CloudMatrix 384 system rivals Nvidia’s most advanced offerings. Dylan Patel, founder of semiconductor research group SemiAnalysis, said in an April article that Huawei now had AI system capabilities that could beat Nvidia’s AI system. The CloudMatrix 384 incorporates 384 of Huawei’s latest 910C chips and outperforms Nvidia’s GB200 NVL72 on some metrics, which uses 72 B200 chips, according to SemiAnalysis. The performance stems from Huawei’s system design capabilities, which compensate for weaker individual chip performance through the use of more chips and system-level innovations, SemiAnalysis said.
Huawei has become widely regarded as China’s most promising domestic supplier of chips essential for AI development, even though the company faces U.S. export restrictions. Nvidia CEO Jensen Huang told Bloomberg in May that Huawei had been “moving quite fast” and named the CloudMatrix as an example.
Huawei says the system uses “supernode” architecture that allows the chips to interconnect at super-high speeds and in June, Huawei Cloud CEO Zhang Pingan said the CloudMatrix 384 system was operational on Huawei’s cloud platform.
According to Huawei, the Ascend AI chip-based CloudMatrix 384 with three important benefits:
- Ultra-large bandwidth
- Ultra-Low Latency
- Ultra-Strong Performance
These three perks can help enterprises achieve better AI training as well as stable reasoning performance for models. They could further retain long-term reliability.
References:
https://www.huaweicentral.com/huawei-launches-cloudmatrix-384-ai-chip-cluster-against-nvidia-nvl72/
https://semianalysis.com/2025/04/16/huawei-ai-cloudmatrix-384-chinas-answer-to-nvidia-gb200-nvl72/
U.S. export controls on Nvidia H20 AI chips enables Huawei’s 910C GPU to be favored by AI tech giants in China
Huawei’s “FOUR NEW strategy” for carriers to be successful in AI era
FT: Nvidia invested $1bn in AI start-ups in 2024
Gen AI eroding critical thinking skills; AI threatens more telecom job losses
Two alarming research studies this year have drawn attention to the damage that Gen AI agents like ChatGPT are doing to our brains:
The first study, published in February, by Microsoft and Carnegie Mellon University, surveyed 319 knowledge workers and concluded that “while GenAI can improve worker efficiency, it can inhibit critical engagement with work and can potentially lead to long-term overreliance on the tool and diminished skills for independent problem-solving.”
An MIT study divided participants into three essay-writing groups. One group had access to Gen AI and another to Internet search engines while the third group had access to neither. This “brain” group, as MIT’s researchers called it, outperformed the others on measures of cognitive ability. By contrast, participants in the group using a Gen AI large language model (LLM) did the worst. “Brain connectivity systematically scaled down with the amount of external support,” said the report’s authors.
Across the 20 companies regularly tracked by Light Reading, headcount fell by 51,700 last year. Since 2015, it has dropped by more than 476,600, more than a quarter of the previous total.
Source: Light Reading
………………………………………………………………………………………………………………………………………………
Doing More with Less:
- In 2015, Verizon generated sales of $131.6 billion with a workforce of 177,700 employees. Last year, it made $134.8 billion with fewer than 100,000. Revenues per employee, accordingly, have risen from about $741,000 to more than $1.35 million over this period.
- AT&T made nearly $868,000 per employee last year, compared with less than $522,000 in 2015.
- Deutsche Telekom, buoyed by its T-Mobile US business, has grown its revenue per employee from about $356,000 to more than $677,000 over the same time period.
- Orange’s revenue per employee has risen from $298,000 to $368,000.
Significant workforce reductions have happened at all those companies, especially AT&T which finished last year with 141,000 employees – about half the number it had in 2015!
Not to be outdone, headcount at network equipment companies are also shrinking. Ericsson, Europe’s biggest 5G vendor, cut 6,000 jobs or 6% of its workforce last year and has slashed 13,000 jobs since 2023. Nokia’s headcount fell from 86,700 in 2023 to 75,600 at the end of last year. The latest message from Börje Ekholm, Ericsson’s CEO, is that AI will help the company operate with an even smaller workforce in future. “We also see and expect big benefits from the use of AI, and that is one reason why we expect restructuring costs to remain elevated during the year,” he said on this week’s earnings call with analysts.
………………………………………………………………………………………………………………………………………………
Other Voices:
Light Reading’s Iain Morris wrote, “An erosion of brainpower and ceding of tasks to AI would entail a loss of control as people are taken out of the mix. If AI can substitute for a junior coder, as experts say it can, the entry-level job for programming will vanish with inevitable consequences for the entire profession. And as AI assumes responsibility for the jobs once done by humans, a shrinking pool of individuals will understand how networks function.
“If you can’t understand how the AI is making that decision, and why it is making that decision, we could end up with scenarios where when something goes wrong, we simply just can’t understand it,” said Nik Willetts, the CEO of a standards group called the TM Forum, during a recent conversation with Light Reading. “It is a bit of an extreme to just assume no one understands how it works,” he added. “It is a risk, though.”
………………………………………………………………………………………………………………………………………………
References:
AI spending is surging; companies accelerate AI adoption, but job cuts loom large
Verizon and AT&T cut 5,100 more jobs with a combined 214,350 fewer employees than 2015
Big Tech post strong earnings and revenue growth, but cuts jobs along with Telecom Vendors
Nokia (like Ericsson) announces fresh wave of job cuts; Ericsson lays off 240 more in China
Deutsche Telekom exec: AI poses massive challenges for telecom industry
