Detailed write-ups
1. Samsung and KDDI complete AI-powered network optimization trial on commercial 5G SA
Samsung Newsroom · Jun 30, 2026
Samsung and KDDI announced the successful completion of a multi-month field trial of Samsung’s AI-powered RAN Speed Optimizer (RSO) on KDDI’s live commercial 5G Standalone network across dense urban, suburban and rural areas of Tokyo. The headline number is concrete: an average 31% increase in peak-hour 5G downlink throughput across the trial area, with gains reaching 52% in dense urban zones. What makes this notable for NetOps readers is the mechanism — RSO applies distinct, AI-tuned parameter settings per individual cell rather than the conventional cluster-level approach that forces the same configuration across many cells. That’s a genuine methodological shift, and one of the few AI-network claims this week backed by a live commercial-network result rather than a lab demo or vendor pitch deck. RSO is part of Samsung’s broader CognitiV Network Operations Suite, and the companies plan to keep evaluating AI-based optimization for wider commercial deployment.
Read the article →
Sources: Samsung Newsroom
2. Indosat outlines AI Grid vision as 5G modernization targets nationwide AI-ready network
RCR Wireless · Jul 2, 2026
Indosat Ooredoo Hutchison CEO Vikram Sinha detailed how the operator’s nationwide 5G modernization with Nokia — deploying Habrok and Pandion radios, Levante basebands, Centralized RAN and network automation, with low-band 5G nationwide and mid-band reaching roughly 80% of the network over three and a half years — is designed as the architectural foundation for an “AI Grid” rather than a coverage exercise alone. The framing treats AI-RAN as the common substrate combining network connectivity with distributed AI compute, already supporting the Indonesian-language Sahabat AI model in education, healthcare and government services. Indosat, Nokia and NVIDIA are also collaborating on AI-RAN field trials by end of 2026. For NetOps readers, this is a useful data point on how operators are sequencing modernization: the RAN upgrade is being sold internally and externally as AI infrastructure investment, not just a capacity refresh.
Read the article →
Sources: RCR Wireless
3. Verizon CTO touts generative AI-fueled network autonomy efforts
SDxCentral · Jul 2, 2026
Verizon CTO Yago Tenorio gave one of the more candid operator accounts of the autonomy journey this year, describing a path that began with virtualizing core and RAN assets a decade ago, progressed to closing 70 million automation “loops” a year without human intervention, and is now accelerating via Claude Code, Anthropic and Gemini models rolled out broadly across Verizon’s technology organization. Crucially, Verizon built its own model-agnostic autonomy platform rather than buying a vendor stack, explicitly to avoid lock-in to any single LLM as the field evolves. Tenorio said specialized agents doing root-cause analysis and remediation are already cutting real-life resolution tasks from hours to 90 seconds in prototype, with production rollout targeted within three months starting in the radio, then transport and core. His own framing: he wants his agents to “graduate from university” and “find a job in operations” within six months — a vivid way of describing the leap from pilot to production that the whole segment is chasing.
Read the article →
Sources: SDxCentral
4. Tech Mahindra partners with Microsoft on AI-driven 5G network digital twin
TechAfrica News · Jun 30, 2026
Tech Mahindra and Microsoft unveiled an advanced Network Digital Twin solution that integrates Azure, Microsoft Fabric and Azure Digital Twin services to unify high-volume network telemetry into a real-time, AI-ready data estate, then layers on Microsoft Foundry, Fabric IQ and agentic AI frameworks for reasoning, autonomous decision-making and closed-loop orchestration. The pitch, in the words of Tech Mahindra’s chief transformation officer, is moving digital twins “from a visualization layer into an active decision-making platform.” That’s the right distinction for NetOps readers to hold onto: most network digital twins to date have been passive simulation and planning tools, and the value claim here is closing the loop so the twin can drive real-time actions like SLA-driven network slicing and edge orchestration for enterprise 5G monetization, not just visualize the network for humans.
Read the article →
Sources: TechAfrica News
5. China Tower to accelerate shift toward “digital” and “intelligent” towers
RCR Wireless · Jun 30, 2026
Speaking at MWC Shanghai, China Tower chairman Zhang Zhiyong committed the company — which operates roughly 6.2 million base station sites, including 3.26 million 5G sites, across China — to accelerating the transformation of traditional towers into “digital and intelligent” infrastructure. About 250,000 towers have already been upgraded, supporting applications from environmental monitoring to emergency response. The plan combines shared communications, computing, power and security resources at each site, extends into low-altitude infrastructure for drones and UAVs, and applies AI to maintenance so infrastructure becomes “visible, manageable and controllable.” For NetOps readers outside China, the scale is the story: this is passive infrastructure — towers, not RAN software — being reimagined as a shared, AI-managed compute and connectivity layer, a model that shared-infrastructure providers elsewhere may eventually look to replicate.
Read the article →
Sources: RCR Wireless
6. Netgear brings AI-driven network management to SMEs and MSPs
Network World · Jun 30, 2026
Netgear launched Insight 10.0, its cloud-based network management platform, adding AI-powered operations, contextual insights, proactive troubleshooting recommendations and AI-assisted workflows aimed squarely at small and midsize enterprises and the managed service providers that serve them. Netgear frames the release as a foundation for future “AI-defined networking,” explicitly extending the AIOps and agentic-management wave that has already reached Cisco, HPE and Extreme into a market segment that has historically lacked the IT resources to adopt enterprise-grade automation. A beta customer credited the release with faster onboarding and less time lost to post-deployment troubleshooting and manual configuration — exactly the kind of unglamorous, high-frequency pain that AI-assisted workflows are best positioned to address. Worth watching as a signal that agentic network management is no longer a large-enterprise-only story.
Read the article →
Sources: Network World
7. Tencent Cloud used AI agents for XLSMART migration
RCR Wireless · Jul 1, 2026
Tencent Cloud disclosed one of the more striking agentic-AI case studies of the year: a 4.5-month, zero-downtime cloud migration of Indonesian telecom XLSMART’s core systems — over 60 applications, 1,200 microservices, 1,100 APIs and more than 15TB of data — using a portfolio of 20-plus proprietary agentic AI “skills.” Tools like CodeBuddy, WorkBuddy, TokenHub and DatabaseClaw handled resource discovery, architecture mapping, low-level design generation and live cutover monitoring across a single AI-driven pipeline, rather than the siloed, manual, spreadsheet-driven process migrations of this scale traditionally require. XLSMART, formed from the 2025 merger of XL Axiata and Smartfren and serving 69 million users, is a genuinely mission-critical proving ground. Tencent has now packaged the approach into a reusable migration platform tied to its Agent Development Platform 3.0 — a notable data point on agentic AI operating in a high-stakes telecom environment without a reported incident, and a competitive marker against AWS, Azure and Google Cloud in Southeast Asia.
Read the article →
Sources: RCR Wireless
8. U.S. Open powers up AI-ready network in challenging environment
Network World · Jun 29, 2026
Cisco and the USGA’s network for the 2026 U.S. Open at Shinnecock Hills is a useful real-world stress test of AI-ready networking principles: a fully redundant, mobile-core design built in pairs, ring-topology fiber, roughly 500 Wi-Fi 7 access points and 100+ switches, and strict per-client segmentation treating every fan, vendor and broadcast device as untrusted. The AI-specific angle is the USGA’s rules assistant, a knowledge-graph-backed application wrapped in Cisco AI Defense controls against prompt injection and misuse, with a hard rule that low-confidence answers escalate to human rules experts rather than guessing. The broader lesson for NetOps readers, explicit in the piece: AI assistants embedded in mission-critical workflows need the same observability, security and redundancy rigor as core network infrastructure — and automation should scale expert judgment, not replace it.
Read the article →
Sources: Network World
9. Enterprise network teams are falling behind as AI raises the stakes (FOUNDATIONAL)
Network World · Jun 8, 2026
The essential counterweight to this issue’s optimism: EMA’s Network Management Megatrends 2026 survey of 352 IT professionals found only 31% consider their network operations strategy completely successful, down from 42% two years ago, even as AI workloads arrive on networks not built for them. The talent shortage has worsened sharply — 52% now find it difficult to hire network experts, up from 26% in 2022 — and skills gaps are the top-cited barrier to automation. Nearly half of respondents already run AI training or inference workloads on their networks, but only 35% say their observability tools are ready to manage them. For NetOps readers, this is the reality check underneath every autonomy headline this week: 55% of organizations now require AI features when evaluating new tools, but successful teams are distinguished less by tool count than by data quality discipline and MCP-based integration across tool sprawl.
Read the article →
Sources: Network World
10. Nokia, Google Cloud unleash Gemini-powered AI agents to fix telco network headaches (FOUNDATIONAL)
Fierce Network · Jun 22, 2026
Ahead of DTW Ignite, Nokia and Google Cloud detailed six specialized Gemini-powered agents — a router agent, event triage agent, KPI selector agent, anomaly reasoner, action reasoner and dashboard agent — designed to move telecom operators past manual troubleshooting. Nokia claims the agents can cut problem-solving time 50% to 80%, with the router and event triage agents launching via Nokia Assurance Center on Google Cloud Marketplace in September. The most useful detail for NetOps readers is the “glass box” design philosophy: full observability and human oversight paired with autonomous capability, where most actions surface as a recommendation for human approval rather than executing unilaterally. Nokia executives argue this explainability, not raw capability, is precisely what has been missing to build operator trust in agentic systems — a direct throughline to the trust debate that dominated last week’s DTW coverage.
Read the article →
Sources: Fierce Network
11. NVIDIA brings trusted, 24/7 AI agents to telecom operations (FOUNDATIONAL)
NVIDIA Blog · Jun 22, 2026
NVIDIA’s DTW Ignite showcase laid out the infrastructure stack underneath this week’s operator announcements: synthetic, privacy-safe training data (used by SoftBank to fine-tune telecom models without exposing customer records); secure agent runtimes like NemoClaw and OpenShell that give long-running agents policy-based guardrails and auditable, sandboxed access to telecom systems; and GPU-accelerated simulation that lets agents validate network changes before touching live systems. Partners spanning AdaptKey, Amdocs, NTT DATA, ServiceNow, TCS, Forsk, VIAVI and KDDI are all building on these pieces for use cases from self-healing 5G RAN to proactive customer care to RAN digital twins. Read this as the connective infrastructure layer beneath the Samsung/KDDI trial, the Verizon platform, and the Tech Mahindra/Microsoft digital twin elsewhere in this issue — the shared plumbing that turns individual agentic pilots into a coherent autonomy stack.
Read the article →
Sources: NVIDIA Blog
12. TelcoAgent grounds 5G network forecasting in 3GPP standards (FOUNDATIONAL)
RCR Wireless · Jun 23, 2026
A June 2026 arXiv preprint introduces TelcoAgent, a research framework combining a time-series foundation model, a multi-agent LLM reasoning layer, and an automatically constructed 3GPP knowledge graph to forecast seven 5G KPMs across 200 cells in zero-shot fashion — without site-specific retraining — then explain likely causes and propose remediations grounded in standards language. Evaluated on a three-month, 200-cell U.S. operator dataset, TelcoAgent reportedly beats established forecasting baselines and produces standards-traceable explanations. The valuable part of this write-up for NetOps readers is its candor about limits: the evaluation covers a single operator over three months, there is no demonstrated closed-loop production deployment or NMS integration, and governance and certification frameworks for autonomous telecom agents simply don’t exist yet. A useful corrective to read alongside this week’s more polished vendor and operator claims.
Read the article →
Sources: RCR Wireless
13. Ericsson makes AI agents a first-class part of its OSS/BSS stack (FOUNDATIONAL)
RCR Wireless · Jun 26, 2026
Ericsson formalized a new OSS/BSS architectural blueprint that treats AI agents as a first-class component rather than a bolted-on Gen-AI feature, introducing an “agentic AI service experience layer” where a single natural-language intent — such as “reduce churn among high-value customers” — fans out across specialized Experience, Revenue and Network agents to generate consistent catalog, charging and provisioning changes. The framework links to Ericsson’s Intelligent Automation Platform (expanded into the core network, as covered in prior issues) and the Telco DataOps Platform for real-time data, closing the loop between network and business operations. Worth flagging for NetOps leaders: the most developed tooling runs on Amazon Bedrock, and while Ericsson offers broader cloud-agnostic options, the reference implementation’s AWS dependency raises real lock-in questions for operators not already committed to that cloud.
Read the article →
Sources: RCR Wireless
|
