Payman Samadi of eino.ai and Parm Sandhu of NTT DATA join Bob Laliberte of theCUBE Research for a NetworkANGLE CUBE Conversation that explores physical AI and evolving requirements for enterprise wireless networks. The discussion examines how artificial intelligence AI applied to physical systems, sensor fusion, agentic systems, digital twin models and agentic network operations influence the design, deployment and operations of private 5G, WiFi and edge computing environments.Sandhu emphasizes that physical AI requires deterministic connectivity, predictable latency and continuous observability to support agentic devices and autonomous robotics, and that wireless infrastructure must be resilient and managed as part of AI infrastructure. Samadi recommends that enterprises unify multi-technology networks, employ digital twins for design and validation and construct knowledge graphs and agentic automation for lifecycle management. Laliberte highlights the need for new operational models, governance and security to scale physical AI safely.

In this CUBE Conversation, Haseeb Budhani, co-founder and chief executive officer of Rafay Systems, joins theCUBE's John Furrier to discuss the operational complexity behind the global AI infrastructure buildup. Budhani explains how neoclouds, sovereign clouds and telcos are all converging on the same requirement: operationalizing massive GPU deployments so they can be monetized as a service. He points to sovereignty of compute and data as the primary driver of adoption across dozens of countries, with enterprises increasingly demanding the same guardrails, quotas and auditing capabilities that hyperscalers spent 15 years building.The conversation explores how Rafay addresses multi-tenancy at every layer, from network-level VPC constructs to Kubernetes, VMs and bare metal, so new AI clouds can deliver a true self-service experience rather than custom infrastructure hosting. Budhani details Rafay's "portfolio theory" approach to AI clouds, blending bare metal, Kubernetes, serverless and token-based consumption to maximize margins. He shares striking growth examples, including a customer scaling from 128 GPUs to a planned 200,000, and describes Rafay's delivery muscle — onboarding new customers in as little as 10 days. From Zero Trust Kubernetes access to deep ecosystem partnerships with NVIDIA, Dell and storage providers, Budhani outlines why speed, security and heterogeneous support are now the deciding factors in the AI infrastructure race.

RK Anand of Tensordyne, founder and chief product officer, joins theCUBE Conversation to discuss the company's Napier 3 nanometer chip and TDN72 rack-scale inference system for energy-efficient artificial intelligence. Anand explains the Napier log-math approach, which converts multiplications into additions to increase compute density and reduce power consumption, and describes partnerships with HPE and Juniper as well as engagements with Neocloud.Anand states Tensordyne reports up to 8x lower power consumption, 6x smaller floor space and up to 6x more revenue per footprint. They outline tape-out status and timelines: silicon return is expected in October, alpha and beta trials occur in Q1 2027 and production begins in mid-2027. Host John Furrier and theCUBE Research team frame economics as the primary market driver and discuss implications for agentic AI deployment and AI infrastructure involving TSMC and Broadcom collaborations.

In this CUBE Conversation, Dave Donatelli, chief executive officer of Riverbed, joins theCUBE's John Furrier to discuss the company's third-generation AI platform and its push toward full IT autonomy. Donatelli frames 2025 as the year when foundational bets made years earlier begin paying off in real, measurable outcomes. He details how Riverbed's centralized data store — collecting telemetry across applications, networks and devices — underpins a unified agent that collapses APM, NPM and device management into a single platform. With over 300 million autonomous operations completed and 59 customers running at least 10,000 automated workflows per month, Donatelli underscores that AI at Riverbed means machines resolving issues before humans ever notice them.The conversation also explores the six new capabilities Riverbed is launching, anchored by IQ 4.0, a second-generation agentic intelligence layer that adds contextual awareness and MCP-driven LLM routing to cut past the 25% error rates Gartner attributes to generic large language models. Donatelli breaks down AI Assurance — Riverbed's new framework for tracking which AI models are running inside an enterprise, how much they cost and how agentic applications are performing. He also details Data Express, a SaaS-based solution that moves data 10 times faster, compressing 30-day model reloads down to three days. On the financial side, Donatelli shares that customers are seeing five-to-one returns on their Riverbed investment, with some reaching ten-to-one through smart hardware refresh and software utilization analytics. From collapsing IT silos with a unified agent strategy to giving every practitioner a natural language interface into complex observability data, Donatelli outlines a roadmap for how enterprises can achieve genuine autonomous operations while keeping ROI firmly in focus.

Krista Case of TheCUBE Research, principal analyst and practice lead for cyber resilience and security, provides a concise analysis of recent announcements at SailPoint and VeeamON 2026. Case outlines how artificial intelligence agents emerge as operational actors, how SailPoint's Agentic Fabric advances identity governance, and how Veeam frames backup and recovery as an operational trust layer for AI-driven systems. They emphasize the operational and governance shifts required for resilient AI deployments.Case highlights that enterprises must shift from protecting users to protecting decisions and machine-generated outcomes by establishing visibility, accountability, lineage and recoverability. Practical implications include governance for non-human identities, behavioral analytics, privilege management and the emergence of resilience operations to prioritize and orchestrate AI-aware recovery and remediation across security, IT, data and compliance teams.This discussion addresses identity governance, cyber resilience, data protection and backup and recovery for AI-driven environments. It provides practitioners and decision makers with considerations for implementing governance, operational trust and coordinated recovery strategies to support resilient, secure AI systems.

This theCUBE Research interview, hosted by Bob Laliberte of theCUBE Research, principal analyst, features Viral Mehta of Wind River, vice president of engineering; Grant Challenger of Wind River, global director of ServiceNow solutions; and Neeraj Jain of ServiceNow, senior director of product management. The panel draws on telecom and cloud engineering experience to discuss ServiceNow’s private stack on Wind River, sovereign on-premises deployments, lifecycle automation, observability, hybrid cloud models and the implications of artificial intelligence for data locality and regulatory compliance.Jain notes that software as a service remains the primary model but demand for sovereign private stack accelerates, driven by geopolitics, regulation and AI locality; they emphasize the importance of lifecycle automation and observability for regulated and distributed deployments. Mehta highlights Wind River’s cloud-like operating model, automated blueprints, multi-site orchestration and layered self-healing that support six-nines availability; they stress how automated recovery and comprehensive observability enable predictable operations across sites. Challenger observes reduced operational cost, simplified vendor support and new regional revenue opportunities for service providers deploying the joint solution; they outline edge AI use cases and distributed deployment models that preserve data sovereignty while enabling cloud-native scale.

This conversation examines the transition from artificial intelligence experimentation to production. Paul Nashawaty of theCUBE Research, practice lead and principal analyst, interviews Kevin Cochrane of Vultr, chief marketing officer, at HumanX 2026 on AI infrastructure and production-ready inference.Cochrane outlines Vultr's global AI infrastructure strategy, Graphics Processing Unit and Central Processing Unit performance-per-dollar considerations, composable AI stacks, and the role of Kubernetes and partner ecosystems in enabling enterprise-scale training and inference. They highlight regional data sovereignty requirements and the need to adopt platform engineering to industrialize AI-native applications.Nashawaty emphasizes research showing 64% of organizations increase AI investment and the growing demand for portability governance and composability to move workloads from pilots to production.This discussion provides practical guidance for technology leaders responsible for AI infrastructure, platform engineering and cloud computing, including model serving performance, cost optimization and data sovereignty strategies.

Andrew Joiner of Hyperscience joins Scott Hebner of theCUBE Research to discuss the inference inflection point and to preview insights from Google Next. Joiner and Hebner examine how enterprises move artificial intelligence from experimentation to production by deploying AI factories, layered inference strategies and document-centric workflows and by meeting platform requirements for secure auditable deployment.Joiner states that inference, rather than models alone, is the operational bottleneck and that trusted data is the primary constraint; they outline a dual-loop architecture, cost-aware layered inference and human on the loop governance to ensure explainability and compliance. Hebner emphasizes that enterprises must optimize inference cost, invest in document operations infrastructure and prioritize security and auditability when deploying AI into core workflows.This discussion highlights practical factors to consider for operationalizing inference at scale, including architecture choices, cost management, governance frameworks and FedRAMP considerations for regulated environments. Viewers learn strategies for document intelligence, doc-ops implementation and platform selection that enable secure scalable AI deployment.