Key Findings of This Report
- The convergence of observability, AIOps (AI for Operations), and agentic AI is changing the way that enterprises and cloud providers manage infrastructure. That will require a new way of thinking to bridge the gap among infrastructure systems to drive automation in a safe and secure way.
- AIOps leverages AI and machine learning (ML) to automate and optimize network and infrastructure operations. It uses AI to analyze vast amounts of data from various sources, identify patterns, predict potential issues, and automate responses, ultimately improving efficiency.
- Agentic AI has changed the infrastructure automation game, placing a premium on validating data, governance, and orchestration across platforms. To safely employ agentic automation, enterprises will need to implement comprehensive safety, compliance, and governance.
- As infrastructure becomes more distributed, dynamic, and mission-critical, legacy tools that rely on dashboards, alerts, and manual interventions are no longer sufficient. Enterprises need a full observability and automation stack that moves beyond just visibility and insights to deliver agentic orchestration.
- Automation and orchestration platforms will be required to consolidate control over disparate AI and automation systems, while providing control and governance. A patchwork of scripting and infrastructure as code (IaC) tools no longer suffices to drive automation across infrastructure.
- Some of the companies highlighted in this report: Alkira, Arista Networks, Aviatrix, BE Networks, Cisco, Dynatrace, Extreme Networks, F5, Gigamon, HPE (Juniper), IBM (Ansible, HashiCorp), Itential, LogicMonitor, NetBox Labs, Nokia, Pulumi, Riverbed, Selector AI, Spacelift, System Initiative.
Introduction: Agentic AI Changes the Game in Autonomous Infrastructure
The convergence of observability, AIOps, and agentic AI is changing the way that enterprises and cloud providers manage infrastructure. That will require a new way of thinking to bridge the gap among infrastructure systems to drive automation in a safe and secure way.
Full infrastructure automation requires many technologies: comprehensive telemetry data, data management, AI/ML, and infrastructure as code (IaC) tooling. More importantly, however, these technology tools need to be unified to build orchestration platforms. As islands to themselves, they won’t provide governance, safety, and coordination of automation across hybrid infrastructure.
As infrastructure becomes more distributed, dynamic, and mission-critical, these trends will only become more important. Enterprises need an observability and automation stack that moves beyond visibility and insights to deliver agentic orchestration. This will require building orchestration platforms to help AI agents detect, decide, and trigger changes across tools in real time.
Futuriom believes the rapid emergence of AIOps and agentic AI technology signals a transformation in how network and infrastructure operations are executed. These innovations reflect a broader trend: the rise of AI-augmented automation systems that integrate observability, configuration management, execution, and policy into a single, programmable control plane. To adapt to this transformation, organizations need to standardize infrastructure data, moving from vendor-specific CLI outputs into structured formats that AI systems can understand. This foundation is critical to enabling effective AI decision-making, policy enforcement, and closed-loop automation across domains.
For this report, Futuriom analysts gathered data from networking case studies, user interviews, and vendor presentations, as well as public information of record, to put together a view on how advances in agentic AI will drive autonomous infrastructure. These automation technologies are evolving to support the next transformation of cloud, communications, and enterprise environments.
Networks Enter the Age of Agentic AI
Let’s talk about how autonomous infrastructure and networks are evolving. This is, of course, a complicated topic, but it’s evolving fast.
In theory, Agentic AI can enable the automation of cloud infrastructure by enabling selfremediating systems that go beyond static scripts and fixed playbooks – methods that have dominated the industry now for a couple of decades. This will be an evolution in addition to a revolution. Automating infrastructure is a delicate process, and practitioners need to get things right with data architectures and deep integrations among tools and infrastructure vendors.
First, let’s take a broader look at how cloud and enterprise infrastructure is changing and why that is likely to spur demand for better automation.
Hybrid Infrastructure Environments & Taxonomy
The distributed nature of enterprise infrastructure has created infrastructure and networking silos. If you are configuring an app to run across this infrastructure, it’s likely that an infrastructure team needs to do a lot of work to connect the moving parts, whether that’s connecting an enterprise datacenter to the cloud or enabling applications to run across diverse environments.
There are at least four major buckets of infrastructure environments: traditional enterprise networking, datacenter networking, cloud networking, and telecommunications infrastructure. Connecting these diverse silos may require separate technologies and methods. Here are several of the categories and approaches that need to be considered in building a more automated infrastructure:
Private Cloud & Datacenter
A private cloud is a datacenter built with current cloud technologies that runs on-premises or is hosted and managed by an organization or an enterprise itself, rather than in a public cloud.
Public Cloud
A public cloud is usually a distributed array of cloud resources and infrastructure run by a large platform-as-a-service (PaaS), infrastructure-as-a-service (IaaS), or software-as-a-service (SaaS) company, providing services to other organizations.
Multicloud
Enterprises might need services or resources from multiple IaaS or PaaS services, such as Amazon Web Services (AWS), Microsoft Azure, or Google Cloud Platform (GCP). In this case they need to connect their networking infrastructure to multiple public cloud infrastructures, all of which have specific requirements.
Hybrid Cloud
When enterprises build distributed applications that share resources on both private and public cloud infrastructure, it is referred to as hybrid cloud.
Service Provider Infrastructure & NaaS
Service providers – including communications service providers, cloud providers, and datacenter providers – supply global infrastructure and networking services to connect regions, points of presence (PoPs), and communications services such as 5G or networking. Network-as-a-service (NaaS) is a model in which networking or communications services can be purchased on demand – for example, using dedicated Internet access (DIA) or Ethernet services to connect datacenter PoPs or public cloud onramps.
What Does AI Mean for Infrastructure Exactly?
As modern applications become more distributed and infrastructure becomes more ephemeral, the need for on-demand infrastructure will only increase. Infrastructure is becoming more diverse and complex, with a combination of cloud infrastructure, AI infrastructure, datacenters, and traditional enterprise infrastructure all being needed to provide data and connectivity to enable AI applications. Organizations are considering using hybrid and multicloud approaches to enable many forms of cloud services to provide better flexibility, greater scale, and lower cloud costs.
But it’s not going to be easy. The scale of data, connectivity, and storage is vast. As organizations seek more flexible and affordable cloud platforms for managing data and applications in both on-premises and public cloud environments, the demand for network automation technologies to connect diverse resources is greater than ever before.
Enterprise and service provider practitioners alike will need to architect agentic orchestration – building a system in which AI agents are empowered to detect, decide, and act autonomously. Legacy tools that rely on dashboards, alerts, and manual interventions will no longer be sufficient.
To achieve these goals, organizations need to standardize observability, AIOps tools, and agentic data to create structured data systems to drive AI. Driving these systems into multivendor environments will be especially important, as there is a risk that proprietary platform fragmentation jeopardizes automation in hybrid environments.
More importantly, organizations will need a centralized system of observing and monitoring agentic AI and infrastructure orchestration. They will need a platform that provides safety guardrails and governance.
Goals & ROI for Hybrid Infrastructure Automation
The world is moving fast. Clouds, datacenters, and enterprise applications are expanding as never before. The instances of hybrid environments are increasing. This puts new demands on infrastructure automation.
Here are some of the goals for autonomous network and infrastructure automation:
Building Autonomy & Proactive Remediation
Agentic AI systems autonomously monitor, optimize, and manage cloud infrastructure, including provisioning, scaling, performance tuning, and issue remediation – all without human intervention. These systems use advanced analytics and real-time monitoring to predict failures and optimize resource usage, often preventing incidents before they escalate.
Contextual Decision-Making
Instead of executing fixed workflows, AI agents can add the capability to assess situations, set goals, reason about context, and take actions in line with organizational policy. This can speed detection and response, support continuous optimization (cost, performance, compliance), and drive operational resilience.
Automation at Scale
In modern IT environments – especially those using hybrid or multicloud architectures – the volume and velocity of changes can overwhelm traditional automation. Agentic AI can orchestrate dozens of specialized agents to help remediate or prevent incidents, reducing errors and operational overhead.
Security
Although agentic AI can introduce cybersecurity threats of its own, agentic AI is used broadly to autonomously detect threats, enforce security policies, and respond to incidents in real time, providing a powerful tool to improve the security posture.
In this report, well take a deeper look at these trends as well as how the practitioner and vendor ecosystem is responding. Our input includes presentations from dozens of practitioners, as well as interviews with key networking experts and vendors.
Trends in Network Automation
The topic of network automation was covered in detail at two of the AutoCon events, including in Prague last May and more recently in Austin in November. AutoCon, hosted by the Network Automation Forum (NAF), is a gathering of global network practitioners and operators.
In a keynote at AutoCon 3, Claudia de Luna, Advanced Technical Consultant, Networking and Automation, with Enterprise Infrastructure Acceleration, pointed out that it’s not possible to jump into automated operations without laying proper groundwork. She pointed out that many IT domains, including cloud, have moved to controller-based architectures, but networking remains fragmented. The danger to an enterprise is that they look to implement automation without asking fundamental questions.
In a summary of AutoCon, Chris Grundemann, one of the AutoCon organizers and a chairperson, offered this great summary of the some of the findings from speakers at the event. Some of the key themes included:
- Trust and User Experience (Damien Garros, Cofounder and CEO of OpsMill): “I truly believe that trust is actually one of the main issues.” Automation fails when builders create “black boxes” that work for them but inspire no confidence in users.
- Foundation Problems (Claudia de Luna): “We went from network automation with a little ‘n’ and little ‘a’ to Network Automation with capital letters, but really, what we’ve been talking about all this time is software development.” Many initiatives fail because they skip foundational design work.
- Tool Limitations (Josh Saul, Network Product Leader, BE Networks): “We haven’t had good tools. We’ve had rough knives, we’ve had hammers when we needed screwdrivers.”
- Skills and Learning Barriers (Emre Cinar): Junior engineers face overwhelming choices and limited guidance, while the community often makes assumptions about baseline knowledge.
- Enterprise Realities (Robert Blake): “Perfect is the enemy of good enough.”-Organizations get paralyzed trying to solve every edge case instead of making incremental progress.
We recently interviewed Greg Freeman, VP, Network and Customer Transformation, for Lumen, who told us that the world is moving toward using APIs and software-defined networks to provide better automation and orchestration of networks. He said the process can be slowed by large installed bases of legacy equipment, but the company is making progress using automation and orchestration tools from the likes of Ansible (IBM), Itential, and Selector.
We have equipment from 30 years ago. We want to maximize our investment. A few years ago, we wanted 80% of all our configuration changes to be machine-to-machine changes by 2025. We are exceeding that goal. Itential is helping us do that. We’re early in our journey with AI agents. We are seeing value, but AI is non-deterministic, and we need to make it deterministic. I’ve gotten great results from deterministic workflows with Itential.
Greg Freeman | Vice President Network and Customer Transformation – Lumen
Challenges for Hybrid Infrastructure Automation
Some of these quotes from AutoCon illustrate the increased challenge of dealing with complex hybrid infrastructure automation. We have been presented with the view that if we sprinkle a little agentic AI into the infrastructure, it will all be fixed. But in practice, this is far from the truth. Practitioners regularly express frustration at the growing complexity and competing goals of specific tools, vendors, and approaches.
There’s no single “infrastructure automation” magic bullet. To achieve autonomous infrastructure, a lot of groundwork must be built, and organizations need to be prepared for a different mindset.
In looking for solutions to automate infrastructure, practitioners have many challenges and obstacles to overcome. Some of these include:
- Fragmented Infrastructure.
Use of diverse tools and infrastructure platforms, including those from many different vendors. Implementing automation requires consistency across environments and platforms, with integrations using APIs and data platforms. - Cultural Divisions.
One of the largest challenges in organizations are divisions of the operational silos, such as the infrastructure, DevOps, and security organizations. The platform engineering movement is seeking to align these divisions, but the truth is that many organizations use different tools and approaches. Networking is often the most difficult piece of the puzzle to complete. - Design & Operational Vision.
Because of the complexity in building a consistent architecture for automation, it requires everybody getting on the same page. This needs to come from the top down of the technical organization, from a CTO or CIO.
Tools & Architectures for Building Autonomous Networks
How does an organization get on the path to true agentic orchestration? This should be defined as a way to extend automation control across compute, network, and storage resources in multicloud and hybrid cloud environments.
Let’s take a closer look at specific areas that need to be addressed so that AIOps and autonomous networks can be implemented.
Key Technology Components for Infrastructure Automation
As infrastructure automation continues to accelerate, look for the continued convergence of observability, AIOps, and NetDevOps functions. In 2025, several advancements are taking shape across the landscape. Let’s dive into these varied buckets.
Infrastructure Monitoring.
AIOps and automation tools may be positioned to detect anomalies that indicate the imminent failure of applications, using AI/ML to diagnose the issue and offer remediation suggestions. AIOps can also be applied to networking infrastructure in a similar way; for instance, applying intelligence to determine the source and remedy for a failed connection.
Network & Cloud Observability.
You can’t automate what you can’t see. The field of observability is growing, but there are of course many components of observability, including log observability and analytics, applications monitoring, and network telemetry tools. Network observability is a key building block to network automation. It can also use AI to correlate logs, traces, and metrics across all layers, enabling predictive operations and rapid anomaly detection. To deliver true AIOps, end-to-end visibility is needed with as much data and telemetry as possible. This holistic approach improves system stability and reduces downtime. This requires gathering, evaluating, and displaying data from those components and getting insights into the health, effectiveness, and security of the network.
In this area, we have been watching some innovative startups mentioned by end users as providing new use cases for network observability and connecting this to automation. In one example of specialized tools, NetBox Cloud provides a programming interface to network functions such as IP address management, datacenter resources management, and other networking tasks. It also enables automated testing of new networking resources as well as monitoring. It’s been embraced by large infrastructure operators such as GPU cloud CoreWeave.
Alkira is integrating automation and AI into its cloud networking platform, which delivers multicloud and hybrid connectivity through a fully managed, cloud-hosted fabric. Its approach emphasizes intent-driven automation, abstracted infrastructure, and increasing use of AI-driven assistants to simplify complex network operations. With Cloud Exchange Points (CXPs) worldwide, enterprises can seamlessly connect users, sites, and clouds while integrating SD-WAN, firewalls, and SASE.
Selector AI combines observability, AI, and root-cause analysis to deliver correlated, contextual insights across network, infrastructure, and application domains. Its purpose-built Network Language Model (NLM) and Copilot interface enable natural language interactions, event correlation, and integration with existing tools, reducing tools sprawl and accelerating resolution. Selector’s platform integrates tightly with Itential to enable closed-loop automation through APIdriven workflows that support policy enforcement and automated error correction.
Gigamon recently announced Gigamon Insights, an agentic AI application integrated with AWS, Elastic, and Splunk (amongst others) to help security, IT, and NetOps teams detect threats, troubleshoot network and application performance issues, and close compliance gaps faster, at scale.
LogicMonitor is an intriguing private company with a large portfolio of observability and automation tools. Its recently announced acquisition of Catchpoint expands the company’s capability to monitor Internet performance and user experience. Catchpoint operates more than 3,000 probes worldwide that track outages and network disruptions in real time. LogicMonitor AIOps capabilities can automatically detect anomalies, predict issues, and correlate root causes without human intervention. Futuriom sees LogicMonitor as uniquely positioned to provide endto- end monitoring, observability, and AIOps across hybrid environments.
Riverbed has been pursuing a low-code and no-code automation strategy using “runbooks.” Runbooks allow IT teams to build workflows that automate common tasks. These runbooks are driven by AI and machine learning (ML) analysis of network telemetry data, providing intelligence for tasks like incident response, troubleshooting, and forensic investigations.
Vendors to Watch in Network Observability
Alkira, Aviatrix, Cisco, Chronosphere, Dynatrace, Datadog, Gigamon, Kentik, LogicMonitor, NetBox, Riverbed, Selector AI
Open-source projects in networking and observability continue to gain steam as helpful tools across the spectrum of observability and autonomous infrastructure needs. A growing area of interest is at the cloud-native layer – for example, using eBPF to collect and observe connectivity at the operating system layer.
Projects to Watch
OpenTelemetry, Prometheus, Grafana, Cilium, and eBPF continue to gain traction, providing the foundation for observability and automation in cloud-native environments. OpenTofu, which is used by many platform engineering and IaC vendors such as Spacelift and Env0, is increasingly seen as a productive platform engineering and infrastructure tool for hybrid environments.
Scripting and coding tools. Network automation projects often start as scripting code in Python but then progress as teams conclude they need more. Scripting tools such as Ansible and Puppet are also popular for automating repetitive tasks such as configuration. Practitioners might adopt other tools, languages, and platforms, such as JSON, YAML, or GitHub API to build their own DIY automation suite. However, managing network automation with a roll-your-own strategy can involve many manual tasks and incur management overhead. And typically it delivers automation of specific pockets of the networks (SD-WAN, routers, etc.) but doesn’t provide orchestration across network boundaries and environments.
IaC tools. A set of automation and orchestration tools known as IaC are important to enabling platform engineering and infrastructure automation. IaC involves managing IT infrastructure such as servers, networks, and storage through machine-readable definition files, rather than manual processes. This allows teams to automate the provisioning, configuration, and deployment of infrastructure, making it repeatable, consistent, and scalable. In theory, IaC can be integrated with machine learning and AI to provide context-aware infrastructure provisioning and scaling. This shift supports more intelligent, automated responses to fluctuating demand.
IBM’s Ansible and HashiCorp’s Terraform (also IBM) are still go-to commercial tools for enabling infrastructure automation, though they are typically optimized for specific environments and require integrations to extend to hybrid environments. Terraform is often successful in cloud networking environments (AWS, GCP, etc.) or with common network orchestration integrations, such as Cisco ACI. But to use Terraform correctly, it typically needs to have APIs built into the platform it is accessing. It can get messy after that, as engineers create separate API operations. This can lead to a path of potential failures in complex environments.
Ansible uses Python to build playbooks that can be used to automate the collection of inventory and device configuration data. Ansible can be written to interoperate with other tools and languages such as Python or YAML. For example, an Ansible playbook could be used to automate the configuration of devices such as routers or firewalls if they need to be configured at scale in a standard way. Ansible can be good for repeatable tasks, such as configuring devices with bulk APIs, but it is not necessarily an orchestrator for complex, hybrid environments.
In addition to IBM’s HashiCorp and Terraform, some additional players are emerging in the market, such as Pulumi and Spacelift.
Vendors to Watch
IBM (including Red Hat, HashiCorp, and Ansible), Itential, Pulumi, Spacelift
Platform engineering is maturing to help provide automated platforms for deploying, monitoring, and operating infrastructure. This can be helpful by limiting choices and streamlining infrastructure and compliance. This enables organizations to deliver standardized, self-service tools to DevOps and business teams, reducing operational friction.
Vendors to Watch
IBM (including Red Hat and Hashicorp), Itential, Pulumi, Spacelift, System Initiative
Agentic AI should be looked at as a way to coordinate infrastructure automation and orchestration across many tools and platforms. The risk of agentic AI is when it is thought of too narrowly – for example, as providing orchestration in a single, closed vendor platform.
In the broad vision of agentic orchestration, AI agents continuously collect real-time data from various sources such as telemetry, logs, APIs, and sensors to build a comprehensive model of the system’s current state. This allows them to understand not just what is happening, but why it might be occurring. The AI agents could then reason, plan, act, and continuously learn from the environment, moving beyond traditional rule-based automation to create self-managing systems.
Many types of agentic AI systems are being added to infrastructure platforms by all of the major vendors. In the next section, we’ll provide more specific details about the largest vendor announcements.
Aviatrix is embedding agentic AI into its multicloud networking platform to automate traditionally manual, error-prone infrastructure tasks. It uses AI as an autonomous operations layer sitting above the Aviatrix data plane, leveraging the platform’s end-to-end visibility and control. Because Aviatrix owns and operates a secure data plane (Aviatrix Gateways), its AI has high-fidelity, vendoragnostic telemetry. Competitors relying solely on control-plane APIs lack this visibility.
BE Networks has added SensAI as a partner for network operations. It enables users to ask complex questions, providing multidimensional answers after assessing the current state of the network. The BE product lineup also includes Verity for intent-based automation and Satori for observability.
Itential’s intriguing new FlowAI product looks to help IaC and platform engineering professionals use agentic AI to orchestrate infrastructure across multivendor and hybrid environments, in addition to addressing security and compliance concerns with the agentic AI approach. This is an important new thrust in a world where many products are limited to specific vendor environments and operational silos – while ignoring the operational security and safety risks of agentic AI.
Vendors to Watch
Most major networking vendors, including Arista Networks, Aviatrix, Cisco, Extreme Networks, HPE/Juniper, Itential
MCP can help with network automation by acting as an intermediary server between AI agents and various network devices, providing access to data, interfaces, and APIs. Any vendor or provider can set up an MCP server, which helps abstract the complexity of the environment and provides details about CLI, APIs, and specific vendor implementations, enabling AI agents to interact with commands. We expect to see an acceleration of MCP announcements.
Key vendors have already made MCP announcements. In one example, Itential recently announced its own MCP server, as well as a management platform for MCP services in FlowAI. Peter Sprygada, Chief Architect with Itential, believes MCP offers a unique proposition for network automation by enabling organizations to build AI-native network operations and workflows. HashiCorp said its MCP server enables agent-based Terraform workflows.
Nokia has for many years had one of the leading automation platforms for service provider and enterprise networks with Network Services Platform (NSP). Recently at a Nokia event, I learned that it is building an MCP server for NSP. NSP is a comprehensive platform for managing and automating network services across IP/MPLS, carrier Ethernet, optical, and microwave transport networks, supporting multivendor equipment and unified management.
Recent Agentic AI Strategies from Leading Networking Vendors
The major networking and infrastructure vendors have ramped up their agentic AI stories to further drive network automation and AIOps. The past few months have featured numerous networking announcements from the largest networking vendors, including Arista Networks, Cisco, Extreme Networks, and Juniper (now owned by HPE), among others. Here are some of the recent developments we’ve been tracking or have been briefed on:
Arista Networks recently unveiled the EOS Smart AI Suite, targeting high-performance AI workloads and networking for large ML clusters. Features include Cluster Load Balancing to maximize AI workload efficiency and ensure low-latency flows, and CloudVision Universal Network Observability (CV UNO) to provide end-to-end, real-time AI job-centric observability and troubleshooting.
As mentioned above, IBM’s HashiCorp recently launched MCP servers for Terraform and Vault, which connect LLMs and trusted automation for infrastructure provisioning and secure secrets management using natural language and agentic interfaces. It’s also embarked on projects with IBM to create a unified control plane for automated infrastructure, including hybrid environments.
Cisco recently launched an AgenticOps model unifying real-time telemetry, automation, and a Deep Network Model. Some of the features include products and services such as AI Canvas, AI Assistant, and AI Agents. Cisco says this will support proactive, predictive, and autonomous network actions, with the capability to process 170,000+ alerts per hour at millisecond speed.
Cisco has also worked toward better integration of its platform, with a unified management platform for Meraki and Catalyst devices in the cloud.
Extreme Networks recently announced the availability of Extreme Platform ONE, which it claimed is the industry’s first fully integrated platform for conversational, multimodal, and agentic AI in networking. Extreme says the platform will deliver as much as 90% less manual work, 98% faster resolution, and unified network visualization. AI agents act as always-on “virtual network engineers,” supporting proactive and predictive management.
HPE, which has many networking assets, including Juniper, Aruba, and integrated SASE platform Silver Peak, recently announced a unified secure networking portfolio post-Juniper acquisition, leveraging agentic AI and new SASE copilots for autonomous policy enforcement, network observability, and proactive security across HPE Aruba and Juniper environments. It has also announced expanded support for cross-vendor observability and agentic assistants. The speed with which HPE and Juniper can bring together their AIOps strategy for networking will play a key role in the success of the mergers with Juniper.
Fresh off the finalized merger with HPE, Juniper rolled out agentic AI workflows in its Mist AIOps platform, enabling autonomous real-time troubleshooting across LAN, WAN, and datacenter environments using Marvis AI and new agentic features. It also introduced a generalized Large Experience Model (LEM) for experience-driven networking and digital twin capabilities.
Nokia recently announced a strategic partnership with Supermicro to help cloud providers, hyperscalers, enterprises, and CSPs deploy high-performance, AI-optimized datacenter networking solutions. This partnership has a significant automation component, combining Supermicro’s switching hardware with Nokia’s datacenter automation and network operating system. The companies will deliver an integrated datacenter networking solution built for AI, high-performance computing (HPC), and cloud environments. The combined solution includes Supermicro’s 800G Ethernet switching platforms integrated with Nokia’s Service Router Linux (SR Linux) Network Operating System (NOS) and Nokia’s Event-Driven Automation (EDA).
The following chart summarizes the approach of leading networking vendors that Futuriom is tracking in the community.
| Vendor | Approach and Environment | Differentiation | Risks |
|---|---|---|---|
| Arista Networks (CloudVision/EOS Automation) |
CloudVision and EOS Automation can be used in campus, enterprise, and cloud-scale networks. The CloudVision platform delivers "zero-touch network operations with consistent operations enterprise-wide." Automation is integrated with the Arista operating system, including real feedback with device telemetry. | The Arista platform excels when the network hardware and software are co-designed into the datacenter and network architecture. | Arista CloudVision is tied into the Arista ecosystem and could be seen as less vendor-agnostic than more open orchestration platforms. May not cover legacy heterogeneous multi-vendor environments. |
| Alkira | Through its Cloud Exchange Points, Alkira operates a fully managed overlay fabric that abstracts the underlying network across AWS, Azure, GCP, datacenters, and SD-WAN. Automation and AI act on this virtualized layer rather than on individual cloud primitives. | The platform provides a fully managed, cloud-based networking backbone. It automatically enforces segmentation, firewall insertion, service chaining, and traffic inspection across clouds reducing manual configuration needs. | Customers must adopt Alkira's global network cloud as the operational platform. Some enterprises might want more control over their own networking software and equipment, rather than fully relying on a cloud-based platform. Cloud operators may implement similar functionality over time. |
| Aviatrix | Aviatrix is embedding agentic AI into its multicloud networking platform to automate traditionally manual, error-prone infrastructure tasks. The core philosophy is to use AI as an autonomous operations layer sitting above the Aviatrix data plane, leveraging the platform's end-to-end visibility and control. | Aviatrix focuses its approach on multicloud environments, enabling users to build end-to-end network control across multiple environments, with complete visibility, security, and control of the data plane. | Implementing closed-loop automation in hybrid environments requires tight integration across gateways, controllers, and AI inference systems. Large cloud operators are starting to offer multicloud features, providing potential future competition. |
| BE Networks (Verity IBN Orchestration and SensAI) |
A modern entrant for vendor-agnostic, intent-based networking orchestration aimed at campus/edge and cloud-connected. | Good for enterprises looking for next-gen intent-based network orchestration, especially in edge/campus + cloud domains. SensAI provides agentic AI operations. | This smaller and newer entrant is flying a bit below the radar and doesn't have the reputation of more established vendors. |
| Cisco Systems (Crosswork/NSO) |
Cisco software and hardware is widely deployed, with a large footprint and strong brand. Cisco's automation/orchestration offerings support intent-based machine-learning-driven closed-loop automation for networking and cloud. | Best for large enterprises already heavily invested in Cisco gear. Can help drive broad automation across WAN, branch, datacenter, and cloud. | Because of Cisco's large footprint and scale, integration and new product developments sometimes come at a slower pace. It's convenient for customers with large Cisco footprints, but it may lean toward vendor lock-in. |
| Hewlett Packard Enterprise (HPE)/ Juniper Networking Automation | The new combination of HPE and Juniper offers a variety of automation/orchestration offerings covering campus, wireless, and datacenter. The Mist AIOps platform was a large component of the Juniper acquisition. | HPE and Juniper offer a large footprint and strong support for customers that are tied into that ecosystem. | Customers may have to await key decisions about integration, such as for Aruba campus wireless and Mist solutions. Like Cisco, solutions may push toward vendor lock-in. |
| Itential | A strong vendor-agnostic network automation and orchestration platform; supports multivendor, multi-domain (WAN, cloud, hybrid). Itential also has strong partnerships with leading networking vendors. | Good fit for enterprises with heterogeneous network estate (various vendors + cloud + SD-WAN) wanting a unified automation/orchestration layer. | Enterprise scale may require integration effort; offerings may be less "plug-and-play" compared to hardware-vendor integrated stacks. |
| Red Hat (Ansible Automation Platform) |
Red Hat is strong in enterprise automation, with an emphasis on hybrid cloud and network-device automation. It has great acceptability for device configuration, network modules, cloud automation. | Best for enterprises that already have automation practice, want vendor-agnostic, multi-cloud + on-prem network orchestration. | Ansible requires disciplined automation management; network-specific modules might lag compared to networking vendors. Agentic AI could challenge the Ansible business case. |
Network Orchestration & AIOps: Selected Industry Case Studies
Futuriom regularly examines customer deployments to look at how cutting-edge technology is being deployed in the real world. Let’s look at some activity in selected verticals with some specific customer examples.
In several case studies analyzed by Futuriom, Extreme Networks users cited ROI gains from its “one network, one cloud” operational model. Customers cited ROI in the range of 20%-50%, driven by automation, simplified operations, and improved security and maintenance. They achieved these savings with an integrated network fabric across wired, wireless, SD-WAN, campus, and datacenter, giving organizations a single control plane, consistent operations model, and reduced fragmentation due to legacy network silos. In one example, San Diego Community College District (SDCCD) – one of the two deployments studied by Futuriom – saw as much as a 50% reduction in network-operations time after moving to Extreme’s cloud-networking solution.
Fiserv and Itential. Financial services is a hot area for adoption of autonomous infrastructure because of the scale and demanding time requirements of the industry. Network monitoring and agentic AI can not only help monitor security and compliance in realtime, but it can enable infrastructure managers to respond to enormous network demands with orchestration.
In a presentation with Itential and Packet Pushers, Fiserv recently showed how it could scale network automation by orchestrating multiple automations across a distributed, multivendor environment. The key was to standardize tools and data to deliver network services, an approach we have grouped with platform engineering, whereby an organization can build a specific automation platform for its constituency. Fiserv partnered with Itential to lower the barrier to entry for automation and accelerate the way they deliver services, without compromising on quality.
In another example, Southern California Edison (SCE) said they needed a stronger strategy to deliver operational consistency, automation, and data-driven decision-making. To enable this, SCE implemented a centralized, vendor-agnostic automation and orchestration framework to support everything from network refresh projects to AI-powered incident response. “These efforts are not only improving day-to-day reliability – they are laying the foundation for long-term scalability, cybersecurity, and intelligent grid operations,” wrote SCE in a recent whitepaper on the project. Itential and NetBox were the two primary vendors included in the project, with Itential acting as the orchestration platform for infrastructure automation and NetBox acting as the network asset management platform.
Craft retailer Michaels used Alkira’s cloud-native network-as-a-service (NaaS) to re-architect its retail network across all its stores. Specifically, they connected around 1,400 stores across the U.S. and Canada to a new network backbone using Alkira. The entire deployment was completed within six weeks, during peak season. Michaels’ technical staff said the move has reduced complexity and operational overhead with centralized network provisioning and management. It has also reduced network costs with less on-site hardware and bandwidth costs, leveraging cloud backbone connectivity instead of more expensive connections to datacenters using MPLS.
As a large hospitality company, Vail Resorts manages a complex, global network infrastructure spanning resorts, datacenters, cloud environments, and retail locations. Managing a network of such scale, with environments ranging from on-premise datacenters to multicloud platforms like Azure and GCP manually would be slow and inconsistent. Ensuring that all network deployments are secure and compliant requires a programmatic approach. Vail Resorts adopted IaC principles, using network automation such as Terraform. They also integrated CI/CD pipelines to ensure consistent, policy-compliant deployments across their hybrid environment. The IaC approach allows Vail Resorts to standardize its network configurations, manage its hybrid cloud connectivity more efficiently, and accelerate deployments. This enables the company to better support its strategic initiatives and maintain a secure network.
Adoption Path & AIOps and Agentic Orchestration: Considerations for AI Orchestration Platforms & Governance
The picture we have painted for automated infrastructure operations, AIOps, and agentic orchestration might seem a little complicated and overwhelming. Indeed, it can be. That’s why it requires setting a strategic path of building automation platforms over time.
An automation roadmap could emphasize building a governed and standardized infrastructure before introducing AI. Key steps might include standardizing workflows, embedding security and policy into orchestration, and using AI-driven intelligence to enhance, but not replace, the orchestration layer, enabling closed-loop automation through integrating MCP or related API governance architectures.
Governance & Safety Considerations
For professionals charged with infrastructure, there are many implications of the advance of agentic AI and AIOps. All agentic AI and orchestration are based on data. The big question is: How do you manage the consistency, safety, and governance of data and agentic AI operations?
The data that will be used for AI will be everywhere, including multiple clouds and private datacenters. This will pose a challenge for many organizations.
Full agentic stacks: Agentic AI agents are expected to move networking from reactive incident response to network optimization and remediation across hybrid environments. But to do so effectively, you must have a complete suite of integrated network observability and monitoring tools in place, and organizations need to have full control of how models are collecting and being fed data, along with policy guardrails.
Staffing and skills: This may be the least mentioned but most consistent challenge in network automation. Programming network and infrastructure automation and orchestration requires highlevel tools and specialized skills. Many organizations struggle with staffing in this area and may require the help of third-party professional services teams.
Multi-vendor ecosystem: Another leading challenge continues to be architectural complexity, especially in multivendor environments. Many of the leading platforms may not emphasize open architectures and API extensibility, risking putting customers into a “walled garden” approach. Futuriom believes in open and multivendor systems, which often requires detailed integration across network, cloud, and security domains.
Security and governance: The rise of agentic AI in production networks brings new focus on policy enforcement, compliance, and controls to prevent unintended or unauthorized actions by autonomous agents.
Cost management: Before implementing agentic AI or AIOps, organizations need to conduct a thorough analysis of how this will impact costs, including the costs of all licenses, devices, and data. Observability and log data can spiral rapidly, so you need to know how this will affect your cost structure.
Steps in Building the Foundation
In my conversations with many leading practitioners and vendors, the most sensible thing I have heard to is to take the process of autonomous infrastructure operations step-by-step. Currently, most organizations are filled with manual workflows and data dependencies that will have to be converted to autonomous operations over time. This requires a looking at each step in the operations along the way.
Some key approaches might include the following:
Standardize and govern: Organizations must first standardize manual steps and build guardrails into their existing automation and orchestration. Without a strong, governed foundation, enterprises will struggle to trust AI in the future.
Integrate across silos: The transition from task-centric automation to full lifecycle orchestration might need coordination of integration tools like RPA, observability data, APIs, and IaC. Full infrastructure automation is an integrated strategy.
Embed policy and compliance: AI-driven changes must be governed by security policies, approvals, and compliance standards. Building an orchestration platform can provide the necessary layer to enforce these rules and ensure AI actions are safe and auditable.
Build infrastructure for intelligent agents: The ultimate goal for autonomous operations is a programmable, governed, and transparent infrastructure that can be consumed not only by developers but also by intelligent agents.
Conclusion: Autonomous Infrastructure Gains Momentum
Futuriom believes that enterprise adoption of infrastructure automation and network automation is accelerating, as companies recognize that manually managing ever-more-complex cloud environments is not sustainable. However, significant barriers to adoption exist, including hybrid complexity, multivendor environments, skills gaps, and security. The road to network AIOps and full-service orchestration is not an easy one, with many moving parts.
Most of the customer deployment case studies we have seen involve some combination of end-to-end network observability, orchestration platforms, and some form of IaC, involving multiple vendors. For individual vendors, the rapid emergence of AIOps and agentic AI technology signals a transformation in how network operations are executed. It will separate the efficacy and adoption of enterprise, datacenter, and cloud networking platforms going forward. However, not all vendors are focused on delivering orchestration in multivendor environments, which is a key consideration for many practitioners.
Autonomous networking reflects a broader trend: the rise of AI-augmented automation systems that integrate observability, configuration management, execution, and policy into a single, programmable control plane. But to achieve true closed-loop automation, many pieces need to fall into place. Organizations need to standardize infrastructure data, moving from vendor-specific CLI outputs into structured formats that AI systems can understand. This foundation is critical to enabling effective AI decision-making, policy enforcement, and closed-loop automation across domains.
The most important industry trend to watch is whether networking environments can coalesce around common formats and diverse APIs to build powerful orchestration platforms, or whether specific vendors try to push it down the road to proprietary platforms. The diverse nature of most installations means that many organizations are operating in hybrid and multivendor environments, requiring platforms and tools that can interoperate across many vendor platforms.
About Futuriom
Founded in 2017, Futuriom is the premier research and analysis community focused on next-generation technologies that will have a profound impact on the world. We analyze the companies and technologies that we think will provide the most growth over the next decade, including artificial intelligence, Internet of Things (IoT), cloud computing, and cybersecurity. Futuriom is a subsidiary of Rayno Media Inc. www.futuriom.com
Get the full Building Autonomous Infrastructure with Observability, Orchestration and AIOps report.
Appendix: Autonomous Infrastructure Leaders to Watch
Arista Networks (NYSE: ANET)
Arista is leading networking equipment and software provider that is integrating AIOps through its Arista AVA (Autonomous Virtual Assistant) and CloudVision platforms to automate network management and ensure high performance, especially for demanding AI workloads. Some of the capabilities include data collection & analysis; proactive problem solving; automated root-cause analysis; gen AI assistance; and optimization of AI workloads. This approach can be used by customers to transform IT operations from a reactive to a proactive, predictive, and increasingly autonomous model.
Cisco (AppDynamics, Splunk, and ThousandEyes) [Nasdaq: CSCO]
Cisco has a large observability portfolio with the successive acquisitions of AppDynamics, ThousandEyes, and Splunk. After Cisco recently acquired Splunk, Cisco and Spunk announced new integrations, including a unified observability experience for joint customers and the introduction of Splunk Log Observer Connect for Cisco AppDynamics and Cisco AppDynamics integration with Splunk IT Service Intelligence (ITSI). Cisco believes that its visibility into the network and any environment, coupled with Splunk’s industry-defining log analytics and cloud-native observability capabilities, will enable customers to instrument their business and reduce blind spots. Cisco Crosswork Network Automation offers a “multi-agentic AI framework,” meaning customers can build, deploy, and link multiple agents across network infrastructure.
Aviatrix
For enterprises struggling to secure cloud workloads, Aviatrix offers a single solution for pervasive cloud security. Where current cybersecurity approaches focus on securing entry points to a trusted space, Aviatrix Cloud Native Security Fabric (CNSF) delivers runtime security and enforcement within the cloud application infrastructure itself – closing gaps between existing solutions and helping organizations regain visibility and control. Aviatrix ensures security, cloud, and networking teams are empowering developer velocity, AI, serverless, and what’s next.
Dynatrace (NYSE: DT)
Dynatrace’s unified platform combines broad and deep observability and continuous runtime application security with its Davis AI to provide answers and intelligent automation from data at an enormous scale. This enables innovators to modernize and automate cloud operations, deliver software faster and more securely, and ensure flawless digital experiences. Dynatrace empowers organizations to manage complex, cloud-native applications more efficiently, enabling a focus on rapid feature development and accelerated innovation cycles. Dynatrace processes data from diverse sources, including its OneAgent, open-source collectors, and cloud service providers. Using AI-driven analytics, Dynatrace boosts cloud-native automation and reliability, uniting platform operations, DevOps, security, and business teams. Dynatrace is headquartered in Waltham, Mass., and trades on the NYSE as DT.
F5
F5 is a multicloud application services and security company committed to bringing a better digital world to life. F5 partners with the world’s largest, most advanced organizations to secure and optimize apps and APIs anywhere – on premises, in the cloud, or at the edge. F5 enables organizations to provide exceptional, secure digital experiences for their customers and continuously stay ahead of threats.
Gigamon
Gigamon offers an observability pipeline that efficiently delivers network-derived intelligence to your cloud, security, and observability tools, helping organizations eliminate security blind spots, reduce tool costs, and better secure and manage hybrid cloud infrastructure. Gigamon goes beyond security and observability log-based approaches by extracting real-time network intelligence derived from packets, flows, and application metadata to deliver defense-in-depth and complete performance management. Gigamon has served more than 4,000 customers worldwide, including over 80 percent of Fortune 100 enterprises, 9 of the 10 largest mobile network providers, and hundreds of governments and educational organizations worldwide.
Itential
Itential is the infrastructure and network orchestration platform for the AI era. It’s cloud-native solution unifies lifecycle operations across hybrid network and cloud environments, empowering enterprises to automate everything from configuration management and compliance to service delivery. With support for both low-code orchestration and high-code extensibility, Itential enables teams to build, adapt, and scale automations faster. Through agentic orchestration via MCP and intelligent workflows, Itential connects IT systems, CI/CD pipelines, and network infrastructure to deliver secure, end-to-end orchestration that’s ready for the next generation of digital and AI-driven operations.
Kentik
Kentik is the network observability company. The platform is used by the network front line, whether digital business, corporate IT, or service provider. Network professionals turn to the Kentik Network Observability Platform to plan, run, and fix any network, relying on granularity, AI-driven insights, and fast search. Kentik makes sense of network, cloud, host and container flow, internet routing, performance tests, and network metrics. It shows network pros what they need to know about their network performance, health, and security to make their business-critical services shine. Kentik has raised $105 million in funding and is based in San Francisco.
LogicMonitor
LogicMonitor can help accelerate infrastructure automation with its comprehensive observability platform. It automates data collection and normalization from diverse environments (cloud, hybrid, multicloud), eliminating manual configuration. The platform utilizes AIOps capabilities to automatically detect anomalies, predict issues, and correlate root causes without human intervention. Key features include automated resource discovery and configuration for monitoring dynamic environments. Prebuilt monitoring templates and data sources ensure rapid deployment and configuration. By automating alert routing, contextual data gathering, and integrating seamlessly with ITSM and automation tools, LogicMonitor streamlines incident response, allowing teams to proactively remediate issues with less manual effort. LogicMonitor is owned by Vista Equity Partners.
NetBox
NetBox Labs makes sense of complex networks and infrastructure. As the commercial steward of open source NetBox, the most popular platform for operating, automating, understanding, and securing complex networks and infrastructure, NetBox Labs delivers a world-class portfolio of open, composable products. NetBox Labs fosters and invests in a vibrant community of tens of thousands of network and infrastructure professionals. Top companies like ARM, Cisco, Constant Contact, CoreWeave, J.P. Morgan, Kaiser Permanente, and Riot Games trust NetBox Labs to understand, operate, and transform their critical infrastructure.
Nokia
Nokia provides observability and automation capabilities focused on creating autonomous networks using a “sense, think, act” framework. Observability tools, such as its Autonomous Network Fabric, offer 360-degree, multivendor visibility across 5G, IP, and optical networks. This data, combined with analytics and AI from Nokia’s Data Suite, helps detect issues, predict bottlenecks, and optimize network performance. Automation solutions, including the Network Services Platform and Event-Driven Automation, enable closed-loop, intent-based network operations to ensure reliability and deliver new services.
Pulumi
Pulumi is a single unified platform for everything you run in the cloud. It delivers productive automation, automatic security, and intelligent management acrosshundreds of clouds. Powered by the industry’s leading open-source infrastructure-as-code, programming languages, and generative AIs, developers, infrastructure experts, and security teams collaborate seamlessly to ship faster with built-in confidence and security. Using Pulumi, organizations get to market faster, with less risk and more control, turning the cloud into a competitive advantage.
Selector AI
Selector AI helps organizations prevent and resolve network issues faster by combining observability, AI, and automation. Trusted by enterprises such as NBC, Fiserv, Lumen, and AT&T, Selector delivers instant, correlated insights across metrics, logs, events, and topology, giving teams clear context for action across hybrid and multi-domain environments. Purpose-built AI and network-trained large language models (LLMs), trained and validated in some of the world’s most complex and demanding networks, accelerate root cause analysis and decision-making. Selector is headquartered in Santa Clara, Calif., and has raised a total of $66 million.
Spacelift
Spacelift is an infrastructure orchestration platform that manages your entire infrastructure lifecycle – provisioning, configuration, and governance. Spacelift integrates with all your infrastructure tooling (e.g. Terraform, OpenTofu, Kubernetes, CloudFormation, Pulumi, Ansible) to provide a single integrated workflow so you can deliver secure, cost-effective, and resilient infrastructure, fast. By automating deployment and configuration, providing developer self-service, golden paths with guardrails, and an OPA policy engine, Spacelift empowers businesses to accelerate developer velocity while maintaining control and governance over their infrastructure.
