The rapid adoption of generative AI and agentic AI is spurring demand for AI infrastructure and the need for a rich ecosystem that can deliver the complete AI lifecycle. In this environment, a new breed of AI cloud providers—often referred to as neoclouds—has emerged as agile, technically sophisticated players building highly specialized AI cloud services to serve new markets and needs. At Cisco, we are dedicated to working with the full range of ecosystem participants. This includes the traditional hyperscalers, as well as these newer neocloud providers and foundational model builders. It also includes communications service providers, colocation providers, and sovereign cloud providers that augment AI cloud services by delivering the AI hosting, secure ubiquitous connectivity, and trusted infrastructure and data services enterprises need.
The AI infrastructure market: The surging frontier
The demand for AI infrastructure is growing at an exponential rate. Total global data center IT equipment spend over the five years between 2025 and 2030 is expected to reach as much as 4.7 trillion dollars.1 This encompasses servers, accelerators (XPUs), networking, optics, and storage. Today, it’s estimated that hyperscalers are responsible for over 60% of this infrastructure investment, while neoclouds are responsible for about 17%, which is expected to grow to over 30% over the next ten years.2 The other ecosystem members and enterprises themselves are becoming responsible for an increasing proportion of the AI infrastructure buildout as inferencing and agentic AI, sovereign cloud, and edge AI become more mainstream. According to McKinsey, demand for AI data center capacity will account for 70% of total data center demand by 2030 and is the main driver of a growing infrastructure deficit.3 Neoclouds are rapidly stepping up to fill the gap and are making big moves to drive new business models and technology innovations that focus on the readiness, resilience, differentiation, and profitability of their AI cloud service offerings.
Three approaches to AI cloud services
Neocloud providers are not a monolith. They utilize various business models to deliver AI infrastructure services that cater to different use cases and verticals. Consumption models are changing rapidly, from bare-metal offerings to full-stack hardware and software in addition to “as a service” offerings. We can categorize delivery and consumption models into multiple approaches that reflect diverse enterprise requirements for scalability, flexibility, cost, performance, and data sovereignty. Individual providers are choosing the approaches that best suit their business priorities and customer needs.
Three primary approaches have emerged:
- Dedicated cloud AI IaaS: In this model, enterprises commit to using dedicated AI infrastructure within a neocloud’s data center for a fixed term. The benefits of this approach include guaranteed capacity, better performance isolation, and often significant cost savings compared to on-demand pricing. It’s best suited for predictable and long-running AI workloads, such as large-scale model training or continuous inference tasks, and for organizations with consistent AI resource requirements or those that need exclusive access to compute resources.
- Public AI cloud services: With this model, neoclouds enable enterprise customers to access shared pools of AI-optimized computing resources, such as GPUs and tensor processing units (TPUs). Customers only pay for the resources they consume without long-term commitments. This “pay-as-you-go” approach offers maximum flexibility and scalability, making it ideal for experimental workloads, development, testing, or bursty AI tasks where demand is not consistent.
- Hybrid and edge AI IaaS: This model extends deploying and managing AI infrastructure across a combination of public cloud environments, on-premises or colocation data centers, and edge locations. It brings AI compute closer to where data is generated or consumed, addressing needs like low-latency processing, data sovereignty, and regulatory compliance. This model is particularly beneficial for customers that require real-time decision making, such as industrial internet of things (IoT), autonomous vehicles, or localized analytics.
Seven reasons neoclouds can count on Cisco for AI infrastructure
Regardless of the service model adopted, neocloud providers require a robust, scalable, and secure foundation. Many are finding that Cisco offers the partnering focus, comprehensive portfolio, and vision necessary to succeed. Here are seven key advantages to partnering with Cisco:
- Comprehensive, in-house AI infrastructure portfolio: Cisco offers the broadest homegrown AI infrastructure portfolio both within and between data centers. This includes integrated full-stack secure AI factory solutions, AI-optimized compute, AI networking, optics, and data center interconnect solutions, as well as security and observability for AI. Benefits for neoclouds include architectural flexibility, streamlined procurement, consistent operations and support, and a highly robust supply chain.
- Architectural and operational flexibility: With Cisco, neocloud providers have the flexibility to choose the reference architecture and operating environment that best meets their technical requirements. Whether they want to adhere to the NVIDIA Enterprise Reference Architecture (ERA), the NVIDIA Cloud Partner (NCP) reference architecture, Cisco Cloud Reference Architecture, or simply want to build their own—we provide the necessary flexibility and validated designs to accelerate deployment.
- Industry-leading AI networking fabrics: Cisco AI Networking delivers scalable front-end and back-end fabrics within data centers, together with data center interconnect and scale-across networks between data centers. Neoclouds can choose between the proven reliability of Cisco NX-OS for operational consistency, or SONiC for open networking standards.
- Expanded partnership with NVIDIA: Our expanded partnership with NVIDIA enables NVIDIA Spectrum-X technology integration on Cisco Silicon One–based switches and Cisco N9100 Series Switches that incorporate NVIDIA Spectrum-X Ethernet switch silicon—all unified under the Nexus operating model for seamless integration with Cisco Nexus Dashboard. Nexus is already well known within many neoclouds who have established it in their front-end computing networks.
- Powerful and diverse partner ecosystem: Cisco partners with a broad ecosystem to deliver the best solutions and value. This includes compute and accelerator partners like NVIDIA, AMD, and Intel; storage partners; software partners; liquid cooling partners; system integrators; and many others.
- End-to-end security and observability: Security and reliability are top concerns for enterprises deploying AI. Cisco Secure AI Factory with NVIDIA provides a full-stack solution with security integrated at every layer. With Splunk, Cisco provides Observability for AI for deep visibility into every layer of the environment, including GPU infrastructure, GenAI, LLM outputs, and AI applications. Cisco offers unmatched security and observability for AI deployments, helping neoclouds protect their infrastructure and data.
- Unified silicon architecture with Cisco Silicon One: Cisco Silicon One is a groundbreaking switching and routing silicon architecture. This single architecture is the foundation used across a broad range of networks, from back-end and front-end AI networks to scale across, data center interconnect, wide-area networks, and edge. It delivers advanced congestion control and load balancing for AI fabrics and is programmable to support emerging standards.
Building the future of AI cloud services with Cisco
Cisco is uniquely positioned to support neocloud providers’ AI infrastructure needs across all their service models. By providing cutting-edge technology—from Cisco Silicon One and NVIDIA Spectrum-X silicon-based switching to Cisco UCS compute to integrated security and observability stacks—Cisco enables neoclouds to deliver scalable AI services with speed, reliability, and efficiency.
As the market continues to evolve, Cisco provides the expertise, go-to-market support, and solutions to help neoclouds accelerate time to service, maximize the efficiency of GPU clusters, and ultimately secure a foundational role in the burgeoning AI ecosystem.
1. The cost of compute: A $7 trillion race to scale data centers, McKinsey & Company, April 2025.
2. AI Infrastructure Survey, McKinsey, June 2025. (Internal Cisco survey)
3. AI power: Expanding data center capacity to meet growing demand, McKinsey & Company, October 2024.