Zeus Kerravala argues that NVIDIA’s low placement in the Q3 2026 Forrester Wave for Data Center Networking Solutions reflects a mismatch between the evaluation’s reported priorities and NVIDIA’s specialization in AI networking—not proof that the Wave is invalid or that NVIDIA is the wrong choice for every network. For buyers, the key question is whether a ranking measures the workloads, operating model, and management needs they actually have.
What Kerravala says the Forrester ranking gets wrong
In a September 4, 2026 opinion article, ZK Research founder and principal analyst Zeus Kerravala says NVIDIA landed at the bottom of the “Contenders” quadrant in the Q3 2026 Forrester Wave for Data Center Networking Solutions. He argues that the evaluation is oriented toward traditional enterprise networking, while NVIDIA’s data center networking is focused on high-throughput fabrics for large GPU clusters. Kerravala’s Network World article is the source for the placement and the score details below; the underlying Forrester scorecard was not available for independent confirmation.
As an Amazon Associate I earn from qualifying purchases.
Kerravala reports that NVIDIA received 1 out of 5 for Vision, Enterprise Management Types, Administrative Experience, and Switch Portfolio Scope, and a 5.0 for AI Infrastructure. He also says AI infrastructure support accounted for 2% of the overall evaluation weighting. These are his reported figures, not independently verified scores or weights.
His criticism is about fit: if a buyer’s main concern is broad enterprise networking, criteria such as management experience and portfolio scope may be relevant. If the project is an AI fabric for GPU clusters, a broad matrix may give too little weight to the workload-specific networking capabilities that matter most. That argument does not establish that the Wave is useless; it means buyers should inspect what it measures before treating its placement as a verdict on their use case.
#1 Best Overall
- Supports optical fiber cable to span longer distances and provided high data transmission rates between servers and network components
- 100 Gigabit Ethernet provides high bandwidth performance, ease of use and reliability for your network backbone
- Supports layer 3 switching for enhanced performance and usability
- Power Supply power source for dependable power supply
- Rail mounting is ideal for factory areas, in vehicles and marine applications
Why a general data center matrix may not answer an AI-fabric question
A network evaluation can reward breadth: support for general-purpose enterprise traffic, existing protocols, management integration, and a wide switch portfolio. Kerravala says those kinds of considerations feature in the challenged Wave. They can matter to organizations operating conventional enterprise data center networks, where compatibility with established systems and day-to-day administration are central requirements.
An AI fabric has a different job. Its value depends on how well networking supports the topology and traffic patterns of a particular cluster and AI workload. A buyer may need to assess throughput and latency alongside deployment, observability, lifecycle management, and coordination with compute, storage, software, and cloud or on-premises infrastructure. This is a practical checklist synthesized from Kerravala’s description and Forrester’s separate AI-infrastructure guidance; it is not a verified copy of the Q3 2026 networking Wave rubric.
Rank #2
- Flexible 10/25/40/100GbE Interface Speeds, Support Stacking; Achieve Reliable Links With 25G NVIDIA Mellanox NIC; CE, Reach, RoHS, FCC, WEEE Certified;
That distinction is important because a score is meaningful only in relation to the criteria and their weights. A product can be specialized rather than broad, and a ranking designed to compare broad enterprise networking portfolios may not resolve whether that specialization is suitable for a given AI cluster. Conversely, strong AI-networking credentials alone do not settle operational fit or the ability to integrate with an organization’s existing environment.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →What Forrester’s AI infrastructure work adds—and what it does not
Forrester’s own AI infrastructure materials offer relevant context, but they concern a separate evaluation and should not be conflated with the Q3 2026 data center networking Wave. In its Q4 2025 AI Infrastructure Wave announcement, Forrester describes assessing current offerings and strategy, including training and inference support, operational readiness—deployment, observability, and lifecycle management—and alignment across silicon, software, and cloud services. Forrester also cautions that the evaluation differed from its 2023 version, making year-over-year comparisons potentially misleading. Forrester’s announcement does not verify the criteria or weighting Kerravala attributes to the networking Wave.
Rank #3
- Wired connection speed up to 1000 Mbps
- 5 Gigabit Ethernet auto sensing ports
- Easy plug & play connection
- Quality of Service (QoS) traffic prioritization
- Meets EEE (Energy Efficient Ethernet) 802.3az standard
Forrester’s 2024 AI infrastructure overview defines these solutions as computer systems and cloud services designed to maximize AI workload performance across data preparation, training, and inference. Its practical advice is to match infrastructure to computational needs, recognize that workloads differ, and plan for integration with existing infrastructure management tools. The overview supports a workload-first buying approach, not a claim about how the separate networking report scored NVIDIA.
Forrester analysts’ commentary after GTC 2026 describes NVIDIA’s approach as increasingly system-level, spanning compute architectures, server, storage and networking reference designs, AI software, and data pipelines. The same analysts identify execution constraints such as operational readiness, power, cooling, geopolitics, and competition from custom and sovereign silicon. This is their interpretation of GTC, not an independent product test. Their GTC analysis is a reminder that an AI infrastructure decision reaches beyond the network switch itself.
Rank #4
- Full duplex IEEE 802.3x flow control and half duplex backpressure with intelligent port based congestion detection and broadcast rate control
- All ports support auto MDI/MDI X cable detection
- Non head of line blocking architecture
- QoS: Prioritize video and audio applications for better streaming (IEEE 802.1p)
- Power Save: Advanced power saving mode detects unused ports
How to evaluate networking for an AI data center
Start by writing down the project’s workload and operating requirements, then check whether an analyst comparison uses criteria that reflect them. Forrester describes its Wave format as criteria-based and says clients can use interactive tools to develop custom shortlists aligned with their priorities. Forrester’s Wave methodology page explains that customization, but it does not mean every published ranking automatically reflects a particular buyer’s needs.
Recommended Free Tools
For a general-purpose enterprise network
- Check support for the traffic types and protocols already in use.
- Assess how administration fits the organization’s current operations and management tools.
- Compare portfolio breadth and deployment fit against the actual network footprint.
- Use a broad enterprise networking evaluation when those are the capabilities the project needs to compare.
For an AI fabric
- Define the cluster topology and the training or inference workloads the network must support.
- Set workload-specific throughput and latency requirements rather than assuming one network profile fits every AI system.
- Assess deployment, observability, and lifecycle operations, including how the fabric will be managed in practice.
- Check integration across compute, storage, software, and cloud or on-premises environments.
- Include facility and execution constraints—especially operational readiness, power, and cooling—in the architecture decision.
Then compare the evaluation’s emphasis with those requirements. The useful question is not simply whether a vendor ranks high or low, but whether the report’s criteria and weighting match the decision at hand. A ranking can inform a shortlist; it cannot replace workload and operational requirements.
What the reported scores establish—and what remains uncertain
The reported 2% weighting and NVIDIA scores are attributable to Kerravala’s account of the Q3 2026 Wave. Without the underlying report and scorecard, those figures cannot be independently checked here. Nor do they establish that Forrester endorsed Kerravala’s critique, that the evaluation is invalid, or that a particular network will perform well in a specific AI deployment.
The defensible takeaway is narrower: a comparison built for broad enterprise networking may not answer every buyer’s AI-fabric question. Define the workload and operating environment first, then decide how much weight to give a matrix whose criteria may prioritize different needs.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
