Dell’s AI-Server Acceleration Broadens Enterprise Demand and Cloud Providers in 2027 Q2

Dell’s AI and Server Acceleration: What the Excerpts Establish

Bottom line

The excerpts point to a broadening customer and workload mix, but they do not include management’s answers, so they cannot definitively quantify how much of the acceleration comes from enterprise customers versus Neoclouds or Tier 2 cloud service providers.

The clearest indications are:

• Traditional server growth appears to be increasingly influenced by AI-related CPU infrastructure, including “CPU racks” and agentic-AI servers that may be classified within traditional servers rather than AI compute. ¹
• Enterprise demand remains the historical center of Dell’s traditional-server business, while AI-server demand has historically been concentrated among Neoclouds and Tier 2 CSPs. ¹
• Enterprise customers had previously been growing faster in AI servers, suggesting that the AI opportunity was broadening beyond specialized cloud providers before this quarter’s acceleration. ¹
• The reported growth was unusually strong across both categories: traditional servers were cited as growing 122%, while AI compute and storage were cited as growing 26%. ²
• The excerpts suggest that the demand is not confined to GPU clusters: it includes enterprise data-center modernization, CPU-based agentic-AI infrastructure, networking, storage, and potentially PCs pulled through by the same enterprise projects. ¹³

Customer mix

Enterprise customers

Enterprise customers appear to be the most important source of momentum in the traditional-server business. The traditional server category was described as having been “almost all enterprise customers” historically. ¹ That makes enterprise demand the natural baseline explanation for the traditional-server acceleration, although the inclusion of CPU racks and agentic-AI servers means the category is no longer a pure measure of conventional, non-AI infrastructure. ¹

The excerpts also indicate that enterprise customers were becoming more significant in Dell’s AI-server business. The question refers to Dell’s prior statement that enterprise AI-server growth had been faster than growth among other customer groups. ¹ In addition, enterprise strength was said to be driving pull-through in networking, storage, and PCs, implying that enterprise AI deployments may create a broader infrastructure basket rather than a standalone server sale. ³

This is strategically important because enterprise AI deployments are typically associated with data-center modernization and integrated infrastructure purchases, whereas Neocloud and Tier 2 CSP demand is more concentrated in high-density compute capacity. The excerpts, however, do not provide the actual enterprise percentage of AI-server revenue or orders. ¹

Neoclouds and Tier 2 cloud service providers

Neoclouds and Tier 2 CSPs remain central to the AI-server opportunity. The excerpts state that the majority of Dell’s AI-server revenue and orders had historically come from these customers. ¹ Their role is consistent with the sharp demand for accelerated-computing infrastructure and large-scale deployments.

However, the supplied material does not establish whether these customers still account for the majority of current AI-server revenue or orders. The question explicitly asks whether that historical mix remains intact, but no management response is included. ¹ Therefore, it would be premature to conclude that Neoclouds and Tier 2 CSPs are either losing share to enterprises or continuing to dominate at the same level.

The most defensible interpretation is that these customers remain a major volume driver, while enterprise customers are becoming a more important and potentially faster-growing portion of the AI business. That interpretation is supported by the reference to prior faster enterprise growth, but not by a current-period mix disclosure. ¹

Workload mix behind the acceleration

1. Conventional enterprise compute and data-center modernization

The traditional-server acceleration appears to include standard enterprise infrastructure refresh and modernization. One question specifically frames the transition from older-generation servers such as 14G to newer generations such as 17G and 18G. ⁴ That suggests at least part of the demand reflects replacement and modernization cycles rather than only incremental AI capacity.

The excerpts do not quantify how much of traditional-server growth is replacement demand, new capacity, or share gain. They do, however, raise the possibility that the growth reflects more than pricing or pre-buying, with the underlying workloads and durability of demand identified as key issues. ²

2. CPU-based agentic-AI infrastructure

A significant source of category ambiguity is the inclusion of CPU racks and agentic-AI servers in traditional-server reporting. ¹ This means the reported traditional-server growth may include infrastructure supporting AI workloads that do not rely primarily on GPU systems.

Consequently, the 122% traditional-server growth figure should not automatically be interpreted as purely conventional enterprise server demand. It may partly represent a new AI-adjacent workload category: CPU-heavy systems used for agentic AI, inference, orchestration, or related enterprise applications. The excerpts identify this classification issue but do not provide the size of the CPU-rack contribution. ¹²

3. GPU and accelerated AI compute

AI compute remains a major workload, particularly for Neoclouds and Tier 2 CSPs, which historically represented the majority of Dell’s AI-server revenue and orders. ¹ The excerpts also refer to a sharply increased backlog and ask whether Dell’s growth should track the projected growth of its AI-server partner, indicating that demand for accelerated systems remained exceptionally strong. ⁵

Still, the available material does not distinguish between training, inference, model development, or other AI workloads. It also does not disclose the relative contribution of GPU servers versus CPU-based AI systems. Any precise workload attribution would therefore go beyond the evidence provided.

4. Storage-intensive AI deployments

Storage appears to be an increasingly important complement to AI-server deployments. The excerpts refer to a long-discussed $2-$3 storage attach opportunity per AI server and ask whether that opportunity is beginning to materialize. ⁶ Another question specifically asks about the level of Lightning storage attachment to cloud AI-server deals. ⁷

This suggests that AI deployments are generating demand beyond compute, particularly for data ingestion, training datasets, checkpoints, model outputs, and related data-management requirements. The excerpts support the existence of a storage-attach opportunity, but they do not provide the actual attach rate, revenue per server, or duration of the runway. ⁶⁷

5. Networking and broader enterprise infrastructure

Enterprise AI projects appear to have a broader product footprint. The excerpts state that strength in the core enterprise customer base was driving pull-through in networking, storage, and PCs. ³ This supports the view that enterprise customers are purchasing AI as part of wider infrastructure modernization rather than buying isolated accelerator servers.

That broader basket may also help explain why the acceleration is economically attractive: a single enterprise engagement can extend across servers, storage, networking, and client devices. The excerpts raise the possibility that a larger enterprise share of AI engagements could support a structurally higher margin framework, but they do not include management’s confirmation or quantify the margin effect. ³

Demand quality: volume, pricing, and backlog

The excerpts indicate that pricing is a meaningful tailwind, particularly in storage. ⁶ Analysts also questioned whether traditional-server and storage growth was being driven by pricing, pre-buying, or genuine demand. ² Another question asks how much traditional-server and storage growth is attributable to volume versus pricing and whether customers are beginning to resist higher prices. ⁸

Accordingly, the acceleration should be viewed as a combination of:

1. Volume growth, including enterprise refresh, AI capacity additions, and cloud-provider deployments; ²⁸
2. Pricing increases, especially in storage and potentially constrained components; ⁶⁸
3. Mix effects, including higher-value AI systems, CPU-based AI racks, storage attach, and broader enterprise pull-through; ¹⁶³
4. Potential backlog and supply constraints, which indicate demand is exceeding immediately available supply but do not by themselves prove that all backlog represents durable end demand. ⁵⁹

The excerpts also raise the possibility that budget constraints could cause enterprise customers to split purchases between current orders and later backlog—for example, buying half of a planned deployment immediately and deferring the balance. ¹⁰ No management response is provided, so the extent of such deferral is unknown.

Overall interpretation

The acceleration appears to be driven by three overlapping customer/workload engines:

Conclusion

The excerpts support a view that Dell’s server acceleration is not simply a Neocloud/GPU story. It appears to combine:

• Large-scale AI deployments by Neoclouds and Tier 2 CSPs;
• A strengthening enterprise AI business;
• Conventional enterprise server modernization;
• CPU-based agentic-AI infrastructure reported within traditional servers; and
• Increasing storage, networking, and other infrastructure attachment.

The most important caveat is that the excerpts contain the questions but not management’s responses. Therefore, they establish the historical customer mix, the relevant workload categories, and the issues investors were investigating, but they do not establish the current percentage split among enterprise customers, Neoclouds, and Tier 2 CSPs, nor do they confirm the precise contribution of pricing, volume, backlog, or share gains.