HPE presents the most structurally interesting comparison to Dell in the 4+1 model because it makes genuine software authority claims that Dell does not. Three capabilities differentiate HPE's architectural position: GreenLake Intelligence (agentic AI mesh using domain-specific LLMs via MCP, HPE-owned Layer 2A/2C for IT operations), the $14B Juniper Networks acquisition (full networking IP stack from silicon to software, owned networking depth that Dell entirely delegates to NVIDIA Spectrum), and the Unleash AI program with Kamiwaza as the chosen Layer 2C orchestration partner (validated in production at Town of Vail).
The AI workload runtime (Layer 2B) remains structurally dependent on NVIDIA AI Enterprise, branded as ‘NVIDIA AI Computing by HPE.’ HPE co-engineers more deeply than Dell. Private Cloud AI is a jointly developed product, but the DAPM implication is the same: the Layer 2B runtime is NVIDIA's, Ceded on portability inside a turnkey system, while the runtime call is the model's and the enterprise's own tool code gates it, as on NVIDIA's row. However, HPE brackets the NVIDIA-controlled Layer 2B with governance surfaces above (GreenLake Intelligence at 2A/2C) and below (GreenLake platform at 2A), governance coverage, not governance authority: the bracket is real, and it is a bracket of Ceded surfaces, HPE-owned above and partner-captive (Kamiwaza) beside it.
Kamiwaza's capabilities span multiple layers: context orchestration (1B), governed data pipelines (1C), agent execution coordination (2B), and decision authority placement (2C), making it a multi-layer platform, not a point solution. The Town of Vail deployment serves as a by-proxy assessment of Kamiwaza's capabilities across this full span. The Kamiwaza-provided functions are Ceded to the chosen partner: the platform's orchestration and governance opinions are Kamiwaza-captive, exactly as they score on Kamiwaza's own row. The structural advantage over Dell survives on the capability axis: the function exists, someone is accountable, and it is validated in production. The authority reading is capability-with-capture versus Dell's no-capability-no-capture.
The Cray supercomputing heritage gives HPE a sovereign AI positioning that Dell cannot match: exascale systems for Argonne, HLRS, HammerHAI (EU AI Factory). This is a differentiated 0A capability with implications for sovereign data governance at Layer 1A.
HPE has one of the most credible on-prem AI infrastructure stacks in the market. Its credibility comes from genuine software authority (GreenLake Intelligence, Data Fabric, OpsRamp), depth of networking IP (Juniper/Aruba/Slingshot: three owned fabrics, each captive to its own stack but each representing real engineering investment), sovereign compute heritage (Cray), and a structured ecosystem model (Unleash AI) that deliberately addresses Layer 2C through a chosen partner rather than leaving it unaddressed.
Layer-by-layer status: Layer 0A (ProLiant, the XD690, GB300 NVL72, Cray EX and the HPE AI Factory), Layer 0B (Slingshot 400 for AI and HPC, With Juniper and Aruba Beside It), Layer 0C (Cray K3000 on DAOS and E2000 on Lustre, With Alletra MP X10000), Layer 1A (X10000 Data Intelligence Enrichment and a Polaris Catalog; Classification Not Documented), Layer 1B (Reference Stack + Partner Orchestration), Layer 1C (HPE + Open Source), Layer 2A (HPE Strength), Layer 2B (NVIDIA Runtime + Framework Surface), Layer 2C (IT-Ops Reasoning + Partner AI Orchestration), Layer 3 (+1) (No First-Party AI Applications; GreenLake Marketplace Sells ISV Apps but Doesn’t Support Them).
Assessment framework: 4+1 Layer AI Infrastructure Model. Scoring model: Decision Authority Placement Model (DAPM) — Retained, Delegated, or Ceded. Published by The CTO Advisor LLC (DBA The Advisor Bench). Author: Keith Townsend. Date assessed: October 27, 2026. Version: v1.54 - 4+1 v3: Layer 0 Sub-Layers.
HPE presents the most structurally interesting comparison to Dell in the 4+1 model because it makes genuine software authority claims that Dell does not. Three capabilities differentiate HPE's architectural position: GreenLake Intelligence (agentic AI mesh using domain-specific LLMs via MCP, HPE-owned Layer 2A/2C for IT operations), the $14B Juniper Networks acquisition (full networking IP stack from silicon to software, owned networking depth that Dell entirely delegates to NVIDIA Spectrum), and the Unleash AI program with Kamiwaza as the chosen Layer 2C orchestration partner (validated in production at Town of Vail).
The AI workload runtime (Layer 2B) remains structurally dependent on NVIDIA AI Enterprise, branded as ‘NVIDIA AI Computing by HPE.’ HPE co-engineers more deeply than Dell. Private Cloud AI is a jointly developed product, but the DAPM implication is the same: the Layer 2B runtime is NVIDIA's, Ceded on portability inside a turnkey system, while the runtime call is the model's and the enterprise's own tool code gates it, as on NVIDIA's row. However, HPE brackets the NVIDIA-controlled Layer 2B with governance surfaces above (GreenLake Intelligence at 2A/2C) and below (GreenLake platform at 2A), governance coverage, not governance authority: the bracket is real, and it is a bracket of Ceded surfaces, HPE-owned above and partner-captive (Kamiwaza) beside it.
Kamiwaza's capabilities span multiple layers: context orchestration (1B), governed data pipelines (1C), agent execution coordination (2B), and decision authority placement (2C), making it a multi-layer platform, not a point solution. The Town of Vail deployment serves as a by-proxy assessment of Kamiwaza's capabilities across this full span. The Kamiwaza-provided functions are Ceded to the chosen partner: the platform's orchestration and governance opinions are Kamiwaza-captive, exactly as they score on Kamiwaza's own row. The structural advantage over Dell survives on the capability axis: the function exists, someone is accountable, and it is validated in production. The authority reading is capability-with-capture versus Dell's no-capability-no-capture.
The Cray supercomputing heritage gives HPE a sovereign AI positioning that Dell cannot match: exascale systems for Argonne, HLRS, HammerHAI (EU AI Factory). This is a differentiated 0A capability with implications for sovereign data governance at Layer 1A.
HPE has one of the most credible on-prem AI infrastructure stacks in the market. Its credibility comes from genuine software authority (GreenLake Intelligence, Data Fabric, OpsRamp), depth of networking IP (Juniper/Aruba/Slingshot: three owned fabrics, each captive to its own stack but each representing real engineering investment), sovereign compute heritage (Cray), and a structured ecosystem model (Unleash AI) that deliberately addresses Layer 2C through a chosen partner rather than leaving it unaddressed.
CPUs, accelerators, servers, integrated systems, OS and hypervisor, physical plant
Intel Xeon 6 / AMD EPYC. HPE iLO management silicon (HPE-owned). Improved perf/watt, security. Foundation for Private Cloud, Private Cloud AI, standalone.
Exascale-class supercomputing on the Cray Slingshot interconnect. Its current QuickSpecs list one accelerator blade, the EX254n with "Two 4-socket Nvidia GH200 Grace Hopper Superchip nodes"; the rest are AMD EPYC CPU blades.
Full-stack AI infra: compute, GPUs, networking, liquid cooling, software, services. Blackwell (RTX PRO 6000). Multi-tenancy via MIG with GPU passthrough (Spring 2026). Air-gapped configs for sovereign. NVIDIA Cloud Partner endorsed. STIG-hardened, FIPS-enabled.
Proprietary DLC supporting up to 400kW per rack with warm water operation. As Blackwell/Vera Rubin density increases, cooling becomes the physical constraint. Cray heritage is genuine differentiator.
HPE's KVM-based HVM hypervisor with cluster HA, live compute and storage migration, distributed workload placement and integrated data protection, managing HVM and existing ESXi clusters from one console. VMware image conversion included. Licensed per physical CPU socket; FC, iSCSI and NFS v3 external storage.
HPE's GB300 NVL72 rack, in QuickSpecs since December 1, 2025 (Version 3, September 8, 2026), with system codes S7K95A and S7K96A: "72 NVIDIA Blackwell Ultra GPUs all interconnected with NVIDIA’s 5th-gen NVLink technology," on power shelves and direct liquid cooling. Ceded on the integrated-system rule, the Lenovo NVL-class reading: a rack-scale system built around a captive NVLink fabric can't be lifted as deployed.
Air-cooled server with "8x NVIDIA® Blackwell Ultra/B300 HGX GPUs" and two Intel Xeon 6 processors (6776P or 6767P), chassis SKU S6G97A, in QuickSpecs since November 3, 2025. Standard NVIDIA HGX: a CUDA or PyTorch workload moves to another OEM's HGX B300 server without a rebuild, Retained; the CUDA dependency is read on NVIDIA's row.
Blackwell Ultra in the GB300 NVL72 and the XD690's HGX B300, and RTX PRO 6000 in the AI Factory. Vera Rubin NVL72 (December 2026) and the GX5000's AMD MI430X blades are watch-listed.
HPE sells the compute the enterprise racks and runs. ProLiant Compute Gen12 carries Intel Xeon 6 or AMD EPYC, with HPE's own iLO management silicon. The Compute XD690 is an eight-way HGX B300 server, and HPE sells the GB300 NVL72 rack on its own part numbers. The Cray EX4000 is an exascale-class supercomputer on HPE's own Slingshot interconnect, and the HPE AI Factory bundles Blackwell compute, networking and cooling. Cray direct liquid cooling handles up to 400kW per rack on warm water, and Morpheus VM Essentials adds HPE's own KVM-based hypervisor. The buyer gets current accelerated compute on hardware it owns. The architect's concern is the integrated systems. ProLiant lifts to another OEM; the Cray systems and the AI Factory bundle can't be lifted as deployed. Calibration: Dell, Lenovo and Supermicro read strong on the same OEM shape; AWS and GCP strong on compute the enterprise rents.
Retained where the grade rests. A ProLiant server or Cray system the enterprise owns makes no runtime call on HPE's behalf; the enterprise's own hypervisor, scheduler and policies decide, and iLO reads with each server. Portability splits: ProLiant, the XD690 and the cooling read Retained, while the GB300 NVL72, the Cray EX and the AI Factory read Ceded on the integrated-system rule, and Morpheus VM Essentials reads Ceded on the Nutanix AHV reading.
Watch-list (not scored): HPE Compute XD700, an OCP-inspired AI server on NVIDIA HGX Rubin NVL8, liquid-cooled, early 2027; Vera Rubin NVL72 in the AI Factory, December 2026. Named: Argonne, HLRS, HammerHAI, Hudson River Trading and KISTI are Cray GX5000 customers. Watch-list (October 23, 2026, not scored): the HPE Cray GX5000 platform. HPE's November 2025 release says the GX440n (Vera Rubin), GX250 and GX350a (AMD MI430X) blades "will be available in early 2027"; the GX240 rests on NVIDIA's Vera CPU, which isn't generally available. At availability the rack reads Ceded on the integrated-system rule.
Interconnects, switching and routing, network virtualization
QFX switches for AI data centers, the QFX5240 for "AI Data Center Networking deployments with AI/ML workloads," with 800GbE ports, managed by Apstra. Juniper's validated AI data center design runs three fabrics (a front end, a GPU back end and a storage back end) on QFX leaves, each an Apstra blueprint, with DCQCN (PFC and ECN) keeping the GPU back end lossless. Junos opinions are captive to the Juniper stack: Ceded.
HPE-owned high-performance interconnect delivering 400 Gbps at scale with ultra-low tail latency for AI workloads. Distinct from NVIDIA InfiniBand, this is HPE networking IP for the supercomputing fabric. Slingshot opinions are captive to the Cray/HPE HPC stack: there is no commodity substrate that would let the enterprise move these fabric configurations to InfiniBand or Ethernet without rebuilding.
Campus and edge networking with AI-native Central platform. Being retooled on GreenLake Intelligence agentic mesh. Complementary to Juniper's DC focus and Slingshot's HPC fabric. Aruba opinions are captive to the Aruba/HPE stack: there is no commodity substrate that would let the enterprise move campus networking configurations to Juniper or Cisco without rebuilding.
The captive Slingshot interconnect the Cray EX systems ship with. Integrated-system rule at this facet.
The bundle's integrated networking, captive to the AI Factory as deployed. Integrated-system rule at this facet.
The captive NVLink scale-up fabric inside HPE's GB300 NVL72 rack. Integrated-system rule at this facet.
PTX routers, which Juniper's datasheet calls "a superior spine or leaf in AI data center networking environments," with RoCEv2 and VOQ scheduling, and which carry backbone and peering routing. Junos opinions are captive: Ceded.
"A robust portfolio of software-defined networking (SDN)-enabled routing platforms" for edge and data center gateway routing, on Junos: Ceded.
EX switches, "cloud-ready, high-performance access and distribution/core-layer devices for enterprise" networks, run as campus fabrics from Mist Cloud. Junos and Mist opinions don't lift to another campus stack: Ceded. SRX firewalls are noted, not scored at Layer 0.
Wi-Fi 7 access points (AP47, AP37, AP36, AP66) that "work in conjunction with the Mist™ platform's cloud and Marvis® AI." Mist's opinions don't lift to another wireless stack: Ceded.
ConnectX-9 SuperNICs and BlueField-4 DPUs in the AI Factory; Quantum-X800 InfiniBand for the GX5000 in 2027. NVIDIA's fabric competes with HPE's own Slingshot and Juniper.
The GB300 NVL72's scale-up domain, captive inside the rack.
HPE owns its fabrics outright. Cray Slingshot 400 is HPE networking IP, 400 Gbps at scale with low tail latency, and it's the interconnect inside every Cray EX system. Juniper, acquired in July 2025 for $14 billion, brings Junos, QFX switches, MX routers and Apstra intent-based fabric automation to the data center. Its validated AI design runs separate front-end, GPU back-end and storage back-end fabrics. Aruba covers campus and edge. No other row carries three purpose-built fabrics. The buyer gets an AI fabric HPE designed, not one it resells. The architect's concern is that each fabric is captive to its own stack. Slingshot configurations don't move to InfiniBand or Ethernet, and Junos opinions don't move to Aruba or Cisco, without rebuilding. Three platforms also means three management tools. Calibration: NVIDIA reads strong on Spectrum-X and InfiniBand, AWS on EFA; Dell, Lenovo and Supermicro strong on Spectrum-X on their own paper.
Ceded where the grade rests. Slingshot's routing and congestion control on the Cray systems, and Junos in the data center, are the fabric's own runtime calls, made inside a configuration the enterprise writes with no per-decision override: vendor decides, visible, not overridable, Ceded. Portability: Slingshot, Juniper, Aruba and the Cray and AI Factory fabrics all read Ceded.
Watch-list (not scored): NVIDIA Quantum-X800 InfiniBand for the Cray GX5000, 2027. Named, not scored here: Juniper Mist and Aruba Central, the device-fleet planes inside the Juniper and Aruba chips, read at 2A (October 16, 2026); Apstra stays at 0B because it programs switch state.
Arrays, storage servers and software, front-end protocols: where the data sits
Disaggregated, all-flash, scale-out. Native file + object on single platform. 16 nodes, 23PB raw. 100% availability guarantee. RDMA-enabled for AI pipeline optimization across training, inference, KV cache. 2.5 PB/hr backup ingest. Storage opinions (configs, tiering, performance tuning) are captive to the HPE platform. Native Model Context Protocol (MCP) servers are built into the array so GreenLake Intelligence agents and GreenLake Copilot can reach storage operations directly. They're the array's management half, not a component of their own (Keith, October 16, 2026): MCP is an open protocol, but the servers answer only for the X10000, so they read with it. Per HPE's release notes (October 6, 2026), OS v2.0.0.0 "Added native file storage support, allowing file workloads through NFS v4.1 for qualified Linux clients," v2.1.0.0 "Added support for NFSv4.1 over RDMA," and HPE publishes an "S3-Compatible APIs Protocol Guide" for the system.
Mission-critical block storage. 6-controller-node scaling (50% more perf vs 4-node). Dual-node fault tolerance. 5:1 data reduction guarantee. Real-time agentic support (v10.6.0): coordinated specialized AI agents for semantic understanding, adaptive reasoning, and prescriptive intelligence. Agents draw from system telemetry metadata, best practices, and accumulated product knowledge across installed base. Storage opinions captive to HPE platform. Hosts attach over Fibre Channel or iSCSI: per HPE's installation guide (August 2026), "A minimum of one Fibre Channel interface card or an iSCSI card is required for host connectivity."
Factory-integrated storage on HPE ProLiant DL360 Gen12 storage nodes, deployed alongside the Lustre-based E2000 where AI I/O exceeds roughly 50 million IOPS per rack. Selected for Oak Ridge National Laboratory’s next-generation exascale system "Discovery"; DAOS-based systems hold the top two IO500 positions. HPE documents the system as "fully orchestrated and managed by K3000 management stack," HPE's software on top of the open-source engine. Ceded, the E2000 reading: the management stack and the factory integration are HPE's, and the system lifts only as HPE built it. The DAOS engine underneath, which the enterprise can run without HPE, is its own chip.
The community storage engine the K3000 runs, BSD 2-Clause under the DAOS Foundation at the Linux Foundation, community-owned rather than HPE-owned, with no capacity-based licensing, running on commodity x86 NVMe servers and consumed through POSIX, SMB, NFS and S3. The enterprise can run community DAOS without HPE, and the pools and containers it accumulates run on any other DAOS deployment: Retained, the HPE DAOS reading the DDN and HPE Lustre chips follow. The K3000 management stack stays in the K3000 chip.
Data Fabric's distributed file and object store, its global namespace, and policy-based data placement and tiering across hybrid environments. The Apache Polaris catalog and governance stay at 1A.
Continuous data protection, AI-powered assistant, Microsoft Defender integration, live VMware-to-HPE VM migration. Near-zero RPO/RTO. Replication and recovery opinions are captive to the Zerto/HPE platform.
Lustre-based parallel storage on NVMe Gen5 building blocks, all-flash, all-disk, hybrid or tiered, with OpenZFS or LDISKFS on HPE's parity-declustered GridRAID as the local object store. Tiering's policy engine moves files between flash and disk; data mover nodes scale it out. Slingshot, NDR InfiniBand or 400GbE.
The community parallel file system the E2000 runs. The enterprise can run community Lustre without HPE, and the opinions it accumulates in layouts, pools and quotas run on any other Lustre implementation: Retained, the DDN Lustre and HPE DAOS reading. GridRAID and the tiering policy engine stay in the E2000 chip.
The Lustre client protocol, which independent implementations share: DDN's row names the Lustre client across DDN, HPE and the clouds' managed Lustre services. Delegated, the DDN front-end reading. GridRAID and the tiering policy engine stay in the E2000 chip, and the Lustre core is its own chip.
The captive storage in HPE Private Cloud AI, which the QuickSpecs list as Alletra Storage MP X10000 for the Small, Medium and Large sizes. The Developer System uses internal storage.
S3, per HPE's S3-Compatible APIs Protocol Guide, and NFS v4.1 for qualified Linux clients, including over RDMA, per HPE's release notes, which independent implementations share: hosts and applications re-point to another server or target speaking the same protocols, and the data and provider-specific policies migrate with it. Delegated. The array and its management stay in the X10000 chip, Ceded.
Fibre Channel, NVMe-oF/FC, NVMe-oF/TCP and iSCSI, per HPE's QuickSpecs, which independent implementations share: hosts and applications re-point to another server or target speaking the same protocols, and the data and provider-specific policies migrate with it. Delegated. The array and its management stay in the B10000 chip, Ceded.
RDMA for GPU-direct storage access, the same acceleration Dell and VAST use.
HPE's storage substrate runs from the supercomputer to the enterprise array. The Cray K3000 embeds open-source DAOS on ProLiant DL360 Gen12 servers. DAOS-based systems hold the top two IO500 positions, and Oak Ridge selected the K3000 for its next exascale system. The E2000 is Lustre on HPE's parity-declustered GridRAID. Alletra Storage MP X10000 is disaggregated all-flash file and object, RDMA-enabled for training, inference and KV cache, and the B10000 is mission-critical block. Data Fabric's distributed store and Zerto's continuous replication place and protect the bytes. The buyer gets AI-throughput storage at every tier from its server vendor. The architect's concern is the storage OS. Alletra MP decides placement, protection and rebuild, and those opinions stay on HPE's platform. DAOS is the exception: BSD-licensed and community-owned. Calibration: Dell reads strong on PowerScale and Lightning FS; VAST, WEKA and DDN strong on storage alone.
Ceded where the grade rests. The Alletra MP operating system on the X10000 and B10000 decides placement, protection and rebuild at runtime, and the array's built-in MCP servers answer only for that array: vendor decides, visible, not overridable, Ceded. Portability: the DAOS engine inside the K3000 and the Lustre core inside the E2000 read Retained, though each engine still decides placement; the E2000's Lustre client front end reads Delegated; the K3000 and E2000 systems, the Alletra arrays, Data Fabric, Zerto and the AI Factory's storage read Ceded.
Named, not scored here: Data Fabric's real-time S3-to-S3 object movement, which is data movement and sits closer to 1C; Zerto's live VM migration, which may be a 1C facet.
The data itself, organized and governed: the Governance Catalog that Layer 2C queries
Generally available January 2026; orderable as a metered SaaS subscription (100 TB of metadata and 1,000 vectors, three- and five-year terms) with an HPE datasheet. GPU-backed data intelligence nodes extract, enrich, and store custom metadata and vector embeddings from X10000 objects inline as data lands. HPE’s datasheet and QuickSpecs document no classification, sensitivity, or lineage function, so it carries the cell at moderate, since 1A is graded on a catalog that classifies the data (October 15, 2026). The enrichment schema and catalog structure are captive to the HPE platform.
Policy-based data placement and movement (tiering) across hybrid environments. Conversational interface and agentic AI assistant for natural language access to global namespace. HPE positions metadata intelligence for lineage and governance; the 8.1 documentation describes a federated Apache Polaris catalog and MCP access to metadata, not a classification function. Apache Polaris catalog support for Iceberg tables: consistent governance and compliance across platforms. Real-time S3-to-S3 object movement between any S3-compatible storage systems. Data Fabric governance opinions are captive to the HPE platform. Polaris provides open metadata portability but the orchestration layer is proprietary. The Iceberg REST catalog surface in front of it is its own chip.
The Iceberg REST catalog has independent implementations: code and definitions written to it move to another compatible implementation within the features both support, the data moves by export, replication or rewrite, and the provider's control-plane configuration and operational state don't lift. Delegated. The service behind it is its own chip, Ceded.
RDMA via CX-8/CX-9 SuperNICs for GPU-direct storage access. Same acceleration Dell and VAST also use.
HPE's 1A is inline enrichment, one step short of the bar of a catalog that classifies the data. X10000 Data Intelligence went GA in January 2026, a metered SaaS subscription on GPU-backed data intelligence nodes. It reads objects as they land and writes custom metadata and vector embeddings back to them. That goes further than a file system index, because it touches content. But HPE's datasheet and QuickSpecs describe the output as 'AI-powered metadata enrichment', and neither names classification, sensitivity or lineage. The embeddings answer what the data is similar to, which scores as retrieval at 1B. Data Fabric 8.1 adds a federated Apache Polaris catalog and MCP access to metadata, which governs tables, not what an object contains. The arrays, Data Fabric's store and Zerto's replication protect and hold the bytes, 0C functions; Zerto's live migration reads at 1C. Moderate, at the top of the band. Calibration: Dell reads moderate on MetadataIQ and WEKA on its metadata-only catalog; VAST strong on a Catalog that classifies.
Data Intelligence enriches each object as it lands, with GPU-backed inference on HPE's enrichment schema: vendor decides, visible, not overridable, Ceded. Portability: the enrichment schema and catalog structure stay on HPE's platform, Ceded; Polaris keeps Iceberg table metadata portable across platforms.
Watch-list (HPE Discover 2026, not scored): Alletra MP X10000 auto metadata, governance policies and AI-ready pipelines, Q4 2026. It returns to strong when HPE documents a GA classification or sensitivity function. Named, not scored: a September 25, 2025 partner deck draws a 'Data catalog & classification' box; a pre-GA slide label isn't a product document.
Low-latency retrieval for RAG — vector/hybrid search, context windows
In Town of Vail: manages the full context pipeline for document-centric use cases. Identifies documents requiring processing (Section 508 compliance), ingests and extracts content (housing deeds), prepares contextual inputs for agent consumption. Determines what context each agent needs, from which sources, under what governance constraints. This is retrieval orchestration, above the storage layer, below the agent runtime. Governs cross-departmental context routing (legal, housing, admin). Lift-to-leave: the context-orchestration opinions are Kamiwaza-captive, Ceded to the chosen partner, as on Kamiwaza's own row. HPE sells it inside the HPE Agentic Smart City Solution, which "runs on HPE Private Cloud AI," with "Kamiwaza provides agentic orchestration and decision routing" and the system "supported and evolving as a single integrated capability" (HPE announcement, October 28, 2025; HPE-sponsored CTO Advisor report a00156405enw, January 2026).
Embedding models (NV-EmbedQA-E5-v5, Mistral7B-v2, Arctic-Embed-L) and reranking in unified microservice. GPU-accelerated retrieval for RAG pipelines on Private Cloud AI. Provides the embedding intelligence that HPE’s storage does not.
Research assistant and enterprise data agent blueprint. Connects enterprise data to AI agents via retrieval pipelines. Available on Private Cloud AI.
HPE’s reference RAG architecture uses NeMo Retriever for embedding, Milvus (open-source) for vector database, LangChain for chain serving. HPE does not own any retrieval intelligence component: it provides storage substrate and deployment platform.
Layer 1B is HPE’s thinnest proprietary layer. HPE provides storage infrastructure (Data Fabric namespace, Alletra RDMA) but does not own a vector database, an embedding engine, or a retrieval framework. The retrieval intelligence stack is entirely NVIDIA (NeMo Retriever) + open source (Milvus, LangChain). The three-vendor comparison at Layer 1B: • Dell: storage (PowerScale/ObjectScale) + Elastic (search intelligence, Delegated ISV) + NVIDIA (cuVS acceleration). Three authorities. • HPE: storage (Alletra/Data Fabric) + NVIDIA (NeMo Retriever, embedding) + open source (Milvus, LangChain). No proprietary retrieval intelligence. When Kamiwaza is added via Unleash AI, it provides governed retrieval orchestration above the storage and embedding layers. • VAST: storage + embedding + vector search + retrieval pipeline all in one platform (InsightEngine, native vector search, DataBase). One authority. HPE’s retrieval gap is structural: the company has no analog to Dell’s Elastic partnership or VAST’s native InsightEngine. This is a deliberate architectural choice: HPE provides infrastructure substrate and delegates retrieval intelligence to NVIDIA and open-source components. When Kamiwaza enters via Unleash AI, the retrieval story changes. Kamiwaza provides governed context orchestration that neither the storage layer nor the NVIDIA retrieval components provide independently: cross-departmental context routing, authority-constrained retrieval, and document-pipeline coordination. In the Town of Vail, this means the Section 508 compliance agent receives only the documents it’s authorized to process, with retrieval governed by department boundaries. This is a layer of retrieval intelligence that storage-native search (VAST) and embedding-accelerated search (Dell+Elastic) don’t address: the governance of who receives what context under what authority. The 4+1 model question: is governed context orchestration a Layer 1B function (retrieval) or a Layer 2C function (governance)? Kamiwaza’s context management spans both: it retrieves content (1B) according to governance policies (2C). The assessment classifies the retrieval function at 1B and the governance function at 2C.
Moderate to high. HPE’s own Layer 1B authority is limited to storage infrastructure. Embedding intelligence is NVIDIA (NeMo Retriever). Vector storage is open source (Milvus). Retrieval framework is open source (LangChain). Governed context orchestration is Kamiwaza (Ceded to the partner via Unleash AI). Compare to Dell: Dell delegates retrieval intelligence to Elastic (proprietary ISV partnership) and acceleration to NVIDIA. Dell’s borrowed judgment at 1B is Moderate, split between a proprietary ISV and NVIDIA. Compare to VAST: VAST’s borrowed judgment at 1B is Low: InsightEngine, vector search, and the retrieval pipeline are VAST IP. Only embedding model execution (NIM) is NVIDIA-provided, and InsightEngine is model-agnostic. HPE’s Layer 1B borrowed judgment is the highest of the three vendors because HPE owns the least retrieval IP. The mitigation: Kamiwaza’s governed orchestration adds a unique capability that pure retrieval engines don’t provide.
The HPE Developer Portal’s RAG reference architecture is instructive: NeMo Retriever embedding + Milvus vector DB + LangChain + Llama3-70B. This is a standard NVIDIA reference stack, not an HPE-differentiated architecture. Any NVIDIA partner (Dell, Lenovo, Supermicro) could deploy the identical stack. HPE’s Layer 1B differentiation comes not from the retrieval stack but from the storage substrate below it (Data Fabric governance, Alletra RDMA performance) and the orchestration layer above it (Kamiwaza context governance). The KV cache storage support in Alletra X10000 is worth noting as a Layer 1B/2B bridge: inference state persistence in storage allows agents to maintain context across sessions without holding GPU memory. Dell’s equivalent is the CMX KV cache offload (NVIDIA technology). HPE’s is storage-native. VAST’s CNode-X collocates cache and compute. The Alletra X10000's RDMA file and object access is the storage retrieval reads from, not retrieval, so it isn't scored at 1B. Removed in the 1B dependency audit (October 24, 2026): the HPE Ezmeral Data Fabric retrieval-surface chip. Data Fabric 8.1 documents MCP endpoints "for streamlined data catalog access" and no conversational or natural-language retrieval feature; catalog access is scored at 1A, on the Data Fabric catalog chip. The cell rests on Kamiwaza inside the HPE Agentic Smart City Solution, which Kamiwaza's 1.3.4 documentation of its GA build supports: connector pipelines into vector stores and grounded context retrieval for agents (re-read October 24, 2026).
Move/transform data — ETL/ELT, lineage, cost-aware movement, KV cache tiering
Policy-based data placement considering performance, sovereignty, costs, compliance. Data lineage and compliance tagging. Agentic AI assistant for automated reporting and data placement decisions. Real-time S3-to-S3 object movement for AI data ingestion from external sources. Proprietary pipeline orchestration captive to HPE platform. Not scored within this chip under the GA evidence table: Agentic AI assistant for automated reporting and data placement decisions, Data lineage and compliance tagging (The HPE Data Fabric 8.0 and 8.1 what's-new pages document an MCP server, Object Copier Hub, inline erasure coding (instant tiering) and object lifecycle, but no agentic placement assistant and no lineage or compliance tagging. hpe.com and buy.hpe.com (403s and timeouts) couldn't be reached to check product-page claims.).
Airflow as HPE packages it on Private Cloud AI: one instance per cluster, run by an Airflow Operator, with DAGs pulled from a Git repository, RBAC on DAGs, and DAGs that submit Spark applications. The enterprise authors, schedules, and monitors its own pipelines. HPE AI Essentials is the successor to HPE Ezmeral Unified Analytics (end of sale October 31, 2024), built through BlueData (2018), MapR (2019), Ampool (2021), and Arrikto (2023). Open-source substrate: the DAGs move to any other Airflow.
In Town of Vail: orchestrates governed data flows across department boundaries: housing deeds from ingestion through verification to audit across legal, housing, and admin functions. Decision-driven data movement where pipeline logic is governed by authority constraints and compliance policy, not static ETL schedules. Lift-to-leave: the decision-flow and pipeline opinions are Kamiwaza-captive, Ceded to the chosen partner. HPE sells it inside the HPE Agentic Smart City Solution, which "runs on HPE Private Cloud AI," with "Kamiwaza provides agentic orchestration and decision routing" and the system "supported and evolving as a single integrated capability" (HPE announcement, October 28, 2025; HPE-sponsored CTO Advisor report a00156405enw, January 2026).
Continuous data protection, AI-powered assistant, Microsoft Defender integration, live VMware-to-HPE VM migration. Near-zero RPO/RTO. Replication and recovery opinions are captive to the Zerto/HPE platform.
Two controllers run Spark workloads: a multi-version Spark Operator that takes HPE-curated or the enterprise's own open-source Spark images, and a Livy server for interactive sessions (HPE images only), with the Spark History Server keeping run history. Open-source substrate: the jobs move to any other Spark.
ML pipelines in HPE's Kubeflow, with loop parallelism, dynamic resource limits, and Sub-DAG handling, behind single sign-on. KServe serving, Katib, and notebooks in the same Kubeflow read at Layer 2B. Open-source substrate: the pipeline definitions move to any other Kubeflow Pipelines.
HPE's SQL query engine built on PrestoDB and tuned for federated queries across connected sources (JDBC, Hive, Iceberg, Delta, Snowflake, Oracle, SQL Server, MongoDB, and others), with dataset caching and an MCP server. Queries are Presto SQL over the sources where the data lives.
GPU-accelerated data prep, model training, and visualization within Ezmeral Unified Analytics. Up to 29x faster development. Spark acceleration is the primary NVIDIA contribution at Layer 1C.
Pre-built AI application patterns deployed on Private Cloud AI. Pipeline templates, not pipeline infrastructure.
Three vendors, three distinct architectural strategies for Layer 1C: • Dell: acquired Dataloop (proprietary orchestration, no-code/low-code). Dell’s strongest software move, but the broader pipeline layer depends on ISV partners (ClearML, DataRobot, Starburst). Multiple authority boundaries. • HPE: packages the open-source ML pipeline lifecycle (Airflow → Spark → Kubeflow → Ray → MLflow) as AI Essentials on Private Cloud AI under enterprise-grade guardrails. Four acquisitions (2018–2023) demonstrate deliberate investment. Value is in curation, hardening, support, integration, not proprietary technology. • VAST: built a proprietary DataEngine (event-driven serverless execution on CNodes). Entirely VAST IP. Tightly integrated with storage and retrieval layers. One authority. HPE’s open-source approach creates a specific DAPM trade-off: the enterprise avoids vendor lock-in (Airflow and Kubeflow are portable), but HPE’s authority is in packaging rather than core technology. If Apache Airflow’s community changes direction, HPE is affected. This is a different risk profile than Dell’s (proprietary Dataloop, partner-dependent beyond it) or VAST’s (proprietary DataEngine, VAST-dependent entirely). Data Fabric’s policy-based movement with Apache Polaris governance connects Layer 1C to Layer 2C: data moves according to explicit policies that consider performance, data locality, sovereignty, costs, and compliance. This governance-aware data movement feeds both GreenLake Intelligence (infrastructure decisions) and Kamiwaza (AI workload decisions). Dell’s Dataloop provides orchestration without integrated governance policy. VAST’s DataEngine has Event Broker for data-event-driven movement without an explicit policy engine. When Kamiwaza is selected via Unleash AI, it adds decision-driven pipeline capability: documents move across department boundaries based on decision logic (legal review required? accessibility compliance met? authority approval needed?). This connects infrastructure-level pipeline capabilities to business-level decision flows, where Layer 1C meets Layer 2C. Decision authority: the open pipeline stack carries the grade, because Data Fabric's S3-to-S3 movement and policy placement alone wouldn't hold moderate. Airflow, Spark, and Kubeflow run the DAGs and jobs the enterprise writes, with run history in Airflow and Spark History. Code decides, visible, overridable, Retained, the judgment test (October 4, 2026).
Low for pipeline packaging and integration (HPE owns AI Essentials, Data Fabric, Morpheus). Underlying components are open-source, limiting deep technical authority but also limiting NVIDIA dependency: RAPIDS for Spark is the only NVIDIA contribution at this layer. Open-source components are substitutable by the enterprise without HPE’s permission. Compare to Dell: Dell owns Dataloop (Retained) but depends on partners for everything else at Layer 1C. Four authority boundaries. Compare to VAST: VAST owns everything at Layer 1C (Retained by VAST, Ceded by the enterprise). One authority but total vendor dependency. HPE’s Layer 1C authority model is distinct: HPE curates and supports, the enterprise can substitute, NVIDIA acceleration is additive not required.
The acquisition history (BlueData 2018, MapR 2019, Ampool 2021, Arrikto 2023) shows deliberate multi-year investment in the data pipeline layer. HPE chose to build this capability rather than delegate entirely to partners. Dell’s Dataloop acquisition is a similar strategic move but more recent (2024) and narrower in scope. The NVIDIA RAPIDS Accelerator for Spark (up to 29x faster) is meaningful but optional: the AI Essentials Spark runs without GPU acceleration. Same pattern as VAST’s DataEngine (runs on standard CNodes, CNode-X adds GPU acceleration). Data Fabric’s real-time S3-to-S3 movement bridges Layer 1A and 1C: ingest from external S3-compatible sources into the governed namespace, where policy-based placement takes over. Re-sourced October 5, 2026: the cell named HPE Ezmeral Unified Analytics, end of sale October 31, 2024 (docs.ezmeral.hpe.com release notes, which point customers to HPE Private Cloud AI). The same stack is documented in HPE AI Essentials Software 2026.07.1 on Private Cloud AI, so moderate holds; Feast is not in the current documentation and is dropped. Placement (October 16, 2026): HPE Zerto Software reads here under the function audit and the hardware-management placement rule. HPE Morpheus is "Born to provide simple and consistent provisioning of VMs, Containers, and Applications into any private or public cloud" (QuickSpecs); its migration moves VMs, not data, so it isn't scored at 1C. Out of the HPE Data Fabric Software chip under the GA evidence table (October 24, 2026): Agentic AI assistant for automated reporting and data placement decisions, Data lineage and compliance tagging (The HPE Data Fabric 8.0 and 8.1 what's-new pages document an MCP server, Object Copier Hub, inline erasure coding (instant tiering) and object lifecycle, but no agentic placement assistant and no lineage or compliance tagging. hpe.com and buy.hpe.com (403s and timeouts) couldn't be reached to check product-page claims.). Split by engine (October 24, 2026): HPE AI Essentials reads at 1C as four chips, Airflow, Spark, Kubeflow Pipelines, and EzPresto, each on HPE's 2026.07.1 documentation. Ray and MLflow read at Layer 2B in the ML Runtime chip. Superset is visualization, not data movement, and isn't scored here.
GPU scheduling, quotas, RBAC, fair-share scheduling, utilization optimization
HPE's hybrid cloud control plane (4th gen): unified VM and Kubernetes management, a dashboard for capacity, utilization, and cost, metering, and self-service lifecycle management across HPE infrastructure. The pay-per-use contract it's often bought under isn't a component (Keith, October 16, 2026); this control-plane software is what's scored. Proprietary management platform: orchestration opinions captive to HPE.
HPE-owned agentic AI framework infused across the entire hybrid stack (not a standalone product). Multiple domain-specific LLMs trained on HPE data, communicating via Model Context Protocol (MCP). Agents form an agentic mesh for inter-agent communication with secure, contextual data sharing. Domain agents for networking (Aruba/Juniper), storage (Alletra), compute (OpsRamp), orchestration, and FinOps. Cross-domain correlation: traces performance issues across application to storage to network chain. Can process real-time infrastructure metrics and execute actions across multiple vendor environments. Human-in-the-loop: agents take action subject to approval. Proprietary agentic framework captive to HPE platform.
Multi-domain agentic system coordinating compute, network, storage, virtualization, and software layers. Use cases: root-cause analysis, explainability, capacity planning. AI-driven alerts, incident management, GPU monitoring, workload observability. Operations copilot with conversational product help and agentic command center. CrowdStrike integration for security monitoring. MCP support for connecting to GreenLake Intelligence and third-party tools. Proprietary management platform captive to HPE.
Cloud-native server lifecycle management for the ProLiant fleet. Compute Copilot for AI-assisted infrastructure operations. A fleet management plane sold and run on its own across many servers, so it's a 2A component (Keith, October 16, 2026); the iLO in each ProLiant server is that server's management half and reads with it at Layer 0. Server settings, firmware baselines, and policies are HPE-proprietary and don't lift to another OEM's plane: Ceded.
K8s management with ProLiant Gen12. Unified cloud-native + virtualized workload management. Independent scaling for cloud-native workloads. Upgrade path to Morpheus for hybrid/multicloud.
Intent-based automation for the data center fabric, multi-vendor by design: blueprints rendered to device configuration, continuous validation and rollback. A fleet plane across many switches, the 2A facet of the Juniper chip, read as Nokia's EDA and Arista's CloudVision are. Its blueprints and intents are Apstra objects: Ceded.
GPU fractionalization for multi-tenancy in AI Factory portfolio.
AI Factory at-scale management. Planned later 2026. GPU cluster ops, scheduling, resource allocation.
Lifecycle management for AI software stack. Pre-integrated with ProLiant.
Layer 2A is where HPE makes its most substantial software authority claim. GreenLake Intelligence is not a rebranded monitoring tool. It is an agentic AI framework with domain-specific LLMs communicating via MCP, designed to be infused across the entire HPE hybrid stack. Four characteristics define HPE’s Layer 2A position: (1) Cross-domain agentic correlation. GreenLake Intelligence agents trace performance issues across application → storage → network chains, coordinating remediation across domains. Dell’s OpenManage and NVIDIA’s Run:ai operate within single domains (server fleet management and GPU scheduling respectively). VAST’s Polaris orchestrates VAST clusters but not the broader infrastructure around them. (2) MCP as the inter-agent communication standard. GreenLake Intelligence is compliant with MCP, enabling connection to third-party agents and devices. The X10000's native MCP servers, scored with the array at Layer 1A, extend the same protocol to storage. This means the agentic mesh is architecturally open: ITSM systems can collaborate with GreenLake, and third-party infrastructure can be brought under GreenLake management. HPE positions this as ‘the mesh is open to more stitches.’ (3) Multi-vendor infrastructure support. NAND Research notes that GreenLake Intelligence agents can process real-time metrics and execute actions across multiple vendor environments, not exclusively HPE hardware. This extends Layer 2A authority beyond HPE’s own equipment, a broader orchestration scope than Dell’s OpenManage (Dell hardware only) or VAST’s Polaris (VAST clusters only). (4) FinOps agent for workload placement. Orchestration, networking, and FinOps agents collaborate to determine workload placement across private and public clouds. This is an economic placement decision: where should this workload run based on cost, performance, and policy? This function overlaps with Layer 2C territory. Dell’s Layer 2A is split between Dell-managed rack deployment (OpenManage) and NVIDIA-managed GPU scheduling (Run:ai). HPE’s Layer 2A is unified under GreenLake with OpsRamp providing a multi-domain agentic system that coordinates compute, network, storage, virtualization, and software layers. GreenLake is often bought on a pay-per-use contract. The contract isn't a component (Keith, October 16, 2026); the metering, capacity management, and utilization software under it is, and it's scored on the GreenLake Cloud Platform chip. Who holds authority in a customer's actual incident under that contract is day-two operational authority, which the instrument names but doesn't score yet. The gap: GPU-specific scheduling is still Ceded to NVIDIA (MIG for fractionalization, Mission Control for at-scale management, planned later 2026). HPE orchestrates the infrastructure around the GPU cluster; NVIDIA orchestrates inside it. This is the same Layer 2A boundary as Dell, but HPE’s surrounding orchestration is a unified platform rather than separate point tools. Capability and deployment are separate readings. The grade states what an architect can deploy on this vendor's paper; it does not assert that a given customer consumes it. HPE hardware and GreenLake are frequently bought underneath a customer's own Kubernetes platform, OpenShift being the common one, in which case GreenLake governs the infrastructure estate while the container and AI-workload control plane belongs to somebody else. The GreenLake Intelligence claim is about a capability HPE offers, and the enterprise decides how much of the layer it actually hands over.
Low for infrastructure orchestration (GreenLake platform, GreenLake Intelligence, OpsRamp, Compute Ops Management are HPE-owned IP). Moderate for GPU-specific scheduling (MIG, Mission Control are NVIDIA-controlled). The NAND Research caveat is relevant for the DAPM assessment: tight coupling with GreenLake creates potential vendor lock-in for organizations with diverse infrastructure portfolios. However, MCP compliance and multi-vendor agent support partially mitigate this concern: the agentic mesh can extend beyond HPE hardware. Compare to Dell: Dell’s infrastructure orchestration is fragmented (OpenManage for servers, separate tools for storage and networking). GPU scheduling is fully NVIDIA-controlled (Run:ai). No agentic cross-domain correlation. Compare to VAST: Polaris provides fleet-level VAST cluster orchestration (Retained by VAST). DataEngine provides workload scheduling within the data platform. But Polaris doesn’t orchestrate non-VAST infrastructure. GreenLake Intelligence’s multi-vendor, multi-domain scope is broader.
GreenLake Intelligence’s cross-domain correlation and FinOps-aware placement push beyond traditional Layer 2A into Layer 2C territory for IT operations. The assessment classifies it as spanning 2A–2C for IT ops: infrastructure orchestration (2A) + cross-domain governance decisions and economic placement reasoning (2C). This dual classification is important: GreenLake Intelligence is both an orchestrator and a decision-maker. The MCP openness is architecturally significant: by using an open protocol for agent communication, HPE enables third-party integration without custom APIs. ITSM systems (ServiceNow, BMC) can collaborate with GreenLake agents. Third-party infrastructure can be managed. This is an open-ecosystem approach to infrastructure orchestration that Dell’s proprietary OpenManage and VAST’s proprietary Polaris don’t provide. The X10000 native MCP servers represent infrastructure-level agent communication: the storage array itself participates in the agentic mesh as a first-class agent endpoint, not just a managed resource. They're built into one storage system, so they read with the X10000 chip at Layer 1A, not as a 2A component (Keith, October 16, 2026). This is a specific implementation of the 4+1 model’s vision of infrastructure that is natively agent-aware. Morpheus Central is documented as of October 5, 2026 (support.hpe.com, HPE Morpheus Central Overview; Morpheus 9.0.1 release notes) and deepens a cell that's already strong. Watch-list (HPE Discover 2026, not scored): intent-based closed-loop network automation and GreenLake Intelligence agent orchestration and copilots, rolling out through 2027. Hardware-management scope (Keith, October 16, 2026): Compute Ops Management is the fleet plane (2A, Ceded). OpsRamp and GreenLake Intelligence are vendor-sold day-two tools, read at 2A like any component. iLO reads with ProLiant at Layer 0 and the X10000 MCP servers with the X10000 at Layer 1A. The strong grade rests on GreenLake Intelligence, OpsRamp, and Morpheus and doesn't move. Open question for Keith: Aruba Central, Juniper Mist, and Apstra sit inside their networking chips at Layer 0; whether a network-device fleet plane follows this rule to 2A or stays with the fabric at 0B is unruled. Placement (October 16, 2026): Juniper Mist reads here under the hardware-management placement rule, as a device-fleet plane. Placement (October 16, 2026): Aruba Central reads here under the hardware-management placement rule, as a device-fleet plane. Named, not scored at 2A (October 19, 2026, the Arista scoping ruling): Juniper Mist and Aruba Central, device-fleet management for campus and branch switches and access points; nothing an AI architect deploys runs through them. Their fabric reads at 0B.
Model serving, agent execution, inference APIs, distributed inference
Co-engineered with NVIDIA. Pre-configured HW+SW stack with four right-sized configurations. Air-gapped capable. Scales to 128 GPUs with network expansion racks. OpsRamp integration for AI workload monitoring. Supports NVIDIA AI-Q, Omniverse, NeMo Retriever blueprints. Multi-tenancy via MIG with GPU passthrough (Spring 2026). The NVIDIA runtime is integral to the stack: the enterprise cannot substitute an alternative inference runtime without rebuilding.
Kubeflow for ML pipeline execution, Katib experiments and model serving through KServe, documented in HPE AI Essentials Software 2026.07.1 on Private Cloud AI, the successor to HPE Ezmeral Unified Analytics (end of sale October 31, 2024). Enterprise packaging of open-source tools: the enterprise could swap to another packaging of Kubeflow.
CrewAI integration enables enterprises to build multi-agent solutions on Private Cloud AI. Deloitte Zora AI for Finance deploys on Private Cloud AI as an agentic platform for dynamic executive reporting (financial statement analysis, scenario modeling, competitive analysis). NIM Agent Blueprints provide pre-built agentic workflows. This is an emerging multi-framework agentic surface, not HPE-owned runtime IP but HPE-curated deployment options. Not scored within this chip under the GA evidence table: CrewAI and Deloitte Zora AI (no HPE page shows them offered on HPE's paper).
In Town of Vail: coordinates multiple specialized agents: accessibility identification, alt text generation, remediation guidance formatting. Determines which agents run, in what sequence, with what inputs, under what constraints. Enforces human-in-the-loop checkpoints. Manages agent lifecycle at the execution layer. ‘Coordination of multiple specialized AI agents and human workflows to execute multi-step decisions under explicit authority, governance, and audit constraints.’ Lift-to-leave: the execution-coordination opinions are Kamiwaza-captive, Ceded to the chosen partner. HPE sells it inside the HPE Agentic Smart City Solution, which "runs on HPE Private Cloud AI," with "Kamiwaza provides agentic orchestration and decision routing" and the system "supported and evolving as a single integrated capability" (HPE announcement, October 28, 2025; HPE-sponsored CTO Advisor report a00156405enw, January 2026).
Ray for distributed compute: tasks, actors and objects, the Ray libraries, GPU support and job submission, in HPE AI Essentials Software 2026.07.1. Open-source substrate: Ray workloads move to any other Ray.
MLflow for experiment tracking, model logging and the model registry, with RBAC on experiments, in HPE AI Essentials Software 2026.07.1. Open-source substrate: tracking and registry data move to any other MLflow.
The AI workload runtime: model serving (Triton), guardrails (NeMo Guardrails), distributed inference, training frameworks (NeMo). Pre-integrated on Private Cloud AI. STIG-hardened, FIPS-enabled for sovereign deployments.
Pre-optimized inference microservices for model deployment. Part of AI Enterprise platform. Available to all NVIDIA partners (Dell, Cisco, Lenovo), not HPE-specific.
Pre-built agentic AI application patterns: Multimodal PDF Data Extraction, Digital Twins (Omniverse), AI-Q for enterprise data agents. Deployed on Private Cloud AI. Same blueprints available on Dell AI Factory, Cisco HyperFabric, Lenovo Hybrid AI.
Layer 2B reveals a more nuanced runtime architecture than Dell’s because authority is distributed across three actors rather than two. The three-actor model: • NVIDIA provides model execution: Triton (model serving), NeMo Guardrails (safety), NIM (optimized inference), NeMo (training). This is the compute execution layer. Identical across all NVIDIA partners (Dell, Cisco, Lenovo deploy the same stack). • HPE provides the deployment platform: Private Cloud AI (hardware, cooling, lifecycle), AI Essentials (ML runtime packaging, successor to Ezmeral Unified Analytics), and increasingly an agentic framework surface (CrewAI, Deloitte Zora AI). This is infrastructure + curation. • Kamiwaza provides agent execution coordination (via Unleash AI): determines which agents run, sequences execution, manages inputs/outputs, enforces execution-time constraints (authority boundaries, audit, human-in-the-loop). This is governance-aware agent coordination above model inference but below Layer 2C policy. This creates a layered runtime: NVIDIA executes individual model inference → Kamiwaza coordinates multi-agent workflows and enforces execution governance → Layer 2C (also Kamiwaza) makes policy decisions about what should run where. The 2B/2C boundary: 2B is execution coordination (how agents run), 2C is decision authority (why agents run, under what governance). The structural comparison across vendors: • Dell: NVIDIA at 2B (model execution + NemoClaw/OpenShell agent runtime). No agent coordination layer beyond NVIDIA. Dell provides packaging and services. • HPE: NVIDIA at model execution + Kamiwaza at agent coordination + CrewAI/ISV frameworks for agent building. Three layers of runtime capability from three sources. HPE provides infrastructure + curation. • VAST: AgentEngine provides a unified agent runtime (execution + coordination + lifecycle + observability) as VAST IP. NVIDIA provides GPU acceleration only. One authority. HPE’s ‘NVIDIA AI Computing by HPE’ branding signals co-engineering, but the DAPM question is precise: can HPE modify, extend, or replace NVIDIA runtime components independently? The answer appears to be no: ‘co-engineering’ means deeper integration and joint validation, not shared IP authority. NVIDIA controls the runtime; HPE controls the platform it runs on. The emerging agentic framework ecosystem (CrewAI, Deloitte Zora AI) on Private Cloud AI is worth noting: HPE is becoming a multi-framework agentic deployment surface, not locked to a single agent runtime. This is a platform strategy: provide the substrate that multiple agentic frameworks can run on, rather than a runtime strategy (build the definitive agent runtime, as VAST is attempting with AgentEngine).
High for AI workload runtime. NVIDIA controls model serving, inference optimization, guardrails, and training frameworks. The same NVIDIA AI Enterprise stack runs on Dell, Cisco, and Lenovo. This is not HPE-specific technology. The mitigating factor is the ‘bracketing’ architecture: HPE provides governance at Layer 2A (GreenLake Intelligence, HPE-owned) and sources Layer 2C governance from a chosen partner (Kamiwaza via Unleash AI). The NVIDIA-controlled Layer 2B runtime is sandwiched between two governance layers. The enterprise has governance coverage even where it doesn’t control execution, coverage, not authority: both bracket surfaces are Ceded from the customer's seat, one to HPE and one to Kamiwaza. Compare to Dell: Dell has NVIDIA at 2B with no governance brackets. No Layer 2C (gap). Layer 2A is split between Dell (OpenManage) and NVIDIA (Run:ai). The enterprise has neither governance authority above nor unified governance authority below the NVIDIA runtime. Compare to VAST: VAST owns AgentEngine (2B) and is building PolicyEngine (2C). No bracketing needed because VAST controls both the runtime and the governance layer. The enterprise Cedes both to VAST. HPE’s borrowed judgment at 2B is the highest of any layer in the HPE assessment. The bracketing architecture is the mitigation, not the solution. Decision authority: the runtime is NVIDIA's, and it reads as it does on NVIDIA's row. The model makes the runtime call inside it, and the enterprise's own tool code, executing client-side and able to refuse, gates the effect: model decides, visible, overridable, Delegated. Portability is a separate reading, carried by the components.
The CrewAI and Deloitte Zora AI integrations signal that Private Cloud AI is evolving from a single-stack NVIDIA deployment platform into a multi-framework agentic surface. This is architecturally different from both Dell’s approach (NVIDIA-only runtime) and VAST’s approach (proprietary-only runtime). HPE is positioning Private Cloud AI as the substrate that multiple agent frameworks deploy on. The NIM Agent Blueprints (PDF Extraction, Digital Twins, AI-Q) are available identically on Dell AI Factory, Cisco HyperFabric, and Lenovo Hybrid AI. These do not differentiate HPE at Layer 2B. HPE’s differentiation comes from the bracketing architecture (2A and 2C governance around the NVIDIA 2B runtime) and the emerging multi-framework agent deployment model. The bracketing architecture has a structural analog in HPE’s networking story: NVIDIA InfiniBand handles GPU-to-GPU interconnect (HPE doesn’t control it), but HPE’s Juniper/Aruba/Slingshot handles everything around the GPU fabric (HPE owns it). The pattern: cede the NVIDIA-specific function, retain authority over everything surrounding it. Out of the HPE AI Essentials ML Runtime on Private Cloud AI chip under the GA evidence table (October 24, 2026): Ray, MLflow and MLIS (not named in the HPE AI Essentials 2026.07.1 documentation). Out of the Agentic Framework Ecosystem on Private Cloud AI chip under the GA evidence table (October 24, 2026): CrewAI and Deloitte Zora AI (no HPE page shows them offered on HPE's paper). Out of the AI Essentials ML Runtime chip by the bundle split (October 24, 2026): MLIS, which the 2026.07.1 documentation doesn’t name. The earlier note that Ray and MLflow weren’t named was wrong: the documentation has Ray and MLflow sections (read October 24, 2026).
Policy-driven placement and resource coordination — the Autonomy Layer
MCP-based agent communication across infrastructure domains. Domain-specific agents for networking, storage, compute, operations. Cross-domain correlation and autonomous remediation. Layer 2C for IT infrastructure operations, but not for AI workload placement and policy. Proprietary agentic framework captive to HPE platform.
Agentic orchestration and decision routing for AI workloads. Policy-driven placement, mission decomposition, decision authority placement, cross-agent governance. Distributed Data Engines process data at the source without moving it or compromising security. Cross-environment evaluation: rather than treating anomalies as isolated alerts, Kamiwaza evaluates what else is happening across the environment to determine appropriate response. Town of Vail production agents: ARIA (accessibility auditing: independently audits websites, identifies Section 508 issues, provides developer fixes in days vs $1.5M and months for manual audits). Deed restriction processor (reviews documents spanning 60 years, extracts key data, answers compliance questions, generates Excel/PDF reports, work that previously required weeks of manual review). Fire detection coordinator (works with Vaidio/ProHawk video AI, evaluates cross-environment context, triggers workflows, supports operators as conditions change). HPE's chosen Layer 2C for AI workload orchestration, delivered under single-accountable-provider model. Lift-to-leave: the reasoning-plane opinions (orchestration flows, decision routing, ReBAC governance configurations) are Kamiwaza-captive and cannot be swapped without rebuilding. Deliberate choice and single accountability describe the procurement, not the exit: Ceded to the chosen partner, as on Kamiwaza's own row. HPE sells it inside the HPE Agentic Smart City Solution, which "runs on HPE Private Cloud AI," with "Kamiwaza provides agentic orchestration and decision routing" and the system "supported and evolving as a single integrated capability" (HPE announcement, October 28, 2025; HPE-sponsored CTO Advisor report a00156405enw, January 2026).
GreenLake Intelligence and Kamiwaza are HPE-owned and partner-provided (Ceded to Kamiwaza) respectively. NVIDIA does not control the governance, placement, or policy reasoning layer in the HPE stack.
This is the most analytically interesting layer in the HPE assessment and where HPE’s approach diverges most from Dell’s. HPE has a two-part Layer 2C story: GreenLake Intelligence provides Layer 2C for IT infrastructure operations: correlates signals across networking, storage, compute to diagnose and resolve infrastructure issues. Routes decisions across domains. Takes autonomous action (subject to approval). Uses MCP for agent communication. HPE-owned IP. Kamiwaza provides Layer 2C for AI workload orchestration: agentic orchestration, decision routing, policy-driven placement, cross-agent governance. Distributed Data Engines process data at the source without data movement. Not an accidental ISV partnership: HPE’s deliberate architectural choice, curated, integrated, validated, and delivered under single-accountable-provider model. The Town of Vail as by-proxy Kamiwaza assessment, specific evidence: • ARIA accessibility agent: independently audits municipal websites, identifies Section 508 issues, provides developer fixes. Manual equivalent: $1.5M, months of work. ARIA delivers in days. This demonstrates decision automation with governance (the agent identifies what needs fixing, recommends how, but human developers implement). • Deed restriction processor: reviews housing documents spanning 60 years in disjointed legacy formats, extracts key data, answers compliance questions, generates reports. Previously required weeks of manual review and data entry in Excel. Single processing errors carry serious legal and financial consequences. This demonstrates governance-aware document intelligence with cross-departmental implications (legal, housing, administrative). • Fire detection coordinator: works with Vaidio video analytics and ProHawk enhanced vision. Rather than treating a video anomaly as an isolated alert, Kamiwaza evaluates what else is happening across the environment and determines the appropriate response. Agents surface relevant information, trigger correct workflows, and support operators as conditions change. This demonstrates cross-environment evaluation, the core Layer 2C pattern of reasoning across multiple data sources and agent outputs. • Deployment velocity: concept to first-phase production in three months. 20–30 additional use cases projected in first year. Additional use cases compose from existing primitives (decision flows, authority boundaries, governance constraints) rather than requiring new infrastructure. • Economic model: fixed-cost infrastructure on the town’s own solar/wind-powered data center. No cloud API token pricing. Billions of tokens without variable costs. • RBAC → REBAC governance: emerged from Kamiwaza’s production behavior. Traditional role-based access breaks when autonomous agents operate across department boundaries. Relationship-Based Access Control constrains agent permissions based on context, not just role. Structural comparison, by shape rather than roster. Some vendors ship no 2C and the enterprise owns the function by default. Some productize Intelligence-2C themselves and cede that authority to their own platform. HPE's position is the third shape and the rarer one: it splits the layer, keeping IT-operations 2C in HPE-owned GreenLake Intelligence and sourcing AI-workload 2C from a partner, so the enterprise cedes to the chosen partner rather than to HPE. That is validated in production with named agents and measurable outcomes, which most 2C claims on the instrument cannot show. Nobody, HPE included, ships Infrastructure-2C as a configurable product.
IT ops Layer 2C: GreenLake Intelligence is HPE-owned IP, low borrowed judgment for HPE, but the customer's ops-reasoning opinions are captive to the HPE platform. AI workload Layer 2C: High, Ceded to Kamiwaza. Deliberately chosen, integrated, and delivered under HPE's accountability, but the platform's opinions (orchestration flows, decision routing, ReBAC configurations) are Kamiwaza-captive and cannot be swapped without rebuilding. Structurally stronger than Dell's position on the capability axis because the function exists and someone is accountable; the dependency is captive. The trade is capability-with-capture versus Dell's no-capability-no-capture.
Strategic question: is partner-provided Layer 2C transitional (HPE eventually builds/acquires orchestration IP) or permanent (HPE’s value is ecosystem curation, not owning every layer)? Town of Vail evidence suggests HPE is comfortable with the ecosystem model, and that it works operationally. The RBAC → REBAC governance shift from Town of Vail validates the 4+1 model’s claim that Layer 2C requires governance architecturally distinct from Layer 2A infrastructure RBAC. Watch-list (HPE Discover 2026, not scored): native agent registry + governance via GreenLake Intelligence, and Private Cloud AI secure cross-framework agent registration - core GA July 2026, agentic observability/data intelligence Q4 2026. Directionally shifts HPE 2C from Kamiwaza-provided toward HPE-native, but it is Intelligence-2C (governance), not placement. Revisit at GA. Instrument note: live Infrastructure-2C (a policy engine deciding at request time where inference runs relative to the data, which model serves the request, and how cost, latency, and compliance are arbitrated) does not ship from any vendor on this instrument. It is recorded here as a universal finding about the market, not as a charge against this vendor. The sibling finding applies here as well: you cannot prompt your way to deterministic output. The controls credited in this cell answer whether an action is legal (permissions, schema and policy checks, rate and token limits, approval gates), and none of them validates whether the outcome was right under a graduated escalation policy. Controls that work by instructing the model shift the output distribution without pinning it. That deterministic outcome-validator is missing instrument-wide, and like live placement it is noted as universal rather than charged to one vendor. To notes under the GA evidence table (October 24, 2026): HPE Data Fabric Policy Engine (no owner evidence of a Data Fabric policy engine that places data on performance, sovereignty, cost and compliance).
AI-powered business capabilities — business logic, workflow automation
Pre-built AI application patterns (Multimodal PDF Extraction, Digital Twins, AI-Q) and inference microservices. Deployed on Private Cloud AI. Same blueprints available across all NVIDIA partners.
HPE sells no first-party AI business applications. Since June 12, 2026, HPE has sold independent software vendor (ISV) applications on its own paper through the GreenLake Marketplace: "customers buy GreenLake Marketplace software from HPE," and "HPE manages invoicing, payments, and settlement with the ISV" (HPE, GreenLake Marketplace launches end-to-end commerce capabilities). The marketplace's AI and machine learning section lists business applications such as Aible, the Vaidio AI Vision Platform and Razor DataQI. That's selling. It isn't support: HPE describes curating and validating the software, and no public HPE source found says HPE supports the applications it sells. A Layer 3 partner application counts where the vendor both sells it and supports it, so the layer reads gap, the same as the hyperscaler marketplaces, where the enterprise buys on the provider's paper but the ISV supports the app. The rest of the ecosystem doesn't change that. Unleash AI is a validation program (each ISV "undertakes a comprehensive onboarding process, including validation testing"). Deloitte Zora AI for Finance is announced as a joint solution with no HPE SKU found. HPE Private Cloud AI is the platform applications run on, and CrewAI is a framework; both are read at 2B. Where HPE sells and supports a partner's capability inside a named HPE solution, as with Kamiwaza in the HPE Agentic Smart City Solution, it scores at the layer that capability serves (1B, 1C, 2B and 2C), not here. Calibration: Dell and Supermicro read gap for the same reason, with no applications on their paper. Cisco, Arista and Nokia read moderate on first-party security applications. Public evidence that moves the cell: an HPE support statement covering GreenLake Marketplace applications, or a first-party HPE business application.
None on HPE’s paper that HPE stands behind. The enterprise can buy ISV applications from HPE, but each application’s judgment and its support come from the ISV, so the reading is Absent: HPE offers nothing at this layer’s function that it supports.
Named, not scored: HPE Unleash AI (validation program), Deloitte Zora AI for Finance (joint solution, March 2025, no HPE SKU found), Kamiwaza and other members sold through TD SYNNEX, CrewAI pre-installed on Private Cloud AI (framework, 2B), QuantPi, BigID, Securiti and DagsHub on the marketplace (tooling, not business applications). Marketplace purchasing is US-only (June 12, 2026). Out of the cell under the Layer 3 sells-and-supports rule (Keith, October 24, 2026): ISV AI applications bought from HPE on the GreenLake Marketplace (Aible, Vaidio AI Vision Platform, Razor DataQI), sold on HPE’s paper (US customers with an active GreenLake service) but with no public HPE support statement.
GTC 2026, HPE GreenLake/storage May 2026 announcements, Discover 2025, Juniper acquisition, Town of Vail whitepaper, analyst coverage. v1.6: vendor name updated from 'HPE AI-Native Infrastructure' to 'HPE Private Cloud AI'. v1.7: Kamiwaza components at 1B/1C/2B/2C moved Delegated→Ceded under the instrument's retraction of the OEM-channel DAPM rule (a proprietary single-implementation platform is Ceded-to-partner through any paper; partnership substitutability claims fail the lift-to-leave litmus), calibrated to Kamiwaza's own row, which scores the same surfaces Ceded. 2C statusLabel moved to capability vocabulary. Bracketing narrative corrected: the capability advantage over Dell survives; the authority reading is capability-with-capture versus no-capability-no-capture. No capability grades moved. v1.8 (July 12, 2026, /reconcile): 1A moderate→strong under the whose-paper and frontier rules (owned Alletra/Data Fabric/Polaris/Zerto depth; drift vs. Lenovo 1A strong resolved). 1B/2B statusLabels moved to capability vocabulary. 2C prose corrected: Google’s productized 2C is Ceded per the GCP row, not Retained. /reconcile (September 1, 2026): GX5000 moved Retained to Ceded under the integrated-system rule (blade chassis on a captive fabric); component renamed from a property to the product. K3000 Retained upheld on QuickSpecs a00154232ENW: community-owned BSD-2 DAOS on commodity servers behind POSIX/SMB/NFS/S3, distinguishing it from proprietary-OS storage (ONTAP, VAST) which stays Ceded. /reconcile (September 1, 2026): retired status vocabulary replaced with gap. /reconcile (September 1, 2026): shared 2C findings stated explicitly (live-placement universality, outcome-validation finding). /reconcile (September 1, 2026): hard-coded peer-grade ledger replaced with a shape-based reading; the roster drifted from the rows it described and went stale on every peer move. /reconcile (September 1, 2026): the availability-not-occupancy reading applied at 2A: the grade states deployable capability on this vendor's paper and no longer reads as though the vendor's layer is automatically the customer's. /reconcile (October 5, 2026): Layer 3 calibration citations updated for the Layer 3 re-reads (Nutanix, VAST, VMware to partner; CoreWeave, NVIDIA, AMD to gap) and the MongoDB BI ruling; no grade or authority change on this row. /reconcile (October 5, 2026): Layer 1A re-read against the catalog-classification criterion (the methodology's queued item). Strong held, now resting on X10000 Data Intelligence (inline metadata and vector enrichment, GA January 2026, added as a Ceded component) plus Data Fabric, in place of storage-management agents and owned-platform depth; sensitivity-classification depth flagged as inferred pending the datasheet. /reconcile (October 5, 2026): 2B authority re-read from {decides:vendor,visible:true,overridable:false,boundary:vendor,direction:Ceded} to model / visible / overridable / Delegated, matching NVIDIA's row for the same NVIDIA AI Enterprise runtime (the open-source seam and tool-call ruling, September 2, 2026). Components and grade unchanged. /reconcile (October 5, 2026): 1C authority column (rulings of September 15 and October 4, 2026). 1C re-sourced from HPE Ezmeral Unified Analytics (end of sale October 31, 2024) to HPE AI Essentials Software 2026.07.1 on Private Cloud AI, which documents the same Airflow, Spark, Kubeflow, Ray, and MLflow stack; Feast dropped. Authority vendor / Ceded to code / Retained: the open stack carries the grade and runs the enterprise's DAGs and jobs, and Data Fabric's placement alone wouldn't hold moderate. Grade unchanged. /reconcile (October 5, 2026): authority visible set from the cell's own text where it settles it: layer1a true. Direction, grades, and components unchanged. /ga-check (October 5, 2026): Morpheus Central confirmed documented and moved from the watch-list to notes; the network automation and agent orchestration halves stay watched. Grade unchanged. /reconcile (October 5, 2026): 2B ML runtime chip re-sourced from HPE Ezmeral Unified Analytics (end of sale October 31, 2024) to HPE AI Essentials Software 2026.07.1 on Private Cloud AI, which documents Kubeflow, Ray, MLflow, and MLIS. Grade, DAPM, and authority unchanged. /reconcile (October 5, 2026): calibration text citing NetApp's AI Data Engine updated after its documentation was withdrawn and NetApp's 1A moved to moderate. Text only; grades and readings unchanged. /reconcile (October 5, 2026): authority read against product docs: layer1b visible true. Grades and components unchanged. v1.21 (October 15, 2026, catalog-classification storage re-read): layer1a strong to moderate. X10000 Data Intelligence documents metadata enrichment and vector embeddings (datasheet PSN1014922454AEEN, QuickSpecs July 6, 2026) but no classification, sensitivity, or lineage; Keith ruled that enrichment without documented classification doesn't meet the bar of a catalog that classifies the data. Data Fabric Polaris reads as a table catalog. Calibration paragraph updated for Dell 1A moderate. Authority and components unchanged; component details corrected. v1.22 (October 16, 2026, hardware-management scope ruling): a controller embedded in one system is that system's management half and reads with it; a fleet management plane sold and run on its own across many systems is a 2A component; a vendor-sold day-two tool reads at 2A like any component; a consumption contract isn't a component, so the control-plane software under it is scored (Keith, October 16, 2026, after the AWS, Dell, and HPE pressure tests). HPE Alletra X10000 MCP Servers (Native) moved from layer2a into the X10000 chip at layer1a: they're built into one storage system. Compute Ops Management stays at layer2a as the ProLiant fleet plane, Ceded on proprietary server settings and firmware baselines; iLO reads with ProLiant at Layer 0. OpsRamp and GreenLake Intelligence stay at layer2a as day-two tools. GreenLake Cloud Platform chip re-worded to score the control plane, not the pay-per-use contract. Grades, authority, and the other chips unchanged. v1.23 (Layer 0 sub-layers, October 16, 2026): Layer 0 split into 0A Compute, 0B Network Fabric and 0C Storage Substrate by runtime decision (Keith, October 16, 2026); storage left 1A for 0C, and 1A, now Data Foundation & Governance, is graded on the catalog that classifies the data. 0A strong, 0B strong, 0C strong; summary edits (summary worksheet, Keith, October 16, 2026). v1.24 (October 19, 2026, /reconcile partner ruling): Layer 3 holds partner on ISV applications bought from HPE on the GreenLake Marketplace (HPE, June 12, 2026: customers buy marketplace software from HPE, which invoices and settles with the ISV). Unleash AI is validation and is named, not scored; Private Cloud AI and CrewAI moved to 2B prose; Deloitte Zora named. Keith challenged a draft that read the layer gap from Unleash AI alone: GreenLake commerce is procurement on HPE paper. (Same pass) 0C: a Retained Lustre-core facet added under the Cray E2000, the DDN reading (DDN follow-up 8). 2A: Juniper Mist and Aruba Central chips moved to notes on the campus-and-branch scoping ruling (Keith, October 19, 2026); a Ceded Apstra fleet-plane facet added, the EDA and CloudVision reading. Grades and authority unchanged. (Same pass) Plain-language pass: methodology rule numbers in the prose and audit notes replaced with the reasoning they stood for; no grade, chip or authority reading changed. v1.25 (October 19, 2026, /reconcile voice pass): em-dashes in the prose replaced with periods, colons, commas or parentheses; wording, grades, chips and authority readings unchanged. v1.26 (October 21, 2026, /reconcile): 0C K3000 split like the E2000, from reconcile's peer calibration. HPE's K3000 product page (buy.hpe.com, read October 21, 2026) documents the system as "fully orchestrated and managed by K3000 management stack" on ProLiant nodes, so the system reads Ceded on HPE's management layer and factory integration, and the open-source DAOS engine moves to its own chip, Retained, keeping the HPE DAOS reading. Grade and cell authority unchanged; the cell rests on the Alletra arrays. v1.27 (October 21, 2026, /reconcile): 0C door chip added for the E2000's Lustre client front end, Delegated, from reconcile's peer calibration: DDN carries the same door, and DDN's row names the Lustre client across DDN, HPE and the clouds' managed Lustre services. Grade and cell authority unchanged; the cell rests on the Alletra arrays. v1.28 (October 22, 2026, /reconcile): 0B Juniper detail drops the networking revenue figures, which are earnings data and not fabric evidence, and records Juniper's validated AI data center design: three fabrics (front end, GPU back end, storage back end) with DCQCN on QFX and PTX switches. Found by the 0B fabric research. Grade and authority unchanged. v1.29 (October 23, 2026, /reconcile): the HPE Cray GX5000 and its GX440n, GX350a, GX240 and GX250 blades move to the 0A watch list, dated early 2027 by HPE's November 2025 release. The multi-silicon blade chip comes out at 0A, and its fabric facet comes out at 0B. The Cray chips and their Slingshot facet read as the EX4000. The AI Factory's Vera Rubin date goes to the watch list it was already on. The 0B map follows. Found by the 0A compute research. Grade and authority unchanged. v1.30 (October 23, 2026, /reconcile): 0A adds the NVIDIA GB300 NVL72 by HPE (Ceded, on its QuickSpecs) with a 0B NVLink facet, and the HPE Compute XD690 with eight HGX B300 (Retained, on its QuickSpecs). The GX5000 watch note quotes HPE's release directly and separates the GX240, which rests on NVIDIA's Vera CPU. The confidence flag on unchecked candidates closes. Grade and authority unchanged. v1.31 (October 23, 2026, /reconcile): 0A: The map had no accelerator platform for the EX4000. 0B: The brand-umbrella chip, split into products by function. Found by the Layer 0 data pass after the sniff read. Grade and cell authority unchanged. v1.32 (October 23, 2026, /reconcile): the 0C X10000 chip names its protocols, NFS v4.1 (with NFS over RDMA from v2.1.0.0) and S3-compatible APIs, from HPE's release notes (sd00007906en_us, October 6, 2026). Found by the Layer 0 sniff read. Grade and cell authority unchanged. v1.33 (October 23, 2026, /reconcile): the 0C B10000 chip names its host protocols, Fibre Channel and iSCSI, from HPE's installation guide (sd00002405en_us, Edition 13, August 2026). NVMe/TCP and NVMe/FC aren't in that document, so they're not added. Found by the Layer 0 sniff read. Grade and cell authority unchanged. v1.34 (October 23, 2026, /reconcile): Layer 1 research pass. 1B: rewritten (sold inside a named HPE solution). 1C: rewritten (sold inside a named HPE solution). 2B: rewritten (sold inside a named HPE solution). 2C: rewritten (sold inside a named HPE solution). 1C: "HPE Morpheus Software" removed (Morpheus is "Born to provide simple and consistent provisioning of VMs, Containers, and Applications"; its migration is VM migration, not data movement). 1B: "HPE Alletra X10000 RDMA Storage" removed (X10000 RDMA storage is the storage retrieval reads from, not retrieval). Found by the Layer 1 slot cut. Grade and cell authority unchanged. v1.35 (October 23, 2026, storage door pass): X10000 and B10000 get front-end doors; the arrays stay Ceded. Grade unchanged. v1.36 (October 23, 2026, Layer 1 managed-services pass): HPE Data Fabric Software v8.1 gets a door. Grade unchanged. v1.37 (October 24, 2026, /reconcile): 1B Data Fabric retrieval-surface chip removed: Data Fabric 8.1 documents MCP catalog access, scored at 1A, and no conversational retrieval. Found in the 1B dependency audit under the headend standard. Grade unchanged. v1.38 (October 24, 2026, /reconcile): 1B note updated: Kamiwaza's GA-build documentation supports the Kamiwaza chip the cell rests on. Found in the 1B dependency audit follow-ups. Grade unchanged. v1.39 (October 24, 2026, /reconcile): Layer 1B under the GA evidence table version 2: GA evidence on every chip, preview items and unaffirmed chips to notes. Grade unchanged. v1.40 (October 24, 2026, /reconcile): GA evidence recorded on 2 of 3 Layer 1A chips under the GA evidence table version 2. Grade unchanged. v1.41 (October 24, 2026, /reconcile): GA evidence recorded on 2 of 4 Layer 1C chips under the GA evidence table version 2. Grade unchanged. v1.42 (October 24, 2026, /reconcile): GA evidence recorded on 9 of 10 Layer 0B chips under the GA evidence table version 2. Grade unchanged. v1.43 (October 24, 2026, /reconcile): GA evidence recorded on 6 of 7 Layer 0A chips under the GA evidence table version 2. Grade unchanged. v1.44 (October 24, 2026, /reconcile): GA evidence recorded on 10 of 12 Layer 0C chips under the GA evidence table version 2. Grade unchanged. v1.45 (October 24, 2026, /reconcile): GA evidence recorded on the HPE AI Factory and X10000 Data Intelligence chips, read on hpe.com. Grade unchanged. v1.46 (October 24, 2026, /reconcile): GA evidence recorded on 4 of 6 Layer 2A chips under the GA evidence table version 2. Grade unchanged. v1.47 (October 24, 2026, /reconcile): GA evidence recorded on 1 of 4 Layer 2B chips under the GA evidence table version 2. Grade unchanged. v1.48 (October 24, 2026, /reconcile): GA evidence recorded on 1 of 1 Layer 3 chips under the GA evidence table version 2. Grade unchanged. v1.49 (October 24, 2026, /reconcile): GA evidence recorded on the Kamiwaza Smart City chips at 2B and 2C (composition). Grade unchanged. v1.50 (October 24, 2026, /reconcile): GA evidence table version 2.1 applied: 8 chips recorded or narrowed, 1 to notes. Grade unchanged. v1.51 (October 24, 2026, /reconcile): HPE AI Essentials split by engine at 1C (Airflow, Spark, Kubeflow Pipelines, EzPresto); Ray and MLflow read at 2B, Superset unscored. Grade unchanged. v1.52 (October 24, 2026, /reconcile): Layer 2 bundle split: HPE AI Essentials ML Runtime split by engine, with Ray and MLflow restored. Grade unchanged. v1.53 (October 24, 2026, /reconcile): Layer 3 from partner to gap under the Layer 3 sells-and-supports rule; the GreenLake Marketplace chip moved to notes. v1.54 (October 27, 2026): each component now carries a durable offering id; nothing else changed.