Databricks is a data + AI lakehouse platform that floats above Layer 0 (it owns no silicon and runs on AWS, Azure, and GCP) and is strong across the entire data and application stack: data and governance (1A), retrieval (1B), pipelines (1C), runtime (2B), and the value plane (3). It moderates only where it does not own the layer: infrastructure orchestration (2A), where it rents cloud compute and its serverless GPU runtime is still in preview. The reasoning plane (2C) reads strong on a complete GA governance plane, though its placement routing is still Beta. Its authority sits in the opinion layer, never the substrate.
The capture mechanism is decoupled and invisible, and Databricks is its purest expression in this series. Every openness claim is true: Delta Lake is open source (Linux Foundation), the data sits in the customer's own cloud bucket in open formats (Delta, Iceberg, Parquet), Spark and MLflow are open source, Unity Catalog was donated to open source, and Delta Sharing is an open protocol. But the value the enterprise accumulates lives in the managed services those open pieces sit beneath: managed Unity Catalog governance (lineage, audit, RBAC, masking, none of which the open-source catalog has), the closed-source Photon engine, Lakeflow declarative pipelines, Mosaic AI serving and agents, and the notebooks and workspace. The open formats keep the bytes portable; the opinions (governance policies, pipelines, agents, dashboards) do not lift.
The DAPM profile makes this exact. Of 25 scored components, 20 are Ceded, 5 are Delegated, and none are Retained. The five Delegated components are precisely the open-source and open-protocol surfaces (open lakehouse storage, Spark, MLflow, Marketplace via Delta Sharing, and Lakebase's PostgreSQL interface); everything that carries a Databricks opinion is Ceded. The enterprise never operates an open substrate itself within Databricks. Even the open formats are read and written through Databricks' managed runtime. This places Databricks in the proprietary-captive cluster with Palantir and VAST, and it shares Palantir's specific mechanism: open storage, captive opinion layer, so the commitment is invisible until you try to leave, and it compounds with every pipeline, governed asset, and agent built.
Databricks is the strongest data-plane vendor in the series: Layers 1A, 1B, and 1C are all strong, with 1C (Lakeflow + Spark + Photon) the most mature data-engineering layer assessed. Its Layer 2B runtime is strong and, unlike the on-prem platform vendors (VAST, Nutanix, VMware, Dell: all moderate at 2B), its agent runtime is generally available rather than preview, which calibrates it to the hyperscalers. The ceiling: it owns no Layer 0 (a gap, like Palantir), and its orchestration is scoped to its own workloads on rented cloud compute, with the serverless GPU runtime in preview (2A moderate). Its reasoning plane reads strong on a complete GA plane (identity, gateway, registry, orchestration, observability), while placement routing, the reasoning half, is still Beta.
The buyer's trade is architectural coherence and a best-in-class data + AI platform in exchange for ceding the entire opinion layer to Databricks under an open-formats banner. You keep your data in open formats in your own cloud, genuinely, and you accumulate governance, pipelines, retrieval, models, agents, and dashboards that are Databricks-shaped and Databricks-bound. Databricks owns the lakehouse and the AI stack on top of it. It does not own the silicon beneath, and its routing of where inference runs is still Beta.
Layer-by-layer status: Layer 0A (Not Databricks' Layer (By Design)), Layer 0B (Not Databricks' Layer (By Design)), Layer 0C (Not Databricks' Layer (By Design)), Layer 1A (Lakehouse Strength: Open Formats, Captive Governance), Layer 1B (Lakehouse-Native Governed Retrieval), Layer 1C (Data Engineering Heartland), Layer 2A (Managed Workload Orchestration; Serverless GPU Pre-GA), Layer 2B (Mosaic AI Runtime: Serving + Agents (GA)), Layer 2C (Complete GA Agent-Governance Plane; Placement Routing Beta), Layer 3 (+1) (First-Party Data Apps + Marketplace Ecosystem).
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: v1.29 - 4+1 v3: Layer 0 Sub-Layers.
Databricks is a data + AI lakehouse platform that floats above Layer 0 (it owns no silicon and runs on AWS, Azure, and GCP) and is strong across the entire data and application stack: data and governance (1A), retrieval (1B), pipelines (1C), runtime (2B), and the value plane (3). It moderates only where it does not own the layer: infrastructure orchestration (2A), where it rents cloud compute and its serverless GPU runtime is still in preview. The reasoning plane (2C) reads strong on a complete GA governance plane, though its placement routing is still Beta. Its authority sits in the opinion layer, never the substrate.
The capture mechanism is decoupled and invisible, and Databricks is its purest expression in this series. Every openness claim is true: Delta Lake is open source (Linux Foundation), the data sits in the customer's own cloud bucket in open formats (Delta, Iceberg, Parquet), Spark and MLflow are open source, Unity Catalog was donated to open source, and Delta Sharing is an open protocol. But the value the enterprise accumulates lives in the managed services those open pieces sit beneath: managed Unity Catalog governance (lineage, audit, RBAC, masking, none of which the open-source catalog has), the closed-source Photon engine, Lakeflow declarative pipelines, Mosaic AI serving and agents, and the notebooks and workspace. The open formats keep the bytes portable; the opinions (governance policies, pipelines, agents, dashboards) do not lift.
The DAPM profile makes this exact. Of 25 scored components, 20 are Ceded, 5 are Delegated, and none are Retained. The five Delegated components are precisely the open-source and open-protocol surfaces (open lakehouse storage, Spark, MLflow, Marketplace via Delta Sharing, and Lakebase's PostgreSQL interface); everything that carries a Databricks opinion is Ceded. The enterprise never operates an open substrate itself within Databricks. Even the open formats are read and written through Databricks' managed runtime. This places Databricks in the proprietary-captive cluster with Palantir and VAST, and it shares Palantir's specific mechanism: open storage, captive opinion layer, so the commitment is invisible until you try to leave, and it compounds with every pipeline, governed asset, and agent built.
Databricks is the strongest data-plane vendor in the series: Layers 1A, 1B, and 1C are all strong, with 1C (Lakeflow + Spark + Photon) the most mature data-engineering layer assessed. Its Layer 2B runtime is strong and, unlike the on-prem platform vendors (VAST, Nutanix, VMware, Dell: all moderate at 2B), its agent runtime is generally available rather than preview, which calibrates it to the hyperscalers. The ceiling: it owns no Layer 0 (a gap, like Palantir), and its orchestration is scoped to its own workloads on rented cloud compute, with the serverless GPU runtime in preview (2A moderate). Its reasoning plane reads strong on a complete GA plane (identity, gateway, registry, orchestration, observability), while placement routing, the reasoning half, is still Beta.
The buyer's trade is architectural coherence and a best-in-class data + AI platform in exchange for ceding the entire opinion layer to Databricks under an open-formats banner. You keep your data in open formats in your own cloud, genuinely, and you accumulate governance, pipelines, retrieval, models, agents, and dashboards that are Databricks-shaped and Databricks-bound. Databricks owns the lakehouse and the AI stack on top of it. It does not own the silicon beneath, and its routing of where inference runs is still Beta.
CPUs, accelerators, servers, integrated systems, OS and hypervisor, physical plant
ML training and inference run on NVIDIA GPUs, but only as mediated by the hyperscaler's IaaS (AWS/Azure/GCP). There is no structural Databricks-NVIDIA dependency the way the on-prem vendors (Dell, VAST, Nutanix) have. The GPU relationship belongs to the cloud. Serverless GPU ('AI Runtime') is Public Preview, not GA.
Layer 0 is not Databricks' layer, by design. Databricks owns no silicon, no networking, and no datacenters: it is a software platform layered entirely on AWS, Azure, and GCP IaaS, split into a control plane (Databricks' own cloud accounts) and a compute plane (classic clusters in the customer's own VPC, or serverless in Databricks' account). The buyer never thinks about Layer 0, and that is the value proposition. The consequence for the 4+1 model is the same as Palantir's: a Databricks adoption decision resolves no Layer 0 authority question. Whatever capture exists at the silicon and fabric layer belongs to the chosen hyperscaler (a different row on this map), not to Databricks. The contrast with the infrastructure vendors is clean: Dell and Cisco are strong here because they own or design hardware; VAST and Nutanix are moderate as software-defined/HCI abstraction layers; Databricks, like Palantir, simply floats above. The one captive compute-adjacent asset, Photon (the closed-source vectorized execution engine), is a query/processing engine rather than silicon or fabric, and is scored at Layer 1C where it does its work, not here.
Total at Layer 0, and irrelevant to the value proposition by design. Databricks inherits all silicon, networking, and acceleration judgment from the host cloud. The enterprise's Layer 0 authority position is set by its hyperscaler choice (a different vendor's row), and adopting Databricks does not, by itself, resolve it.
Watch-list (Preview, not scored): serverless GPU / 'AI Runtime' (A10/H100): Public Preview. Classic GPU clusters are long-GA but run on the cloud's GPU instances. Photon is named and scored at Layer 1C, not here.
Interconnects, switching and routing, network virtualization
Assessment pending
Databricks Data Intelligence Platform sells no network fabric to the customer; the one under its service is its own or its providers', out of the customer's view.
The service absorbs this sub-layer and the customer gets no view into it: vendor decides, not visible, not overridable, Ceded.
Arrays, storage servers and software, front-end protocols: where the data sits
Assessment pending
Databricks Data Intelligence Platform sells no storage substrate to the customer; the one under its service is its own or its providers', out of the customer's view.
The service absorbs this sub-layer and the customer gets no view into it: vendor decides, not visible, not overridable, Ceded.
The data itself, organized and governed: the Governance Catalog that Layer 2C queries
Single governance layer across data and AI assets: column-level lineage, audit logging, RBAC, ABAC, row-filtering and column-masking, data products, and the catalog namespace. The governance opinions are a proprietary managed-service surface and do not lift: the open-source Unity Catalog donation lacks lineage, audit, RBAC, and masking. Proprietary Databricks platform: opinions captive, no open exit.
Open-source Delta Lake (Linux Foundation), native managed Iceberg read/write plus an Iceberg REST catalog and UniForm interop, and the Delta Sharing open protocol, with data physically in the customer's own cloud object storage. The Delta Lake and Apache Iceberg surface is its own chip. The service behind it, its operations, management and configuration, is Databricks' and doesn't lift: Ceded.
Serverless PostgreSQL (built on Neon) integrated with the lakehouse: database branching and sync to Delta. GA on AWS; Azure Beta; GCP not yet. The PostgreSQL surface is its own chip. The service behind it, its operations, management and configuration, is Databricks' and doesn't lift: Ceded. Not scored within this chip under the GA evidence table: sync to Delta (Lakebase Change Data Feed, Postgres changes stored as Delta) ("Lakebase Change Data Feed (CDF) is now in Public Preview").
Delta Lake and Apache Iceberg have independent implementations: code and definitions written to them 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.
PostgreSQL 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. Standard PostgreSQL clients, schemas and SQL move to another PostgreSQL by dump or replication; Lakebase's branching, retained history and Delta sync don't. The service behind it is its own chip, Ceded.
Unity Catalog, Delta Lake, and lineage are Databricks IP or open source. NVIDIA contributes nothing to the governance layer.
This is the Lakehouse, and it is Databricks' center of gravity. The buyer gets a unified governed data foundation: Delta Lake tables in their own cloud bucket, Unity Catalog governing every data and AI asset (lineage, audit, access control, ABAC), native managed Iceberg read/write plus an Iceberg REST catalog and UniForm interop, and Delta Sharing for cross-organization sharing. It feels maximally open: your data, open formats, your cloud account, share with anyone, no proprietary storage. This is the textbook decoupled capture, and every openness claim is true and is exactly what hides the commitment. The data is genuinely open: Delta is open source (Linux Foundation), Iceberg is an open standard, Parquet sits in the customer's own S3/ADLS/GCS. But the governance is captive: managed Unity Catalog is where the value accumulates (column-level lineage, audit, RBAC, row-filtering and column-masking, data products, the catalog namespace), and none of that ships in the open-source Unity Catalog donation, which has the APIs but not the lineage, audit, RBAC, masking, or UI. The methodology names this precisely: a metadata-governance catalog layered beyond the open format is Ceded. Data portability is a decoy here; the dependence accumulates in the catalog and compounds with every governed asset, policy, and lineage edge. The calibration is Palantir's Layer 1A almost line for line: open formats Delegated, the storage service and governance Ceded. Both are software-platform peers with the same decoupled-capture shape, and Databricks is, if anything, more open on the data side (real open-source Delta in the customer's own bucket versus Palantir's Ontology-mediated virtual tables), which sharpens rather than softens the finding. Peer also to VAST and Dell at strong, but those are coupled (your data lives in their storage) where Databricks and Palantir are decoupled.
Low for the governance logic: Unity Catalog, lineage, and ABAC are Databricks IP (Ceded to Databricks). The open formats are a Delegated door (open-source formats in the customer's own cloud); the managed storage service behind them reads Ceded. The capture is decoupled: data open, governance captive. Decision authority holds vendor / Delegated, on classification rather than policy evaluation: policies the enterprise writes are only executed (code, on their own), but Data Classification is an agentic system that infers what the data is from the data, and each wrong detection carries a per-detection Exclude, the per-object override (judgment test and defaults clause, October 4 and 5, 2026).
The open-source vs. managed Unity Catalog gap is the capture proof point: OSS UC provides core APIs (Hive-compatible + Iceberg REST) but lacks automatic column-level lineage, audit logging, RBAC/row-filtering/column-masking, and the Catalog Explorer UI. The governance enterprises actually depend on is managed-only. Watch-list (not scored): Lakebase Azure (Beta) and GCP (not yet); external write to managed Delta tables (Public Preview); Governance Hub (Private Preview); RBAC in OSS UC (coming). Managed Iceberg, Iceberg REST, Iceberg v3, and external lineage reached GA at Data + AI Summit 2026. Out of the Lakebase chip under the GA evidence table (October 24, 2026): sync to Delta (Lakebase Change Data Feed, Postgres changes stored as Delta) ("Lakebase Change Data Feed (CDF) is now in Public Preview").
Low-latency retrieval for RAG — vector/hybrid search, context windows
GA. Vector and hybrid (reciprocal-rank-fusion) search over Delta-synced indexes; managed or BYO embeddings; Unity Catalog-governed permission-aware retrieval. A proprietary managed retrieval engine: the index, API, and sync pipeline do not lift. Proprietary Databricks platform: opinions captive, no open exit.
Embeddings and search run on cloud compute; GPU acceleration is incidental and hyperscaler-mediated. Vector Search and hybrid retrieval are Databricks IP.
RAG without a separate vector database. The buyer gets Mosaic AI Vector Search (GA): index Delta tables directly, governed by Unity Catalog, with hybrid keyword + vector retrieval (GA) and managed-or-BYO embeddings, all serverless. Because the index is synced from governed Delta tables and access flows through Unity Catalog, retrieval is permission-aware by construction. A Feature & Function Serving path (Online Feature Store, Lakebase-backed) is in Public Preview with no GA affirmation and sits in the notes. The decoupled split repeats one layer up: the source data (Delta) is open, but the vector index and retrieval engine are proprietary and captive. Mosaic AI Vector Search's index format, retrieval API, and sync-from-Delta pipeline do not lift: leaving means rebuilding RAG on Pinecone, Weaviate, or pgvector and re-wiring permission propagation. The score calibrates to VAST and Palantir at Layer 1B, both strong for native, governed, permission-aware retrieval owned by the platform. Databricks matches: GA native vector search, hybrid retrieval, UC-governed permission inheritance, and MLflow 3 GenAI evaluation/tracing for retrieval-quality observability. It sits above Dell, VMware, and Nutanix at 1B (all moderate), whose retrieval is Delegated (Elastic, pgvector). Databricks' is proprietary, native, and GA.
Low: Vector Search and hybrid retrieval are Databricks IP, governed by Unity Catalog so retrieval inherits the Layer 1A permission model. Embeddings are optionally Databricks-served or BYO; NVIDIA is not required. Decision authority under the override rule (September 4, 2026): Vector Search ranks, but each query sets query_type (ann, hybrid, or full_text), filters, and result count. The reranker is Beta and isn't counted. Per-request overrides: vendor decides, visible, overridable, Delegated, the Elastic and SAP reading.
Online tables are deprecated; the replacement Online Feature Store is Lakebase-backed and therefore inherits Lakebase's cloud-gating (AWS GA, Azure Beta, GCP not yet). Mosaic AI Vector Search GA May 2024; hybrid search (RRF) GA Aug 2024. GA evidence table (October 24, 2026): the Online Feature Store goes to notes, announced in Public Preview September 5, 2025 with no GA affirmation since; it returns on one.
Move/transform data — ETL/ELT, lineage, cost-aware movement, KV cache tiering
GA. Declarative ETL/ELT with built-in data-quality expectations, incremental processing, and a managed runtime. A proprietary declarative framework: pipelines do not lift; the open-source Spark Declarative Pipelines donation is not yet GA. Proprietary Databricks platform: opinions captive, no open exit.
GA. Managed ingestion connectors (SQL Server, Salesforce, Workday, ServiceNow GA; SharePoint, PostgreSQL, Oracle in Preview). Proprietary managed connectors: ingestion opinions are captive. Proprietary Databricks platform: opinions captive, no open exit. Not scored within this chip under the GA evidence table: PostgreSQL, Oracle ("The PostgreSQL connector is in Public Preview. Contact your Databricks account team to request access.").
The open-source distributed processing engine and streaming framework (DBR 17.x = Spark 4.0), run by Databricks with the Databricks Runtime packaging and Photon acceleration. The Spark API surface is its own chip. The service behind it, its operations, management and configuration, is Databricks' and doesn't lift: Ceded.
Incremental and streaming file ingestion from cloud storage. A proprietary Databricks Runtime feature, not part of open-source Spark, so it does not lift. Proprietary Databricks platform: opinions captive, no open exit.
Proprietary closed-source C++ engine delivering 3-10x over JVM Spark SQL; the execution engine beneath pipelines and Databricks SQL. Off Databricks the customer falls back to open-source Spark on the JVM with no Photon. Proprietary Databricks platform: opinions captive, no open exit.
The Spark API 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.
Pipelines run on CPU compute; Photon is a CPU-vectorized engine. No structural NVIDIA dependency at the pipeline layer.
This is Databricks' birthplace and arguably its single strongest layer. The buyer gets Lakeflow, the unified pipeline platform, all GA since June 2025: Lakeflow Connect (managed ingestion connectors), Lakeflow Declarative Pipelines (formerly Delta Live Tables, declarative ETL/ELT with built-in data-quality expectations and incremental processing), and Lakeflow Jobs (orchestration). Plus Auto Loader for incremental file ingestion, Apache Spark and Structured Streaming as the engine, Photon accelerating it, dbt integration, and Unity Catalog column-level lineage threaded through all of it. For data engineering, this is best-in-class. The decoupled split is sharp here. Spark is open source and portable: you can run it anywhere. But the value-bearing pipeline opinions are captive: Lakeflow Declarative Pipelines is a proprietary declarative framework and managed runtime (its internals are being donated to Apache Spark as Spark Declarative Pipelines, but that is not yet GA in open-source Spark: the product you run today is proprietary); Auto Loader is a proprietary Databricks Runtime feature, not in open-source Spark; Photon is a closed-source C++ engine; and Lakeflow Connect connectors and Unity Catalog lineage are proprietary. You can lift raw Spark jobs, but your declarative pipelines, ingestion connectors, data-quality rules, incremental logic, and lineage graph do not lift without rebuilding. The score is peer to VAST at strong (DataEngine) and above Dell at moderate (Dataloop). Databricks essentially defines this category: the most mature data-pipeline platform in the instrument, and the most unambiguous strong in the row.
Low: Lakeflow, Auto Loader, Photon, and Unity Catalog lineage are Databricks IP. Spark is the open-source substrate (Delegated), but the Databricks Runtime packaging, Photon acceleration, and declarative-pipeline layer are proprietary. Decoupled pattern: open Spark API, captive pipeline and runtime opinions. Decision authority reads vendor / Delegated under the override rule (ratified September 4, 2026): the Lakeflow runtime chooses incremental versus full recomputation and the execution plan inside pipelines the enterprise declared, and per-table flow types, triggered versus continuous mode, and full-refresh actions are overrides the engine must honor. Portability stays Ceded on the declarative layer.
Lakeflow (Connect + Declarative Pipelines + Jobs) reached GA June 2025. Watch-list (not scored): Spark Declarative Pipelines (the DLT core donated to Apache Spark) is not yet GA in OSS Spark; Lakeflow Connect connectors vary by GA (SQL Server, Salesforce, Workday, ServiceNow GA; SharePoint, PostgreSQL, Oracle in Preview). DBR 17.x ships Spark 4.0. Out of the Lakeflow Connect chip under the GA evidence table (October 24, 2026): PostgreSQL, Oracle ("The PostgreSQL connector is in Public Preview. Contact your Databricks account team to request access.").
GPU scheduling, quotas, RBAC, fair-share scheduling, utilization optimization
Autoscaling clusters in the customer's VPC (classic) or in Databricks' account (serverless, GA on all clouds); proprietary provisioning and lifecycle management on rented cloud VMs. Cluster and serverless orchestration opinions do not lift. Proprietary Databricks platform: opinions captive, no open exit.
GA. Multi-task workflow DAGs with scheduling, retries, and dependencies across notebooks, pipelines, and SQL. Job and workflow definitions are Databricks-specific. Proprietary Databricks platform: opinions captive, no open exit.
GA. Infrastructure-as-code (YAML) for jobs, pipelines, notebooks, dashboards, and serving endpoints. The CLI is open-sourced, but the bundle opinions deploy to Databricks: the deployment target does not lift. Proprietary Databricks platform: opinions captive, no open exit.
GPU scheduling depends on the underlying cloud; Databricks' own serverless GPU ('AI Runtime') is Public Preview, not GA. No structural NVIDIA orchestration dependency.
Fully managed compute orchestration the buyer never operates. They spin up autoscaling clusters (into their own VPC, or fully serverless, GA across all three clouds), run Lakeflow Jobs/Workflows with complex dependency DAGs, and codify it as infrastructure-as-code via Asset Bundles (now Declarative Automation Bundles). No VM lifecycle, no scheduler to run: Databricks handles provisioning, autoscaling, and job orchestration on top of the cloud. But this is proprietary orchestration on rented cloud IaaS. The cluster manager, Jobs scheduler, serverless layer, and bundle definitions are Databricks IP: job graphs, cluster configs, and orchestration opinions do not lift; leaving means rebuilding on the cloud's native orchestration or Kubernetes. And the AI-relevant part of this layer, GPU scheduling and fair-share, Databricks does not own: its serverless GPU is Public Preview, not GA, so dedicated GPU scheduling is still the hyperscaler's. The score calibrates to VAST at 2A (moderate): VAST owns infrastructure orchestration (Polaris) but GPU scheduling is partial; Databricks owns workload orchestration (Jobs, clusters, serverless) but GPU scheduling is pre-GA and it runs on the cloud's compute. It sits below VMware and Nutanix at 2A (strong), which are mature general-purpose orchestration platforms managing all infrastructure. Databricks orchestrates only its own data and AI workloads. This is where the data-plane strong streak correctly stops: 2A is not Databricks' to own.
Moderate. Databricks owns the workload orchestration (clusters, Jobs, serverless, Asset Bundles: Databricks IP, Ceded to Databricks), but the underlying compute capacity and GPU fair-share scheduling are the hyperscaler's: Databricks provisions cloud VMs and its own GPU scheduler is pre-GA. Orchestration authority is Databricks'; capacity and GPU-scheduling authority are the cloud's. Decision authority under the override rule (September 4, 2026): classic compute carries the grade, and the enterprise can resize a running cluster through the Resize cluster API and pin a job to an existing cluster (existing_cluster_id), the documented resize and reassign the rule names. Serverless sizing is Databricks' call but doesn't carry the grade. Vendor decides, visible (Spark UI, job run history), overridable, Delegated.
Watch-list (Preview, not scored): serverless GPU / 'AI Runtime'. Lakeflow Jobs (formerly Workflows) and serverless compute on all three clouds are GA. Asset Bundles GA since 2024; CLI is open-sourced but bundle definitions deploy to Databricks.
Model serving, agent execution, inference APIs, distributed inference
GA. Real-time and batch model serving with unified OpenAI-compatible endpoints for hosted (Llama, Gemma, GPT-OSS) and external (Claude, GPT, Gemini) models. A managed service behind a genuine multi-vendor standard interface: integration opinions built against the endpoints lift to any compatible provider, so the serving interface is Delegated. The capture sits in the surfaces beyond the interface: the AI Gateway policy layer (scored Ceded at 2C) and Unity Catalog governance bindings, not in the serving contract itself.
GA (~Mar 2025). Author and deploy agents with tool-calling and evaluation, governed by Unity Catalog. Agent authoring and runtime opinions are Databricks-specific. Proprietary Databricks platform: opinions captive, no open exit.
GA (Jan-Feb 2026). Turnkey agents plus multi-agent orchestration (Supervisor Agent). Proprietary turnkey agents: captive. Proprietary Databricks platform: opinions captive, no open exit.
Open-source (Apache 2.0) experiment tracking, model registry, and GenAI tracing (MLflow 3). Core MLflow APIs are open and portable; the managed-MLflow governance and monitoring are the captive add-on. The open core is Delegated.
Serving and training run on cloud NVIDIA GPUs (e.g. managed H100), but there is no NIM/NemoClaw-equivalent software dependency. Unlike Dell, where NVIDIA owns the 2B runtime, Databricks owns its runtime and NVIDIA is just the underlying GPU.
A complete, GA model-serving and agent runtime. The buyer serves any model: their own, open source (Llama, Gemma, GPT-OSS), or external (Claude, GPT, Gemini), behind unified OpenAI-compatible endpoints via Mosaic AI Model Serving and Foundation Model APIs (both GA); builds agents with the Mosaic AI Agent Framework (GA); deploys turnkey agents via Agent Bricks (Knowledge Assistant, Document Intelligence, and the multi-agent Supervisor Agent: all GA); and tracks everything in MLflow (including MLflow 3 GenAI tracing). All governed by Unity Catalog. The decoupled pattern reaches the execution layer: open models, captive runtime. Model choice is genuinely open (any OSS or external model), and MLflow is open source (experiment and model tracking lift). But the serving infrastructure, the Agent Framework, and the Agent Bricks products are proprietary Databricks managed services: agents and endpoints built here do not lift. The score calibrates to the hyperscalers at 2B (strong: comprehensive managed runtimes), not to the on-prem platform vendors. The deciding factor under the GA-gate: Databricks' agent runtime is GA (Agent Framework plus several Agent Bricks), whereas Nutanix's agents are preview, Dell's runtime is NVIDIA-owned, VAST's AgentEngine is newer and less proven, and VMware's is foundational. Databricks owns a GA serving and agent stack, the strong bar the clouds set; managed fine-tuning has reached end of life and was never the deciding factor.
Low. Serving, Agent Framework, and Agent Bricks are Databricks IP (Ceded to Databricks); MLflow is open source (Delegated). Models are open or external (portable choice); the runtime is captive, and unlike the on-prem peers, the runtime is Databricks' own (not NVIDIA's) and is GA, not preview.
Watch-list (Beta/Preview, not scored): Agent Bricks Data + AI Summit 2026 additions (Agent Memory, Sandbox, MCP Catalog, Contextual Policies), Omnigent; serverless-GPU training path (Preview). The old Foundation Model Fine-tuning is deprecated (removal Aug 2026), folded into Mosaic AI Model Training; DBRX dropped from the Foundation Model APIs list. To notes under the GA evidence table (October 24, 2026): Customer-Created and Managed Tools (Function Calling / Client-Side) (Databricks' current function-calling doc (updated Oct 19, 2026) still describes the feature as in Public Preview, and the chip rests on that function-calling path. ("During Public Preview, function calling on Databricks is optimized for single turn function calling")). To notes under the GA evidence table (October 24, 2026): Mosaic AI Model Training & Fine-Tuning (Databricks has ended support for its managed fine-tuning product and says the UI and package are no longer available; it points to AI Runtime instead. ("Foundation Model Fine-tuning has reached end of life (EoL) and is no longer supported.")).
Policy-driven placement and resource coordination — the Autonomy Layer
GA. Agents, tools, and models governed as Unity Catalog assets; every tool and data call is permission-checked against the requesting user's UC entitlements: the agent is governed like the user. Proprietary Databricks platform: opinions captive, no open exit. Not scored within this chip under the GA evidence table: On-Behalf-Of Auth ("User authorization is in Public Preview . Your workspace admin must enable it before you can use user authorization.").
GA. Rate-limiting, usage tracking, payload and inference-table audit logging, fallbacks, and traffic-splitting in front of serving endpoints. This is static routing configuration, not per-request placement reasoning. Proprietary Databricks platform: opinions captive, no open exit.
GA (Feb 2026). Coordinates Genie spaces, Knowledge Assistant agents, MCP servers, Unity Catalog functions, and custom agents under UC governance. Multi-agent orchestration, not infrastructure placement. Proprietary Databricks platform: opinions captive, no open exit.
Agent governance and orchestration are Databricks IP. NVIDIA controls no identity, governance, or routing here: the same pattern as the clouds' 2C.
Databricks governs agents with the same authority it governs data. Unity Catalog agent governance (GA) registers agents, tools, and models as governed assets, and each agent runs under its own identity: app authorization, GA, gives every agent an automatically created service principal with consistent permissions. On-Behalf-Of auth, which checks each call against the requesting user's permissions, is Public Preview and isn't scored. The Mosaic AI Gateway for serving endpoints (GA) adds rate-limiting, usage tracking, payload and inference-table audit logging, fallbacks, and traffic-splitting. And the Agent Bricks Supervisor Agent (GA, Feb 2026) coordinates multiple agents (Genie, Knowledge Assistants, MCP servers, Unity Catalog functions, custom agents) under Unity Catalog governance. Applying the 'Routing Is Not Reasoning' test: what ships GA is Intelligence-2C (governance of which agent may act, on what, under which policy) plus multi-agent orchestration. What does not ship is Infrastructure-2C: live, per-request placement reasoning. The AI Gateway's routing is static configuration (fallback order and traffic-split weights), not per-request cost/quality/compliance arbitration. Unity AI Gateway itself went GA on August 4, 2026, and it carried rate limits, spend caps, MCP-server governance including tool filtering, asset and traffic governance, and cost observability with it. What stayed behind is the part this cell turns on. Smart Routing, the engine that would place each request against quality, cost, performance, availability, and budget, is explicitly Beta, as are service policies and the built-in PII, prompt-injection, and unsafe-content guardrails now folded into them. The routing mechanism is published rather than announcement-only, so this is no longer an evidence gap. It is a shipping gap. There is still no GA engine that reasons per-request and places inference. The score lands in the moderate cohort: AWS (AgentCore Policy/Guardrails), IBM (watsonx.governance + Bob), OCI, Cisco, HPE, all of which have productized, multi-component Intelligence-2C governance without live placement. Governance here is unusually deep because it inherits the full Unity Catalog authority (On-Behalf-Of), the same structural-governance property that earned Palantir and VAST credit. It sits above the Nutanix/VMware/Dell/VAST gap cohort because it adds GA multi-agent orchestration and UC-native agent identity rather than a lone gateway, and below Azure/Google/Palantir (strong) because the placement engine is still Beta and there is no shipping Infrastructure-2C placement reasoning. The August 4, 2026 Gateway GA does not close that distance: it hardened the governance and cost-control surface this cell already credited, and left Smart Routing behind. Re-graded October 5, 2026 under the Layer 2C strong criterion (September 15, 2026): strong is a reasoning mechanism or a complete GA plane, and Databricks has the complete plane, the route Google Cloud took. Identity is the per-agent service principal; the gateway is Unity AI Gateway, GA August 4, 2026; the registry is Unity Catalog; cross-agent orchestration is the Supervisor Agent; observability is MLflow tracing of production agents. Smart Routing, the placement-reasoning leg, is still Beta, so the label names it. Routing isn't reasoning, but the cell no longer needs it for strong.
Low for what is provided: Unity Catalog agent governance, the AI Gateway, and the Supervisor Agent are Databricks IP (Ceded to Databricks), with no NVIDIA dependency. But this is low borrowed judgment for partial 2C: Intelligence-2C (governance and orchestration) is productized and GA; Infrastructure-2C (live placement reasoning) is a gap, not borrowed. The live per-inference placement gap is universal across the instrument, noted rather than penalized.
Watch-list (Beta, not scored): Unity AI Gateway Smart Routing (per-request selection on quality, cost, performance, availability, and budget), Beta; service policies, including the built-in PII, prompt-injection, and unsafe-content guardrails folded into them, Beta. Everything else in the Gateway went GA August 4, 2026: rate limits, spend caps, MCP-server governance with tool filtering, asset and traffic governance, usage tracking, cost observability. Smart Routing at GA would add a placement mechanism to a cell that's already strong on the complete plane. 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. Not adopted (October 5, 2026): the identity sub-test logged in Snowflake's 2C notes (an agent scoped below the delegating user, per task, at the effect boundary). Databricks' down-scoped delegation is Public Preview, but the sub-test isn't part of the ratified criterion, and adopting it would have to be applied to Google Cloud and Azure as well. Out of the Unity Catalog Agent Governance chip under the GA evidence table (October 24, 2026): On-Behalf-Of Auth ("User authorization is in Public Preview . Your workspace admin must enable it before you can use user authorization.").
AI-powered business capabilities — business logic, workflow automation
GA. Natural-language analytics over governed data (Genie) and the default BI dashboards layer. Proprietary first-party applications on Unity Catalog and the lakehouse: they do not lift. Proprietary Databricks platform: opinions captive, no open exit.
GA. Serverless hosting for custom data apps (Streamlit, Dash, Flask, React) in the workspace, with Unity Catalog-governed data access. The app code uses open-source frameworks (portable), but the hosting runtime and UC integration are captive. Proprietary Databricks platform: opinions captive, no open exit.
GA. Datasets, models, notebooks, apps, and MCP servers, built on the open Delta Sharing protocol. An ecosystem of substitutable third-party providers over an open sharing protocol: Delegated. Not scored within this chip under the GA evidence table: MCP servers ("List MCP servers in Databricks Marketplace (Public Preview)").
The value-plane applications are Databricks IP or ecosystem-provided. NVIDIA contributes nothing here.
Unlike the platform-enabler vendors, Databricks ships genuine first-party value-plane applications, not just tooling. The buyer gets AI/BI Genie (GA: ask questions of governed lakehouse data in natural language), AI/BI Dashboards (GA: the default BI layer), Databricks Apps (GA: build and host custom data apps serverlessly in the workspace), Databricks One (workspace-GA: a business-user experience), Databricks Assistant (GA), and the Databricks Marketplace (GA: datasets, models, notebooks, apps, and MCP servers, built on Delta Sharing). Real business capability, GA today. The first-party apps (Genie, AI/BI, One, Apps) are proprietary and built on Unity Catalog and the lakehouse: they do not lift; the value plane is captive even though the data beneath is open. The Marketplace is the open exception (Delta Sharing protocol, substitutable third-party providers). The value plane is also domain-bounded: data, analytics, and BI-centric (Genie, AI/BI) rather than the broad operational business applications of Palantir's Foundry or the hyperscalers' breadth. The score is above VMware and Nutanix at Layer 3 (partner: platform enablement earns no Layer 3 grade, and the layer arrives through their ISV ecosystems), whereas Genie is an application, not a platform service. It is not partner (Dell, Cisco, HPE), because Databricks' Layer 3 is not addressed entirely through an ISV ecosystem. Closest to Palantir and the clouds at strong (first-party apps), with the honest caveat that Databricks' value plane is analytics and data-app-centric.
Mixed, weighted to Databricks. The first-party apps (Genie, AI/BI, One, Apps) are Databricks IP (Ceded to Databricks). The value plane is platform-provided, not merely enabled, which separates Databricks from the platform-enabler moderates. The Marketplace ecosystem is Delegated (third-party providers over the open Delta Sharing protocol). Decision authority, read like any other layer (October 5, 2026): Genie's model writes the SQL behind each answer. Anyone with CAN EDIT on the Genie space can show the generated code, edit it, and run it, which reverses the model's call for that answer, and can save the fix as an instruction (Tune Genie quality docs). Model decides, visible, overridable, Delegated.
Watch-list (Preview/announced, not scored): Genie Deep Research (Preview); Genie One / Agents / Ontology (announced); Databricks One account-level (Beta); Databricks Assistant Agent mode (Preview). Genie and Apps reached GA June 2025; AI/BI Dashboards GA June 2024; Marketplace GA June 2023; Databricks One workspace-GA Mar 2026. To notes under the GA evidence table (October 24, 2026): Databricks One (Databricks One (now Genie One) entered Public Preview Sept 17, 2025; the current doc shows no badge at workspace level (account-level is Beta), but no GA affirmation appears in release notes through Oct 2026. ("Genie One , a simplified user interface designed for business users, is now in Public Preview.")). Out of the Databricks Marketplace chip under the GA evidence table (October 24, 2026): MCP servers ("List MCP servers in Databricks Marketplace (Public Preview)").
docs.databricks.com, Data + AI Summit 2025 & 2026 announcements, Databricks blog/newsroom (Lakeflow GA, Lakebase GA, Agent Bricks, Unity AI Gateway, Unity Catalog open-source), MLflow 3 release notes, analyst coverage, published 4+1 model Inference-interface reconciliation (July 23, 2026): the completions-interface ruling (OpenAI v1.0) and its Messages-API extension (Anthropic v1.0) applied: model access consumed through a genuine multi-vendor standard inference interface is Delegated; proprietary surfaces beyond the interface remain Ceded. Mosaic AI Model Serving re-chipped Ceded→Delegated (OpenAI-compatible endpoints; the AI Gateway keeps its Ceded chip at 2C). v1.2 (Unity AI Gateway GA re-read, August 8, 2026): Unity AI Gateway went GA August 4, 2026 (docs.databricks.com/aws/en/ai-gateway, GA blog post). Rate limits, spend caps, MCP-server governance with tool filtering, asset and traffic governance, usage tracking, and cost observability are GA; Smart Routing and service policies (with the PII, prompt-injection, and unsafe-content guardrails folded into them) remain Beta. 2C prose corrected for three stale claims (spend caps were listed as Beta, smart routing as mechanism-unspecified, guardrails under retired 'AI Guardrails Preview' vocabulary). Score unchanged: the 'Routing Is Not Reasoning' test holds because the placement engine did not ship. Retired status vocabulary removed from the 2C borrowed-judgment note. /reconcile (September 1, 2026): shared 2C findings stated explicitly (outcome-validation finding). /reconcile (September 4, 2026, override rule ratified on the Elastic row): Layer 1C authority Ceded to Delegated. Engine judgment on recomputation and planning inside enterprise-declared pipelines, with per-table and per-run overrides the engine must honor. Portability unchanged. No status changed. /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 3 calibration citation updated for the IBM Layer 3 re-read (partner to strong under the company-paper rule). No grade, component, or authority change. /reconcile (October 5, 2026): override rule applied to a cell scored before it; layer1b authority Ceded to Delegated on documented pins or per-request overrides. Grades and components unchanged. /reconcile (October 5, 2026): Layer 3 authority vendor / Ceded to model / Delegated: Genie's model writes the SQL and editors can edit and rerun it (the Layer 3 class convention retired). 1A holds vendor / Delegated, re-reasoned on Data Classification with per-detection Exclude. Grades and components unchanged. /reconcile (October 5, 2026): 2A authority vendor / Ceded to Delegated on documented cluster resize and per-job cluster pinning; statusLabel corrected from 'GPU Scheduling Pre-GA' (classic GPU clusters are long GA; the serverless GPU runtime is the Preview item). Grade unchanged. /reconcile (October 5, 2026): Layer 2C moderate to strong (Keith, October 5, 2026) on the complete-GA-plane route of the September 15 criterion: per-agent service-principal identity (app authorization), Unity AI Gateway (GA August 4, 2026), Unity Catalog registry, Supervisor Agent, MLflow tracing. Corrected: On-Behalf-Of auth is Public Preview, not GA (agent-authentication docs). The identity sub-test isn't adopted. Summary updated. Authority unchanged. v1.13 (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. three gap cells carrying the row's Layer 0 authority; summary edits (summary worksheet, Keith, October 16, 2026). v1.14 (October 19, 2026, /reconcile): 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. v1.15 (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. v1.16 (October 23, 2026, Layer 1 managed-services pass): Open Lakehouse Storage splits; Lakebase splits; Apache Spark + Structured Streaming splits. Grade unchanged. v1.17 (October 24, 2026, /reconcile): 1B calibration updated: VAST reads moderate at 1B, InsightEngine unscored without a product guide. Found in the 1B dependency audit follow-ups. Grade unchanged. v1.18 (October 24, 2026, /reconcile): 1B calibration restored: VAST reads strong at 1B again, InsightEngine GA on VAST's own announcement. Grade unchanged. v1.19 (October 24, 2026, /reconcile): GA evidence recorded on 1 of 2 Layer 1B chips under the GA evidence table. Grade unchanged. v1.20 (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. v1.21 (October 24, 2026, /reconcile): GA evidence recorded on 3 of 5 Layer 1A chips under the GA evidence table version 2. Grade unchanged. v1.22 (October 24, 2026, /reconcile): GA evidence recorded on 5 of 6 Layer 1C chips under the GA evidence table version 2. Grade unchanged. v1.23 (October 24, 2026, /reconcile): GA evidence recorded on 3 of 3 Layer 2A chips under the GA evidence table version 2. Grade unchanged. v1.24 (October 24, 2026, /reconcile): GA evidence recorded on 4 of 6 Layer 2B chips under the GA evidence table version 2. Grade unchanged. v1.25 (October 24, 2026, /reconcile): GA evidence recorded on 2 of 4 Layer 3 chips under the GA evidence table version 2. Grade unchanged. v1.26 (October 24, 2026, /reconcile): GA evidence recorded on 2 of 3 Layer 2C chips under the GA evidence table version 2. Grade unchanged. v1.27 (October 24, 2026, /reconcile): GA evidence table version 2.1 applied: 5 chips recorded or narrowed, 3 to notes. Grade unchanged. v1.28 (October 24, 2026, /reconcile): 2B prose: managed fine-tuning reached end of life; the strong case rests on GA serving and agents. Grade unchanged. v1.29 (October 27, 2026): each component now carries a durable offering id; nothing else changed.