The direct answer: DP-750 is for data engineers who build and manage pipelines on Azure Databricks. DP-800 is for SQL developers who want to embed AI capabilities directly into SQL Server and Azure SQL databases. They target completely different roles, different platforms, and different skill sets — and most candidates should only pursue one of them.
Both launched in beta in March 2026 and go generally available in May 2026. Neither replaces the other. Neither replaces DP-100. They are two entirely new certifications filling gaps that did not previously exist in Microsoft’s data and AI certification portfolio.
What Is the Difference Between DP-750 and DP-800?
| Factor | DP-750 | DP-800 |
| Official name | Implementing Data Engineering Solutions Using Azure Databricks | Developing AI-Enabled Database Solutions |
| Certification earned | Microsoft Certified: Azure Databricks Data Engineer Associate | Microsoft Certified: SQL AI Developer Associate |
| Primary platform | Azure Databricks | SQL Server, Azure SQL Database, Microsoft Fabric SQL |
| Primary language | Python and SQL (Spark context) | T-SQL, with Python/R for AI integration |
| Primary audience | Data engineers building big data pipelines | SQL developers adding AI to database applications |
| Core technology | Apache Spark, Delta Lake, Unity Catalog, Databricks Workflows | Azure SQL, SQL Server, vector search, RAG pipelines, Azure OpenAI integration |
| AI coverage | Indirect — AI-ready data foundations | Direct — embedding AI models, vector search, RAG in SQL |
| Databricks | Core — every domain | Not covered |
| T-SQL depth | Basic SQL for data engineering | Deep — advanced T-SQL, query optimization, programmable objects |
| Vector search | Not covered | Core topic — VECTOR_SEARCH, embeddings in SQL |
| Azure OpenAI integration | Not covered | Core topic — calling Azure OpenAI from database solutions |
| Beta available | March 2026 | March 2026 |
| General availability | May 2026 | May 2026 |
| Exam cost | $165 USD | $165 USD |
| Passing score | 700/1000 | 700/1000 |
| Duration | 100 minutes | 100 minutes |
| Is it a replacement for DP-100? | No — different role entirely | No — different role entirely |
| Is it a replacement for anything? | Fills a new gap — no retiring exam | Fills a new gap — no retiring exam |
What Is DP-750?
DP-750 is Microsoft’s first dedicated certification for Azure Databricks data engineers. It validates your ability to implement production-grade data engineering solutions using Azure Databricks — building and managing the data pipelines that power analytics and AI at enterprise scale.
This certification fills a gap that data engineers have been pointing at for years. Microsoft and Databricks have had a deep partnership, but there was no Microsoft-issued credential that validated Databricks expertise. DP-750 changes that.
DP-750 Exam Domains
| Domain | Weight | What You Do |
| Set up and configure an Azure Databricks environment | 15–20% | Workspace configuration, compute setup (job compute, serverless, warehouse, classic), cluster performance, Photon acceleration, Git integration, connectivity to Azure services |
| Secure and govern Unity Catalog objects | 15–20% | Unity Catalog structure, privilege management for users/service principals/groups, row-level and column-level security, ABAC with tags, Key Vault integration, data lineage tracking |
| Prepare and process data | 30–35% | Data ingestion via Lakeflow Connect, ADF, and notebooks; batch and streaming loading; Delta table formats and operations; data transformation with SQL and Python; data quality management; schema evolution |
| Deploy and maintain data pipelines and workloads | 30–35% | Databricks Workflows and job scheduling, Lakeflow Jobs, pipeline orchestration, performance tuning, troubleshooting Spark jobs, caching, skewing, and spilling, monitoring with Azure Monitor |
The two heaviest domains — preparing and processing data, and deploying and maintaining pipelines — together represent 60 to 70 percent of the exam. This is a deeply operational exam. It tests whether you can build and run data engineering systems in production, not just demonstrate awareness of Databricks features.
Core tools tested: Azure Databricks, Apache Spark, Delta Lake, Unity Catalog, Databricks Workflows, Lakeflow, Azure Data Factory, Azure Monitor, Azure Key Vault, Microsoft Entra ID, Git.
Who DP-750 is for: Data engineers, big data engineers, Databricks engineers, and analytics engineers who design, deploy, and maintain data pipelines using Azure Databricks as their primary platform.
What Is DP-800?
DP-800 is Microsoft’s first certification specifically for SQL developers integrating AI capabilities into database solutions. It validates your ability to design and build AI-enabled database applications using Microsoft’s SQL platforms — embedding AI models, vector search, RAG pipelines, and Azure OpenAI directly into SQL-based solutions.
This certification reflects a fundamental shift in what SQL development means in 2026. Azure SQL and SQL Server now ship with built-in vector search, native AI model calling from T-SQL, and direct integration with Azure OpenAI. SQL developers can build generative AI applications without moving data to a separate AI platform. DP-800 validates this exact skill set.
DP-800 Exam Domains
| Domain | Weight | What You Do |
| Design and develop database solutions | 35–40% | Design database architecture for AI workloads; implement database objects including tables, temporal tables, graph tables, JSON columns, columnstore indexes; create programmable objects including views, stored procedures, functions, triggers; implement T-SQL advanced features; use SQL Database Projects with CI/CD |
| Implement AI capabilities in database solutions | 25–30% | Implement vector search with VECTOR_SEARCH, VECTOR_DISTANCE, VECTOR_NORMALIZE; choose embedding maintenance methods; integrate Azure OpenAI from T-SQL; implement RAG pipelines using SQL as data layer; evaluate and deploy external AI models; connect to MCP server endpoints |
| Secure, optimize, and deploy database solutions | 35–40% | Implement Always Encrypted, Dynamic Data Masking, row-level security; query performance tuning with query plans, DMVs, Query Store; manage transactions and concurrency; configure testing strategy with unit and integration tests; manage source control and branching for database projects |
The AI capabilities domain is the newest and most distinctive content area. Everything else in DP-800 builds on SQL development skills that experienced SQL professionals already have. The AI integration domain — vector search, embedding management, RAG in SQL, Azure OpenAI calling from T-SQL — is what makes this certification genuinely new.
Core tools tested: SQL Server, Azure SQL Database, Microsoft Fabric SQL analytics endpoint, Azure OpenAI, vector functions (VECTOR_SEARCH, VECTOR_DISTANCE), SQL Database Projects, GitHub Copilot for SQL, MCP server connections, Azure Data Studio, Query Store.
Who DP-800 is for: SQL developers, database developers, database architects, and backend developers who work primarily with Microsoft SQL platforms and want to add AI integration skills to their SQL expertise without migrating to new platforms.
The Core Philosophical Difference
DP-750 asks: Can you build and run the data pipelines that feed AI systems?
DP-800 asks: Can you embed AI capabilities directly into SQL-based applications?
These represent two different approaches to the convergence of data and AI.
DP-750 professionals build the data foundations — the ingestion pipelines, data lakes, Delta tables, and streaming architectures — that AI models and applications consume. They work upstream of AI, making sure the data is clean, available, and governed correctly.
DP-800 professionals bring AI into the data layer itself — embedding vector search, RAG workflows, and AI model calls directly into database applications. Instead of sending data to an AI platform, they bring AI to where the data already lives. They work inside SQL, extending what SQL can do rather than replacing it.
Both roles are in high demand. They rarely overlap. Most data professionals sit clearly in one camp or the other based on their platform experience and daily work.
How DP-750 and DP-800 Relate to DP-100
Many candidates wondering about DP-750 and DP-800 are coming from the DP-100 context — either they hold it and it is retiring, or they were studying for it when the retirement was announced.
| Certification | Role | Platform | Relationship to DP-100 |
| DP-100 (retiring June 1) | Data Scientist | Azure Machine Learning | Build and train ML models |
| AI-300 (replacing DP-100) | MLOps Engineer | Azure ML + Microsoft Foundry | Operationalize ML and GenAI |
| DP-750 (new) | Data Engineer | Azure Databricks | Build data pipelines that feed ML models |
| DP-800 (new) | SQL AI Developer | SQL Server / Azure SQL | Embed AI into SQL database applications |
DP-750 and DP-800 are not replacements for DP-100. They are different roles on different platforms. If your DP-100 is retiring and you want the direct replacement credential, that is AI-300. DP-750 and DP-800 are separate career paths that exist alongside the DP-100 to AI-300 transition.
For the full context of the DP-100 retirement and AI-300 replacement, our DP-100 vs AI-300 guide covers that decision in complete detail.
Who Should Take DP-750?
Take DP-750 if:
You use Azure Databricks as your primary data engineering platform. If your day involves writing Spark jobs, building Delta pipelines, configuring Unity Catalog governance, or managing Databricks Workflows in production, DP-750 directly validates what you do every day.
You are a data engineer at an organization that runs analytics or AI on Databricks. The certification signals specifically to employers running Databricks-heavy data stacks that you have validated, production-grade expertise on their primary platform.
You work at the intersection of data engineering and data science. Data engineers who build the pipelines that data scientists consume are in high demand at every organization running ML workloads. DP-750 validates the engineering side of that partnership.
You are coming from Azure Data Factory, Azure Synapse, or Microsoft Fabric and want to formalize your Databricks expertise. If you have worked with Azure data services broadly and Databricks is increasingly central to your stack, DP-750 gives you the Microsoft-issued credential that validates that specific platform depth.
Do NOT take DP-750 if your primary work is SQL database development, if you have never worked seriously with Databricks, or if your data engineering work is primarily Fabric-based rather than Databricks-based.
Who Should Take DP-800?
Take DP-800 if:
You are a SQL developer who wants to add AI capabilities to your work without changing platforms. DP-800 is specifically designed for professionals who already know SQL deeply and want to extend that expertise into AI — building vector search, RAG pipelines, and Azure OpenAI integrations from T-SQL rather than from Python in a separate environment.
Your organization runs SQL Server or Azure SQL Database and is moving toward AI-powered applications. As enterprises deploy generative AI, many teams are discovering that their data already lives in SQL databases and they do not need to move it to build AI features. DP-800 validates the skills to build AI directly on that existing foundation.
You are a database developer or architect designing AI-ready database schemas. Graph tables, JSON columns, columnstore indexes, temporal tables, and vector-enabled schemas are all DP-800 content. If you are the person who designs database architectures that can support both traditional and AI workloads, this certification validates that architectural thinking.
You want to differentiate yourself in the SQL developer market. SQL development has been considered a mature, stable skill area. DP-800 represents a genuine inflection point — AI-enabled SQL is a new capability area where certification early movers will have a market advantage for the next several years.
Do NOT take DP-800 if your primary platform is Databricks or Python-based ML, if your data engineering work is primarily pipeline-focused, or if you have minimal SQL Server or Azure SQL experience.
DP-750 vs DP-800: Can You Take Both?
Yes. They do not conflict — they are complementary credentials for different but adjacent roles.
Some data platform professionals genuinely work across both Databricks-based pipeline engineering and SQL-based AI development. For these candidates, holding both DP-750 and DP-800 communicates full-stack data and AI competency across Microsoft’s two primary data platforms.
Most professionals, however, sit clearly in one specialization. Taking both makes sense only if your actual daily work spans both platforms at meaningful depth. Taking DP-800 just to add a second credential without real SQL AI experience does not serve your career as well as going deeper on your primary platform.
How DP-750 and DP-800 Fit the Microsoft Data Certification Landscape
| Certification | Platform | Level | Role |
| DP-900 Azure Data Fundamentals | Multiple | Fundamentals | Beginner overview |
| DP-100 (retiring June 1, 2026) | Azure ML | Associate | Data Scientist |
| AI-300 (replacing DP-100) | Azure ML + Foundry | Associate | MLOps Engineer |
| DP-600 Fabric Analytics Engineer | Microsoft Fabric | Associate | Analytics/BI Engineer |
| DP-700 Fabric Data Engineer | Microsoft Fabric | Associate | Fabric Data Engineer |
| DP-750 (new May 2026) | Azure Databricks | Associate | Databricks Data Engineer |
| DP-800 (new May 2026) | SQL / Azure SQL | Associate | SQL AI Developer |
Microsoft has now filled two significant gaps in its data certification landscape. Previously, data professionals working heavily with Databricks had no Microsoft-issued credential for that platform. And SQL developers building AI features in SQL had no certification path for that emerging skill set. DP-750 and DP-800 fill both gaps simultaneously.
Salary and Career Context
| Role | Average US Salary |
| Azure Databricks Data Engineer (DP-750 level) | $115,000 to $155,000 |
| Senior Databricks Engineer / Architect | $145,000 to $185,000 |
| SQL AI Developer (DP-800 level) | $100,000 to $140,000 |
| Senior Database Developer with AI specialization | $120,000 to $160,000 |
Databricks data engineering is one of the highest-compensated data specializations in the market. The platform’s dominance in large-scale data processing and its deep integration with ML and AI workloads means practitioners who can operate it at production scale are in sustained high demand.
SQL AI development is an emerging premium specialization. As enterprises deploy AI on their existing SQL infrastructure rather than migrating to new platforms, the SQL developers who can build these solutions will command increasing premiums over traditional SQL developers.
How to Prepare for DP-750
Step 1: Get hands-on with Azure Databricks using a free trial or your organization’s environment. DP-750 is an operational exam. Reading about Databricks is not sufficient. Set up a Databricks workspace, configure Unity Catalog, build a Delta pipeline end-to-end, and troubleshoot a real Spark job. This hands-on experience is what the exam’s scenario-based questions are testing.
Step 2: Focus preparation on the two heavyweight domains. Prepare and process data (30–35%) and Deploy and maintain data pipelines (30–35%) together account for approximately 65 percent of the exam. Build real pipelines — ingest from multiple sources, transform with both SQL and Python, implement Delta table operations including OPTIMIZE and VACUUM, and configure Databricks Workflows to orchestrate multiple jobs.
Step 3: Master Unity Catalog governance deeply. Unity Catalog is Microsoft and Databricks’s shared direction for data governance. The securing and governing domain (15–20%) is conceptually distinct from the rest of the exam — it is about governance architecture, privilege models, and security patterns rather than data processing. Study ABAC, column-level security, row-level security, and data lineage separately from your pipeline work.
Step 4: Know your Azure integrations. DP-750 expects you to understand how Databricks connects to the broader Azure ecosystem — Azure Data Factory for orchestration, Azure Monitor for observability, Azure Key Vault for secrets management, and Microsoft Entra ID for identity. These integration points appear throughout the exam.
Step 5: Use current practice materials. Our Microsoft exam preparation section covers current active Microsoft certifications. Check for DP-750 practice materials as they become available following the May 2026 general availability launch.
How to Prepare for DP-800
Step 1: Build genuine SQL depth first. DP-800 expects candidates to already be comfortable with T-SQL — advanced queries, stored procedures, indexes, transactions, and performance tuning. If SQL is not your daily language, build that foundation before attempting DP-800. The AI integration content sits on top of strong SQL fundamentals, not alongside weak ones.
Step 2: Get hands-on with vector search in Azure SQL. This is the genuinely new content area that most SQL developers will not have prior experience with. Use Azure SQL Database (which now supports vector functions natively) to implement a real vector search scenario — store embeddings, implement VECTOR_SEARCH queries, and build a basic RAG pipeline using SQL as the data layer.
Step 3: Integrate Azure OpenAI from T-SQL. Practice calling Azure OpenAI models from stored procedures and T-SQL queries. Understand how to pass data to models, parse responses, and store results in SQL tables. This workflow is central to the AI capabilities domain.
Step 4: Practice with SQL Database Projects and CI/CD. The exam tests modern database development practices — SQL Database Projects, source control integration, branching, and deployment pipelines. Set up a SQL Database Project in Azure Data Studio or Visual Studio Code, connect it to Git, and run it through a basic CI/CD workflow.
Step 5: Study GitHub Copilot for SQL development. The DP-800 study guide explicitly includes configuring GitHub Copilot for SQL contexts and connecting to MCP server endpoints. This reflects how SQL development is actually done in 2026 — with AI assistance built into the development environment.
Decision Framework: DP-750 vs DP-800
| Your Situation | Take This Exam |
| You build data pipelines on Azure Databricks daily | DP-750 |
| You write T-SQL and work with SQL Server or Azure SQL daily | DP-800 |
| Your work is primarily Spark-based processing | DP-750 |
| You want to add vector search to existing SQL applications | DP-800 |
| You work with Unity Catalog and Delta Lake | DP-750 |
| You want to build RAG pipelines without leaving SQL | DP-800 |
| You are a data engineer at a Databricks-first organization | DP-750 |
| You are a database developer at an enterprise SQL shop | DP-800 |
| Your DP-100 is retiring and you want its direct replacement | Neither — take AI-300 |
| You work across both Databricks and SQL platforms deeply | Consider both |
| You are new to both platforms | Build platform experience first, then certify |
Frequently Asked Questions: DP-750 vs DP-800
What is the difference between DP-750 and DP-800?
DP-750 validates Azure Databricks data engineering skills — building and managing big data pipelines with Spark, Delta Lake, and Unity Catalog. DP-800 validates SQL AI developer skills — integrating AI capabilities like vector search, RAG, and Azure OpenAI directly into SQL Server and Azure SQL database applications. They are different certifications for different roles and platforms.
When are DP-750 and DP-800 available?
Both went into beta in March 2026 and are expected to reach general availability in May 2026. Both are currently available in beta through Microsoft’s certification exam portal.
Is DP-750 a replacement for DP-100?
No. DP-750 is for Databricks data engineers. DP-100 (Azure Data Scientist Associate) is being replaced by AI-300 (MLOps Engineer Associate). These are separate roles and separate platforms.
Is DP-800 a replacement for any retiring exam?
No. DP-800 fills a new gap — there was no previous Microsoft certification for SQL developers integrating AI into database solutions. It does not replace any retiring exam.
Can I take both DP-750 and DP-800?
Yes. They do not conflict. Candidates who genuinely work across both Azure Databricks pipelines and SQL-based AI development can pursue both credentials. Most professionals, however, have a clear primary platform and should focus on the relevant certification.
How hard is DP-750 compared to other Microsoft data certifications?
DP-750 is a production-level operational exam. It does not test feature awareness — it tests whether you can build and troubleshoot real Databricks data engineering systems. Candidates with strong hands-on Databricks experience report it aligns well with their daily work. Candidates who have studied Databricks primarily through documentation without real implementation experience find it significantly harder.
How hard is DP-800 for experienced SQL developers?
The SQL foundations — database design, T-SQL, query optimization, security — are familiar ground for experienced SQL professionals. The AI integration domain is genuinely new for most SQL developers. Candidates with strong SQL backgrounds typically need 4 to 6 additional weeks of focused preparation on the AI-specific content areas.
Does DP-750 cover machine learning?
DP-750 covers configuring Azure Databricks for machine learning workloads (compute settings, runtime versions) but is not an ML engineering exam. It focuses on data engineering foundations — pipelines, governance, and platform operations. AI-300 is the MLOps engineering certification.
What is Unity Catalog and why does it matter for DP-750?
Unity Catalog is the unified governance solution for the Databricks Lakehouse Platform. It provides centralized data access controls, data lineage, and data discovery across all Databricks workspaces in an organization. DP-750 tests Unity Catalog governance deeply because it is how enterprise Databricks environments manage data security and compliance.
What is vector search and why does it matter for DP-800?
Vector search is the technology that powers semantic similarity search — finding database records that are conceptually similar to a query rather than exactly matching keywords. It is the foundation of RAG (retrieval-augmented generation) systems that combine database knowledge with AI language models. Azure SQL and SQL Server now support vector search natively through functions like VECTOR_SEARCH and VECTOR_DISTANCE, and DP-800 tests whether you can implement and optimize these capabilities.
