- BFSIBanking, Financial Services & Insurance
- IT & ITESIT services, BPM, GCCs & SaaS
- Retail & e-commerceStore chains, D2C brands & marketplaces
- HealthcareComing soon
- ManufacturingComing soon
Databricks
The Databricks Data + AI Platform: a lakehouse on open formats with SQL warehousing, data engineering, Postgres, governance, BI and AI agents— on AWS, Azure and Google Cloud, with Indian regions on all three, and how each product is priced. This hub covers eight Databricks products.
Buy through TechBag
Same software. Better outcome — at a lower cost.
The company, at a glance
Quick answer
Eight intel pages. Warehousing, pipelines, Postgres, governance, BI and AI agents.
Databricks’ line, product by product — every card is a full intel page, from Lakehouse (Databricks SQL) and Lakeflow to Lakebase, Neon, Unity Catalog, Genie, Genie Code and Agent Bricks. Databricks Apps is covered on the Lakebase page; Lakewatch and CustomerLake are still in Private Preview.
Databricks Lakehouse (Databricks SQL)
Serverless SQL on Delta Lake and Iceberg tables.
Serverless, Pro and Classic SQL warehouses with Photon, AI/BI dashboards at no extra licence, and federation to other systems.
Databricks Lakeflow
Ingest, transform and orchestrate in one place.
Lakeflow Connect connectors, Spark Declarative Pipelines, Lakeflow Jobs and the no-code Lakeflow Designer, classic or serverless.
Databricks Lakebase
Postgres for apps, beside your lakehouse data.
Postgres 16 to 18 with autoscaling, scale to zero, branching and recovery, synced with Unity Catalog; GA since January 2026.
Neon (from Databricks)
Postgres for developers, no workspace needed.
Autoscaling, scale-to-zero Postgres with instant branches, bought on neon.com. No Indian region: the nearest is AWS Singapore.
Databricks Unity Catalog
One set of permissions for data and AI.
Permissions, lineage, auditing and classification for tables, files, models and agents, plus Unity Gateway for AI traffic.
Databricks Genie
Business users ask their data in plain words.
Genie One for chat and search, and no-code Genie Agents a data team curates, in the web app, Slack, Teams, Excel and Sheets.
Databricks Genie Code
An agent that writes and runs data code.
Plans, writes, runs and fixes code in notebooks, the SQL editor and pipelines under Unity Catalog permissions; CLI in Beta.
Databricks Agent Bricks
Build, evaluate and serve production agents.
Agent framework and evaluation, Model Serving, Foundation Model APIs, AI Search and managed MLflow, one name since March 2026.
Lakewatch
Platform & engineAn open agentic SIEM, announced on 24 March 2026 and still in Private Preview, with no price or documentation yet. Databricks completed its Panther acquisition in August 2026 and is combining the two. TechBag covers it on this hub only, and will review it when Databricks announces general availability and a price
CustomerLake
Platform & engineAn agentic customer data platform, announced on 16 June 2026 in Private Preview. Databricks’ product page gives no status beyond that and no price, so TechBag covers it on this hub only until it ships
OpenSharing, Marketplace and Clean Rooms
Platform & engineThe sharing layer: OpenSharing, the open protocol renamed in June 2026, shares live tables across platforms; Marketplace lists data, apps and accelerators; Clean Rooms bill through serverless jobs and storage. TechBag covers them in the platform section below
Databricks Apps
Platform & engineHosting for data and AI apps built with the frameworks you already use, billed as App Capacity at the serverless interactive rate: $0.75 a DBU on AWS Premium, with a Medium app at 0.5 DBU an hour. TechBag covers it on the Lakebase page, as the app half of the same story
Databricks Free Edition
Platform & engineA free edition for learning and experimenting, which replaced Community Edition (a migration tool arrived in November 2025). It is not a production tier; for that, a 14-day trial leads to pay-as-you-go. TechBag mentions it on this hub only
SAP Databricks
Platform & engineDatabricks inside SAP Business Data Cloud, sold by SAP rather than by Databricks, and quote-only. TechBag covers it on the SAP Business Data Cloud page, not here
Databricks sells 4 of the products TechBag carries in DevOps. The DevOps guide shows how the category splits and which part is yours. →
Why “one copy of data, every workload on it” is the whole story
Most data estates keep a warehouse, a lake, an operational database and an AI stack, each with its own copy. Databricks putsone copy of your data in open formats under one governance layer, and runs SQL, pipelines, Postgres, Genie and AI agents on it, all metered in DBUs. That makes which workload moves first, serverless or classic compute, and where AI prompts are processed the three things to settle first.
Open formats, one copy
Data sits once in your cloud storage as Delta Lake or Apache Iceberg tables, “without lock-in” in Databricks’ words, and serves analytics and operational work alike, so other engines can read it too.
Spark engine with Photon
The platform is built on Apache Spark, which Databricks’ founders created; Photon is its massively parallel engine. On classic job clusters Photon burns 2.5 to 2.9 times the DBUs, so test both.
Serverless or classic
Serverless SKUs include the cloud instances and run in Databricks’ compute plane in your region; classic SKUs run in your own cloud account and add its VM bill. Both are billed per second in DBUs.
Sharing and collaboration
OpenSharing shares live tables with other platforms as an open protocol, Marketplace lists data, apps and accelerators, and packaged Clean Rooms became generally available on 31 July 2026.
Start with the workload you can measure in ninety days — a warehouse migration, a pipeline, an app database or an AI agent — then pick the product that runs it.
Peer & market recognition
Every claim on this hub traces to one of these public signals.
Leader, 2026
highest in Ability to Execute and furthest in Completeness of Vision, its second year; report dated 22 June 2026
Leader, 2025
a Leader for the fifth consecutive year, and the first year it was judged on operational criteria too, via Lakebase
Visionary, 2026
placed as a Visionary in the 2026 Magic Quadrant for Analytics and Business Intelligence Platforms, not a Leader
Leader, 2026
a Leader in The Forrester Wave: Data Lakehouses, 2026, after a Leader placement in the Q2 2024 edition
Leader, 2025
a Leader in The Forrester Wave: Data Management for Analytics, 2025
Leader in four 2025 reports
GenAI evaluation, data platforms software, unified AI governance, and Asia/Pacific unified AI platforms
Data Catalogs: Leader, 2025
Leader and Outperformer in the 2025 Data Catalogs Radar, the report closest to Unity Catalog
Lakewatch, Neon, Genie Code
no Gartner or Forrester placement covers Lakewatch, Neon, Genie Code or Lakebase on its own
The company in six figures
See the platform, hear the pitch
Databricks Product Announcements in 5 Minutes | Data + AI Summit 2026
Databricks’ own five-minute run through what it announced at its June 2026 Data + AI Summit.
Defending against a tidal wave of AI attacks with Lakewatch, the agentic security Lakehouse
A June 2026 session on Lakewatch, the agentic SIEM that is still in Private Preview, with no price yet.
Data + AI Summit Keynote 2026 | Day 1
The full day-one keynote from June 2026, with Databricks’ founders setting out the platform and its new products.
Where Databricks sits — the grids
Two company-level views you won’t find on any vendor site — tap any dot for the rationale. The category-level grid lives on the product page.
Databricks Across Its Solution Areas
Each dot is a Databricks solution area: competitive position vs category momentum.
Lakehouse passed a $1.5B revenue run-rate in August 2026, growing more than 100% a year; Gartner named Databricks a Cloud DBMS Leader for a fifth year in 2025.
Workload Breadth × Cloud Choice
How many data and AI jobs one platform covers (warehouse, pipelines, Postgres, BI, agents) vs how freely you choose the cloud it runs on.
Warehouse, pipelines, Postgres, BI, coding and agents on AWS, Azure and Google Cloud, with Indian regions on all three and a published $/DBU per region.
Positions are TechBag’s illustrative synthesis of public review-platform standings and vendor documentation — not a reproduction of any analyst graphic. Verify before relying on it.
New to this? Learn it properly.
Zero-jargon starting points, in reading order. Each links into the deep education on the product page.
Where should you start with Databricks?
Answer three questions; we’ll point you at the right starting product. No email required — this isn’t that kind of quiz.
1. What problem is forcing the decision?
2. Who will own it?
3. What does success look like in 90 days?
Eight products, one copy of data
Every Databricks product reads the same governed tables in your cloud storage. Start with the workload you need first, then see what plugs in around it.
Read →Lakebase or Neon?
Both are serverless Postgres on the same technology. Lakebase lives in a Databricks workspace and runs in Mumbai; Neon needs no workspace but has no Indian region.
Read →The DBU rate is not the bill
Classic SKUs look cheaper per DBU because your cloud bills the VMs separately. Compare the total for the same workload, not the rate alone.
Read →Genie, Genie Code or Agent Bricks?
Genie answers business users, Genie Code writes data code for engineers, and Agent Bricks builds the agents you ship. Each bills on its own meter.
Read →Data in India, prompts elsewhere
An India Geo exists, but Genie One chat and agent modes process prompts outside India unless you switch cross-Geo processing off. Storage stays in-Geo.
Read →Where Databricks fits in DevOps
The DevOps category sets out the buying routes; Databricks sells into two of its guides, with Lakehouse, Lakebase and Neon in databases and Genie Code in developer tools.
Read →The acronym decoder
Every term on these pages, in one place- DBU
- Databricks Unit: a normalised unit of processing power, billed per second at a rate set per product.
- DSU
- Databricks Storage Unit: the unit for storage across products; default storage is 1 DSU per GB-month.
- Lakehouse
- One store of open-format tables that serves both warehouse-style SQL and data science, without copies.
- Delta Lake
- The open table format Databricks created, adding transactions and versioning to files in cloud storage.
- Apache Iceberg
- Another open table format; Databricks reads and writes it alongside Delta Lake.
- Photon
- Databricks’ massively parallel query engine, which speeds up SQL and Spark at a higher DBU rate on classic.
- Serverless SKU
- Compute Databricks runs for you; the price per DBU includes the cloud instances.
- Classic SKU
- Compute in your own cloud account; the DBU rate is lower, but your cloud bills the VMs on top.
- Capacity Unit
- Lakebase’s compute meter, billed per CU-hour, with autoscaling and scale to zero.
- Geo
- A group of cloud regions Databricks uses for data residency; there is an India Geo.
- Cross-Geo processing
- Lets some AI features send prompts outside your Geo; on by default outside the US and EU.
- Committed use
- A spend commitment that lowers list prices and can be used across clouds.
Buy it like you’ve done this before
The procurement playbook TechBag runs with IT buyers — steps, licensing cheat-sheet, and the pitfalls that cost quarters.
Pick the workload, not the platform
Eight products share one platform, but each bills on its own meter. Start with the workload you can measure in 90 days, and add others once it proves itself.
Choose cloud, region and tier
Pick AWS, Azure or Google Cloud and the Indian region first. Tiers are Premium or Enterprise, and Azure Premium matches Enterprise on AWS and Google Cloud.
Size DBUs from a real test run
Databricks itself advises representative test runs to estimate serverless DBUs. Run your own queries and pipelines for a week before you commit to a spend.
Add the cloud bill to classic
Classic and Pro SKUs exclude the VMs your cloud bills separately. Put both bills side by side before you compare classic with serverless compute.
Decide where AI prompts go
In the India Geo, cross-Geo processing is on by default for several AI features. Decide per workspace whether to allow it, and record the choice for DPDP.
Get it priced in rupees
Databricks lists prices in USD only. TechBag sizes the commitment, checks each promotion’s end date and gets the quote itemised in INR with GST.
The licensing cheat-sheet
| Product | Licensing model | How you enter | Best for |
|---|---|---|---|
| Lakehouse (Databricks SQL) | Per DBU, per second · SQL Serverless includes the instances; Pro ($0.55) and Classic ($0.22) add your cloud VM bill | $0.70/DBU US East · $0.77 AWS Mumbai and Azure Pune | SQL warehousing and BI |
| Lakeflow | Per DBU · Jobs, Pipelines (Core, Pro, Advanced) and Connect metered separately · 100 free connector DBUs per workspace a day | Jobs Classic $0.15 + VMs · serverless $0.35 US East, $0.37 Mumbai | Ingestion and pipelines |
| Lakebase | Capacity Unit-hours for compute, DSUs for storage · compute 50% off until 31 Jan 2027 | $0.092/CU-hour US East · $0.104 Mumbai · storage $0.375/GB-month in India | App and agent databases |
| Neon | Self-serve plans on neon.com, per CU-hour plus $0.35 per GB-month of storage · no monthly minimum | Free · Launch $0.106 · Scale $0.222 per CU-hour | Developer Postgres, no workspace |
| Unity Catalog | Included in Premium and Enterprise · classification, quality monitoring and other managed services at the serverless jobs rate | Included · add-ons $0.35/DBU US East, $0.37 Mumbai | Governance and lineage |
| Genie | Serverless Real-time Inference DBUs for LLM use · SQL warehouse compute billed separately · no seat fees | Named users’ LLM use free to 31 Jan 2027 · then $0.07/DBU, $0.074 Mumbai | Business-user questions |
| Genie Code | 150 DBUs per user a month, then inference DBUs · 25% off above the allowance to 31 Jan 2027 | $0.07/DBU US East · $0.074 Mumbai | Data engineering code |
| Agent Bricks | Each underlying product’s DBU rate · pay-per-token or provisioned throughput for models | Serverless inference $0.07/DBU US East · $0.074 Mumbai | Production AI agents |
| Platform add-ons | Enhanced Security and Compliance at 15% of product spend · Mission Critical at 30%, with managed disaster recovery (Public Preview) | Mission Critical 50% off to 30 Jun 2027 | Regulated workloads |
Every price is Databricks’ USD list, pay-as-you-go, as its pricing pages showed on 12 October 2026: SQL Serverless is $0.70 per DBU in US East and $0.77 in AWS Mumbai and Azure Central India, cloud instances included. Classic SKUs add the VM bill your cloud sends separately. The old entry tier is retired, so plans are Premium or Enterprise. Promotions end on 31 January 2027 (Genie LLM use, Lakebase compute, Genie Code) and 30 June 2027 (Mission Critical). Databricks publishes no INR prices; TechBag quotes in INR with GST.
Five pitfalls that cost buyers quarters
Comparing DBU rates without the VM bill
SQL Classic at $0.22 a DBU is not cheaper than Serverless at $0.70 until you add the VMs your cloud bills separately and the four-to-six-minute start-ups. Compare the total for the same workload.
Budgeting on promotional prices
Genie’s free LLM use, Lakebase’s 50% compute discount and Genie Code’s 25% discount all end on 31 January 2027, when Data Quality Monitoring also doubles to 2x. Budget year two at list price.
Reading an Indian region as Indian AI processing
Workspaces in Mumbai or Pune keep data in the India Geo, but Genie One chat, agent modes and pay-per-token models need cross-Geo processing, which is on by default. Decide it per workspace.
Assuming Neon runs in India
Neon has no Indian region; its nearest is AWS Singapore. If Postgres data must stay in India, Lakebase in AWS Mumbai or Azure Central India is the Databricks option.
Treating Lakebase storage as cheap
Lakebase storage costs $0.375 per GB-month in India, several times plain object storage, and history and snapshots are billed on top. Keep cold data in the lakehouse, not in Postgres.
The evaluation kit
Each intel page carries an 8-question vendor checklist and a value calculator:
Skip the homework entirely
Bring your requirements and current tool bills — a TechBag advisor models the whole decision for you.
Book a discovery call →The forces shaping this market
Four trends with momentum scores (TechBag’s read of analyst and market signals) — and what each means for your next decision.
Category momentum — estimated market growth
Illustrative CAGR synthesis**TechBag’s illustrative estimate, not a quoted analyst figure. The takeaway: buyers are moving AI agents and app databases onto the same governed data as their analytics — which is why Databricks added Lakebase and Agent Bricks beside its warehouse.
Agents on the data platform
Agent Bricks has been Databricks’ umbrella for all AI since March 2026, and its AI products passed a $1.4B revenue run-rate in February 2026.
What it means for you
Ask which agent actions need a human approval, and what each one costs.
Postgres joins the lakehouse
Lakebase went GA in January 2026 and passed a $100M revenue run-rate by August 2026; Snowflake made its own Postgres GA in February 2026.
What it means for you
Decide whether app data should live beside your analytics data or apart from it.
Open table formats
Databricks agreed to buy Tabular, founded by Apache Iceberg’s creators, in June 2024, and now stores data as Delta Lake or Iceberg tables.
What it means for you
Keep tables in an open format so that leaving stays possible.
Plain-language analytics
Databricks says customers built over 1.5 million Genie Agents, then called spaces, in 2026, and moved Genie to pay-as-you-go in July 2026.
What it means for you
Govern the data first: plain-language answers are only as good as the definitions under them.
Questions buyers ask about Databricks
Ready to shortlist Databricks?
Open any of the eight intel pages for the deep dive, or let a TechBag advisor build the case with you — the workload to move first, the cloud and Indian region, a DBU estimate from a real test run, and the quote that fits, quotes, trials, GST invoicing and lifecycle support included.
Stats, positions and figures are illustrative syntheses of public materials; verify before purchase.