Databricks for Startups: Up to $200K in Databricks and Neon Credits

Databricks for Startups offers up to $200,000 in credits across Databricks and Neon through a single application. Databricks relaunched the program on June 16, 2026. It's aimed at early-stage, venture-backed startups building on data and AI, especially teams that recently raised pre-seed to Series A funding. Besides credits, you get hands-on technical guidance, founder community access and go-to-market support. Databricks doesn't publish credit tiers, expiry terms or a review timeline.

This guide covers what's included, how Neon fits in, who qualifies, how to apply, what the credits are good for, and how the program compares with other startup offers. Everything was checked against Databricks' official pages on October 17, 2026.

What you get

The headline figure is a maximum, not a standard grant. Databricks says the amount depends on qualification and doesn't say how the total is split between Databricks and Neon. Treat $200K as the ceiling for the strongest applicants, not what every approved startup receives.

Why Neon is part of the deal

Neon is a serverless Postgres company. Databricks announced it would acquire Neon on May 14, 2025, saying more than 80 percent of the databases provisioned on Neon were being created by AI agents rather than humans. A month later, on June 11, 2025, Databricks launched Lakebase, an operational database for AI apps and agents built on Neon technology.

The relaunched startup program pulls both together. Databricks' launch post argues that a startup first needs an app backend, a database and access to foundation models, which Databricks and Neon cover on day one. Later it needs analytics, data warehousing, AI systems and enterprise governance, which the Databricks platform covers. In practice, the Neon credits can pay for your app's Postgres database, while the Databricks credits pay for data engineering, analytics, model serving and AI workloads.

What Neon costs without credits

Neon has a permanent Free plan with no card required: up to 100 projects, 100 compute-unit hours per project, autoscaling up to 2 CU (8 GB RAM), and 1 GB of Postgres storage per project (20 GB across all projects). Paid usage on the Launch plan is $0.106 per CU-hour for compute and $0.35 per GB-month for storage, with no monthly minimum. The Scale plan is $0.222 per CU-hour and adds features such as private networking, HIPAA and SOC 2 options and an uptime SLA.

Worked example using Neon's published Launch rates: a database running at 1 CU for a full 30-day month uses 720 CU-hours. At $0.106 per CU-hour that's $76.32 for compute, plus $3.50 for 10 GB of storage at $0.35 per GB-month, about $79.82 a month in total. Scale-to-zero reduces the compute part if your database is idle for long periods.

Who qualifies

Databricks describes the program as purpose-built for early-stage, venture-backed startups, especially founders who have recently raised institutional funding from pre-seed through Series A and are building on data and AI. Its startup page doesn't list detailed criteria such as company age, headcount or country. Databricks explicitly invites startups that aren't sure whether they qualify to apply anyway, and says it will review each application.

• Venture-backed, pre-seed to Series A: the core audience. Apply directly or ask your investor to nominate you.
• Backed by a partner VC: you have an extra route in. The FAQ names Andreessen Horowitz, Battery, General Catalyst, Insight Partners, Menlo Ventures and NEA among the partner firms.
• Bootstrapped or angel-funded: not the stated target, but Databricks doesn't rule you out. Apply and explain your traction.
• Later-stage companies: Series B and beyond fall outside the stated pre-seed to Series A focus. Larger software companies building products on Databricks may fit the Built on Databricks Partner Program instead.

How to apply

1. Go to the Databricks for Startups page and open the application link. The form is hosted on a Google Sites page run by Databricks.
2. Describe your startup, what you're building on data and AI, and your funding history.
3. If your investor is a Databricks partner, ask them to nominate you as well, or instead of applying yourself.
4. Wait for Databricks to review and respond. No timeline is published.
5. Once approved, confirm the credit amount, how it's split between Databricks and Neon, and the expiry date before you plan workloads around it.

Related Databricks programs

Databricks Free Edition isn't a startup shortcut. Databricks says it's for personal use only, not commercial or production use, with smaller compute and warehouse sizes and fair use limits. Use it to learn the platform, then apply to the startup program for a commercial workspace.

How to get the most out of the credits

• Start with the Solution Accelerators. They're working notebooks for common use cases, so you can test an architecture before writing everything from scratch.
• Put your app database on Neon. Since the credits cover Neon too, you can run your Postgres backend there and avoid paying a separate database bill while you're on credits.
• Turn off idle compute. Databricks clusters and SQL warehouses bill while running. Auto-termination and serverless options stop credits draining overnight.
• Ask for the technical guidance early. The Databricks team can help with architecture decisions that are expensive to change later.
• Track the expiry date. Databricks doesn't publish one, so get it confirmed when you're approved and plan heavy workloads before it.

How Databricks billing works once credits run out

Databricks charges for compute in Databricks Units (DBUs), a normalized measure of processing power. How many DBUs a workload uses depends on the compute resources and the amount of data processed, and rates differ by product, region and cloud provider. Databricks publishes per-SKU rates on its price list rather than on the main pricing page. Storage, networking and related costs vary by service and cloud.

• Pay as you go: no upfront costs, billed per second, and billed monthly to a card on file at price list rates. Azure Databricks is billed by Microsoft instead.
• Commitments: committing to a usage level unlocks discounts. Larger commitments bring more benefits, including using a commitment across several clouds. Terms are negotiated with Databricks sales.
• Trials: Databricks' standard free trial may include credits. Databricks decides eligibility and amount, and when trial credits run out you're automatically moved onto the plan you were using.

The same pattern is likely with startup credits, though Databricks doesn't state it: once credits are used up or expire, expect usage to be billed at standard rates. Set budgets and alerts before that happens. On the Neon side, a project can drop back to Neon's Free plan limits if you don't need a paid plan.

Lakebase vs. Neon

Databricks now offers Postgres in two forms. Neon is the standalone serverless Postgres service with its own pricing and console. Lakebase, launched in Public Preview on June 11, 2025, is a managed Postgres database inside the Databricks platform, built on Neon technology. It separates compute from storage, supports copy-on-write branching, syncs with lakehouse tables and works with Unity Catalog. The startup program names Neon as the credit partner. If you plan to use Lakebase instead, ask Databricks how your credits apply.

How it compares with other startup programs

Databricks workspaces are deployed on AWS, Azure or Google Cloud, so it can make sense to pair Databricks credits with a credit program from the cloud your workspace uses. For a broader plan, see our guide on how to get free AI credits. If your startup also needs GPUs for training, look at NVIDIA Inception.

History and recent changes

• May 14, 2025: Databricks announces its agreement to acquire Neon.
• June 11, 2025: Databricks launches Lakebase, an operational database built on Neon technology.
• June 16, 2026: Databricks announces a major update to its startup program, consolidating it into one application with up to $200K in credits across Databricks and Neon.