Resources // Cloud Providers
DigitalOcean Hatch Program
The offer
Compute credits: up to $100,000 over 12 months, but this is a referral-gated ceiling. The amount varies, is generally capped around $10,000/month, does not roll over, and is forfeited if unused. Credits exclude GPU Droplets and inference (and Paperspace / Cloudways).
GPU: discounted NVIDIA H100 GPU Droplets at $1.90 per GPU/hour for up to 12 months; up to 3 months free GPU usage for select startups (apply separately).
Support: 15 months of free Standard-tier support.
Ecosystem: DigitalOcean Marketplace listing, co-marketing, and a welcome kit of rotating partner perks (e.g. Stripe, Notion, HubSpot).
Who qualifies
- Raised $10M or less (the hard funding criterion; "pre-seed to Series A" is just a rough gloss)
- AI-native startups prioritized; service/agency businesses are NOT eligible
- New to DigitalOcean (no prior credits); new team account with a valid credit card
- Company website plus matching corporate-domain email (no Gmail/Outlook)
- VC/accelerator optional, but a partner referral unlocks higher credit tiers
Community Insights
DigitalOcean (the program is now branded “DigitalOcean Startups”; “Hatch” survives mostly in URLs) is liked by early teams for simplicity and predictable billing versus AWS: easy to stand up a server, understandable costs, good for side projects and MVPs. Two things to know for 2026: the “$100k” is a referral-gated ceiling (direct applicants get materially less, paid monthly, no rollover, GPU excluded), and GPU-droplet capacity has actually sold out across all regions at times, so do not assume the discounted H100s are always available.
Best Practices (from community tips)
- Apply with a corporate-domain email plus live website and describe how you will use cloud/GPU; a partner referral unlocks higher tiers.
- Plan credits across 12 months. They are issued monthly (~$10k/mo cap), do not roll over, and overage is billed.
- Use DO for simplicity (one-server setups, predictable cost); reach for AWS only when you need heavier managed features.
- Do not bank on GPU droplets being available. Capacity has sold out region-wide; have a fallback for AI/training workloads.