If you claimed DigitalOcean credits through the GitHub Student Developer Pack, check your inbox. On June 12, 2026, DigitalOcean began emailing pack users that it is winding down its participation in the program and that all credits provided through the pack will expire on July 31, 2026. Unused balances won't carry over or transfer, and accounts continue on standard billing afterward — meaning any Droplets, databases, or volumes you leave running will start charging the payment method on file.
For a lot of students, this is the server their portfolio, side project, or final-year project lives on. This post covers what the notification says, what is and isn't publicly confirmed yet, a concrete pre-deadline checklist — including the backup detail that surprises most people — and where to host for free after July 31.
What the Notification Says
The email, sent from DigitalOcean's official notification address, makes three things explicit:
- Credits stay usable until July 31, 2026 on eligible DigitalOcean services.
- Unused credit expires after that date — no carryover, no transfer.
- Standard billing takes over. If you keep paid resources running past July 31, they bill to your payment method like any regular account.
One caution before you act on any expiring-credit email: log in to your account directly at the DigitalOcean dashboard rather than through links in the message, and confirm your balance and terms there. Credit-expiration notices are a favorite phishing template, and verifying in the dashboard costs you nothing.
What's Publicly Confirmed — and What Isn't Yet
Here's the part worth being precise about. As of June 12, neither GitHub's Student Developer Pack page nor the official pack changelog records a DigitalOcean exit. The pack page still advertises the familiar offer — $200 in platform credit for one year, for new accounts verified through GitHub Education.
That lag isn't surprising. The changelog has trailed reality before: when 40 partner offers (including Linode, CodeSandbox, and SonarSource) were deprecated in July 2024, the entry itself notes the removals were logged late due to an internal delay. Partner departures are routine — Bump.sh ended its student offer in December 2025 — and announcement emails routinely land before public pages update.
What is verifiable on the pack page is a change made four days before the emails went out. Effective June 8, 2026, DigitalOcean narrowed what student credits can buy: they now explicitly exclude GPU Droplets, NVIDIA H100 GPU Kubernetes, bare-metal GPUs, dedicated and serverless inference, third-party AI models (including Anthropic and OpenAI models hosted on the platform), Paperspace Machines, and Cloudways services. Read together, the sequence is hard to misread — first fence the credits off from the expensive AI infrastructure, then wind the program down. DigitalOcean has been a pack partner since the program launched in 2014, with the credit growing from $50 to $100 to $200 over the years, so this is the end of one of the pack's longest-running offers.
Your Checklist Before July 31
Six weeks is plenty of time if you start now, and barely enough if you start the last weekend of July. Work through these in order.
1. Audit what's actually running. Log in and open the billing section of the control panel, check your remaining credit balance, and list every resource that costs money: Droplets, managed databases, volumes, load balancers, reserved IPs on stopped Droplets, snapshots, and container registries. DigitalOcean's own advice in the email is the right starting point — know what you have before deciding what to do with it. Set a billing alert while you're there, so a forgotten resource can't quietly bill your card in August.
2. Decide what survives. For each resource: keep it and pay, migrate it, or shut it down. Be honest about which projects are alive. A $6/month Droplet is cheap; three forgotten ones plus a managed database is a real student-budget line item.
3. Back up — and know the snapshot trap. This is the step with the surprise in it. DigitalOcean snapshots cannot be downloaded off-platform — a snapshot is only restorable inside DigitalOcean. It's a fine safety net if you're staying, but it is not an exit strategy. To actually get your data out, DigitalOcean's own guidance points to manual tooling:
- Files and configs:
rsyncor SFTP the directories you need to your laptop or new host —rsync -avz root@your-droplet:/var/www/ ./backup/covers a typical web root. - Databases: take proper dumps (
pg_dump,mysqldump,mongodump) rather than copying data directories. - Infrastructure as knowledge: export DNS records, note firewall rules, and capture anything you configured by hand in the panel.
Also know that snapshots themselves cost $0.06 per GB per month — once your credits stop covering the bill, even your "archived" snapshots become a recurring charge.
4. Destroy, don't just power off. Stopped Droplets still incur charges, because the disk and IP stay reserved. Once a project is backed up, destroy the resource. Back up first and verify the backup restores — deleting a project removes its data permanently.
5. Decide before the deadline, not after it. After July 31 there's no credit cushion: every day a resource runs is a billed day. If you want zero risk of charges, remove paid resources and confirm your billing page shows nothing active before the deadline.
Where Students Can Host for Free After July 31
The Student Developer Pack still includes real cloud offers. Two are worth naming with their current verified terms:
- Microsoft Azure for Students — $100 in Azure credit plus free access to more than 25 Azure services, no credit card required for students 18 and over. The closest like-for-like replacement if you want general-purpose VMs and managed services.
- Heroku for GitHub Students — $13 per month in platform credits for 24 months ($312 total), usable on Eco dynos, Postgres, and Key-Value Store. Less raw infrastructure than a Droplet, but ideal for the "deploy my app and stop thinking about servers" use case.
Beyond the pack, the usual suspects — Oracle Cloud's free tier, Cloudflare Pages and Workers, Vercel, Netlify, Fly.io, and Railway — all have free or hobby tiers worth evaluating, but check the current terms yourself: free-tier conditions change often, and this announcement is itself proof of that. For a typical student project, the honest question is whether you need a VPS at all. A static or server-rendered site on a modern hosting platform's free tier (this blog runs that way) removes the server-maintenance burden entirely — a choice we've argued for in building reliable web apps without overengineering.
Quick Answers
Do my DigitalOcean student credits expire? Yes — per the June 12 notification, all credits provided through the GitHub Student Developer Pack expire July 31, 2026, regardless of when you redeemed them. Confirm your exact balance in the billing section of your dashboard.
Will I be charged automatically after July 31? Only if you leave paid resources running. The account itself is free; Droplets, databases, volumes, and snapshots are not. Destroy what you don't need and the bill is zero.
Can I download a snapshot of my Droplet as a backup? No. Snapshots restore only within DigitalOcean. Use rsync, SFTP, and database dumps to take data off-platform.
Can new students still claim the DigitalOcean offer? As of June 12 the pack page still lists it, but with credits expiring July 31 the practical value is shrinking by the day. Azure for Students or Heroku's student offer are safer starting points now.
The Real Lesson: Credits Are Marketing, Not Infrastructure
Step back from the deadline and there's a durable lesson here for anyone who builds on subsidized infrastructure — students today, startups on cloud credit programs tomorrow.
Free credits are a customer-acquisition budget, and budgets get reallocated. DigitalOcean ran this program for twelve years and grew the credit fourfold before winding it down inside seven weeks' notice; the 2024 purge of 40 pack partners shows the same pattern across the industry. None of this is bad faith — it's the predictable lifecycle of promotional spending, and the wave of capital flowing into AI infrastructure makes general-purpose subsidies an obvious place to trim. The June 8 exclusion of GPU and inference workloads from student credits says that part out loud.
The defense is the same one we recommend for any external dependency in what product teams need from engineering partners: treat the relationship as something that can end, and keep your exit cheap. In practice that means three habits, each of which costs almost nothing while the platform is free:
- Keep state exportable. If your data only exists in a format the vendor controls — like a snapshot you can't download — you don't have a backup, you have a hostage.
- Keep setup reproducible. A deploy script, a Dockerfile, or even a careful README turns "migrate the server" from a weekend of archaeology into an hour of work, the same way it pays off in balancing speed and maintainability.
- Know your real monthly number. The moment credits appear, find out what the bill would be without them. That number is the difference between a calm migration and a panicked one.
If you're affected, the timeline is generous enough to act calmly: audit this week, back up next week, migrate or shut down well before the last week of July. The students who treat this as a forced rehearsal of good operational habits will get more value from the exercise than the $200 was ever worth.
