What the Porter Startup Program Gives You
Porter offers up to $5,000 in credits toward its platform, the layer that provisions, deploys to and operates a Kubernetes cluster running inside your own AWS, GCP or Azure account.
AI Perks tracks it alongside $7.7M in credits across 194 companies.
That boundary is the whole story of this program. Porter does not host your application. Your cloud provider does, on infrastructure you own and a bill in your name, while Porter charges a separate platform fee for building, deploying and operating what runs there. The $5,000 covers that fee, not the compute underneath it.
That sounds like a limitation. It is closer to the opposite, as the stacking section shows. Availability and current terms vary, and both sit on getaiperks.com.
What Porter Is Actually For
Porter is an internal developer platform that gives you a Heroku-style workflow - connect a repo, push, get a running service with TLS, logs, metrics, autoscaling and preview environments - on top of Kubernetes that lives in your cloud account rather than the vendor's.
The category question is not Porter versus a rival. It is which of three postures you should be in:
- Fully managed PaaS. Someone else's account, someone else's cluster. Fastest to start, and the ceiling arrives when you need a VPC peer, a specific instance type, a compliance boundary or a committed-spend discount.
- Raw cloud plus your own Kubernetes. Maximum control, and a standing tax in engineer-weeks: cluster upgrades, IAM, ingress, secrets, autoscaler tuning, the on-call rotation that follows.
- The middle. Your account, your VPC, your instance types, your cloud contract, with the platform work operated by a vendor. That is the seat Porter sits in.
The honest test is whether you have a reason to own the account: enterprise buyers asking where data lives, a SOC 2 or HIPAA boundary, a GPU instance type a PaaS will not rent you, a cloud commit you already signed, or a services-heavy architecture that per-service pricing handles badly.
If none of those are true yet, a fully managed PaaS is cheaper and simpler. If any are, the middle posture is worth paying for, because the alternative is not free. It is a platform engineer you have not hired.
How Porter's Costs Behave at Scale
Porter produces two bills that scale on different curves: a platform fee that tracks your team and workload count, and a cloud bill that tracks provisioned capacity. The credit on this page touches the first one only.
For most teams the second bill is the larger one within months.
| Line item | Billed by | What drives it | Which credit applies |
|---|---|---|---|
| Platform fee | Porter | Workloads under management | Porter credits |
| Compute nodes | Your cloud provider | Node count and size, provisioned not used | Cloud credits |
| Kubernetes control plane | Your cloud provider | Flat, per cluster, per hour | Cloud credits |
| Load balancers | Your cloud provider | One per ingress, plus traffic | Cloud credits |
| NAT gateway | Your cloud provider | Hourly, plus every GB processed | Cloud credits |
| Managed databases | Your cloud provider | Instance size and storage | Cloud credits |
| Egress and cross-AZ traffic | Your cloud provider | Chatty services across zones | Cloud credits |
The trap in that table is the floor. A Kubernetes environment costs a fixed amount every month before a single container of yours is scheduled, far above the price of one small PaaS instance:
| Floor component | Approximate list cost | Notes |
|---|---|---|
| Managed control plane | Around $0.10 per hour, roughly $73 per month | Per cluster, idle or saturated |
| Smallest realistic node group | Two nodes, commonly $60 to $150 per month | Billed provisioned, not used |
| Application load balancer | Roughly $16 to $25 per month base | Plus a traffic component |
| NAT gateway | Roughly $32 per month base | Plus per-GB processing |
Illustrative AWS list figures at the time of writing. Rates vary by region and provider, so verify current pricing before modelling.
Call the floor $200 to $300 a month per environment, before your app. Separate staging and production clusters double it. That is not a Porter charge, it is what Kubernetes costs to keep the lights on anywhere, but founders routinely attribute it to the platform and reach the wrong conclusion. AI Perks lists what the credits cover; the floor is yours to design around.
The same structure that creates the floor also flattens the curve above it. Adding a tenth service to an existing cluster is a scheduling decision, not a new $25 line item, which is where per-service PaaS pricing starts to hurt and this model starts to win.
What Porter Credits Stack With
This is the cleanest stack in the DevOps category, because the two bills come from two different companies and neither credit overlaps the other. Porter credits pay the platform fee, cloud credits pay the infrastructure it manages.
That property is rare. On a fully managed PaaS your cloud credits are worthless, because the PaaS is the cloud provider's customer, not you. Deploying into your own account makes you the customer again, and cloud programs are the largest credit awards in startup land.
The full stack around a Porter deployment:
- Cloud credits cover compute, control plane, databases, networking and egress, usually the biggest number here
- Platform credits cover the Porter fee, the $5,000 in question
- Observability credits cover metrics, logs and traces, which get expensive fast at real pod counts
- Database and cache credits cover managed data services, in your account or outside it
- Security and compliance credits cover the scanning and audit tooling that motivated account ownership
A team holding cloud and platform credits together has funded both halves of its infrastructure. Which programs run in parallel, and which quietly disqualify each other, is why AI Perks is maintained as a live list rather than a folder of bookmarks.
What Founders Get Wrong About Deployment Credits
The most expensive mistake is assuming a platform credit covers the infrastructure, then discovering months later that the cloud bill was never subsidised at all.
Five patterns, roughly in order of cost:
Funding the wrong half. A $5,000 platform credit beside an unsubsidised cloud bill is the smaller problem solved. The infrastructure invoice is where the larger subsidy lives.
Adopting the model too early. Owning the account is a cost paid for real reasons: compliance, data residency, instance types, committed spend. Pre-revenue with no enterprise buyer, a simpler PaaS is cheaper and migrating later is a week, not a quarter.
Sizing node groups for peak. Cloud compute bills provisioned capacity, not consumed. A cluster sized for a launch day you have not had bills full rate every quiet night.
Leaving preview environments up. Per-pull-request environments are the best feature here and the easiest way to leak money. Without a time-to-live, a branch merged in March still holds nodes in September.
Treating the exit as free. Your cluster survives cancellation, which is a real advantage over a proprietary PaaS. The operator does not. Keep your manifests and Terraform in your own repo from day one, so leaving is a staffing decision rather than a rebuild.
On timing: model the unsubsidised number at 70% of credit consumed, not at 100%. Two bills means two renewal conversations, and they will not arrive together.
Where Porter Sits in a DevOps Credit Stack
Porter is one line in a DevOps stack, and rarely the largest one. The useful exercise is mapping which program covers which invoice before the architecture is locked in.
The DevOps listings on getaiperks.com put the platform, container and infrastructure programs side by side with their current terms. Three things are worth holding in view while reading them:
- The cloud side is the bigger number. A stack that only funds the platform fee has solved the cheaper problem.
- Partner routes exist. Much infrastructure credit moves through investor and accelerator channels rather than public pages.
- Guardrails are cheap on day one. Autoscaler minimums and preview environment expiry compound, and both resist retrofitting once services are live.
Frequently Asked Questions
How much is the Porter startup program worth?
Up to $5,000 in credits toward the Porter platform fee. How far it goes depends on how many workloads you put under management, not how much traffic you serve. The credit does not reduce your cloud provider bill, which is billed separately. Current amounts and terms are tracked at getaiperks.com.
Do Porter credits cover my AWS or GCP bill?
No. Porter deploys into an account you own, so compute, control plane, load balancers, databases and egress are invoiced directly by your cloud provider. That is the point of the architecture, and it means separate cloud credit programs apply cleanly to the larger half of your spend.
Is Porter cheaper than a fully managed platform?
Not at small scale. A Kubernetes environment carries a fixed monthly floor of roughly $200 to $300 per cluster before your app runs, against a few dollars for one PaaS instance. It gets cheaper as service count rises, because adding a workload to an existing cluster costs scheduling, not a new per-service charge.
What happens to my infrastructure if I stop using Porter?
The cluster keeps running, because it lives in your cloud account rather than the vendor's. What you lose is the operator: upgrades, autoscaling config, deploy pipelines and ingress management come back to you. Keeping your own Terraform and manifests in your repo makes that a staffing decision, not a rebuild.
Can I stack Porter credits with other startup credits?
Yes, and this is one of the cleanest stacks available, because the platform fee and the infrastructure bill come from different companies. Cloud credits, observability credits and database credits all apply to separate invoices. AI Perks tracks $7.7M in credits across 194 companies, including which programs conflict.
When should a startup move to deploying in its own cloud account?
When something forces it: an enterprise customer asking where data lives, a compliance boundary, a GPU instance type a managed platform will not rent you, a signed cloud commit, or enough services that per-service pricing hurts. Before any of those, a simpler platform is cheaper.
Own the account. Let someone else pay for the platform that runs it.