One product. Three common use cases.
Three different reasons to leave, one answer underneath: AWS-compatible APIs on infrastructure you control. These are the three we are asked for most.
Cloud at the edge, for air-gapped and denied environments.
Mission software was built for AWS. Operations happen where AWS isn't. Spinifex closes that gap: an AWS-compatible cloud running on the hardware already on-site, hardened for disconnection.
Cloud-native software assumes the cloud is always there. Defence reality is the opposite. Degrade the link and inference stalls, the EW pipeline backs up behind it, and the command app the operators are actually looking at goes dark.
An AWS-compatible cloud on operator-owned hardware: air-gapped, hardened, deployable to site. It serves the same EC2 / S3 / VPC / EBS APIs your software already calls, so the mission code ships to the edge without a rewrite.
ISR & EW providers, command software ISVs, defence integrators, and sovereign capability programs running mission software at the edge.
Forward operating bases, tactical vehicles, ships, drones, ground stations, sovereign data centres. Where the operation is.
Reclaim sovereignty of your mission-critical workloads.
AWS got you to market. Now leaving is the project most teams never start. Spinifex takes the rewrite out of the equation: repoint, don't re-platform.
The bill scales faster than usage. Every AWS-specific service you adopt deepens the lock-in that produced it, and regulators keep asking harder questions about jurisdiction. The case to leave gets stronger every quarter, and the rewrite cost kills it every quarter.
Move workloads into a data centre you own, a private cloud, or a Neocloud partner by changing an endpoint rather than the software. The APIs do not move, the bill becomes predictable, and the core is open source you can audit line by line.
Regulated industries: financial services, healthcare, gov, energy. Platform teams running mature AWS workloads. Organisations with sovereignty or data-residency mandates.
Smaller, more predictable spend. End-to-end control of your stack. Data in the jurisdiction you choose. No next lock-in, because the core is AGPL and yours to keep.
Break out of the hyperscalers.
Run your AWS-native training and serving stack on a Neocloud's GPU fleet. The code and the APIs stay exactly as they are; the bill halves.
GPU on AWS is the most expensive, queue-bound, region-locked place in cloud. Neoclouds have the H100s, the H200s, and the B200s for less. But your stack is built for AWS, and re-platforming it eats months you don't have.
Spinifex sits on a Neocloud GPU fleet and exposes EC2 / EBS / S3 / VPC. Your existing AWS training and serving pipelines run unmodified. Terraform, the AWS CLI, the SDKs all just work.
AI labs, model-serving startups, ML platform teams running large GPU fleets and looking at their AWS bill.
Neocloud GPU capacity runs 50%+ cheaper than the same iron on a hyperscaler, with none of the egress traps. Spinifex makes that capacity addressable from the stack you already have.