Architecture

On-premise AI for schools: data sovereignty, predictable cost, and the on-site server case study

Updated May 2026 · 7 min readIndiLearn · Architecture
Affordable
on-site server cost — pays backes in under 4 months at 600 students
minimal power
on-site server power draw under AI load vs significantly more power for a GPU rig
$0
Ongoing AI running costs after hardware — no per-query billing ever
Predictable
Fixed cost model — no per-student or per-query billing ever

In this article

  1. The cloud economics problem for schools
  2. Why the on-site server is the right hardware for school AI
  3. Why modern on-site hardware is now capable enough
  4. Cost comparison: cloud vs on-premise at scale
  5. Data sovereignty: the procurement argument cloud AI cannot make
  6. How IndiLearn deploys on the school's on-site server

The standard pitch for AI in education is a cloud subscription. Monthly per-student fees. API costs that scale with usage. Privacy policies that require lawyers to interpret. Terms of service that change without notice.

IndiLearn's architecture is a deliberate rejection of that model. Every school that deploys IndiLearn in its full on-premise configuration gets a on-site server in their server room, runs all AI inference on their own network, has no ongoing AI running costs, and holds a structural guarantee that no student data ever leaves the building.

This is not primarily a cost argument, although the cost case is strong. It is a procurement argument. Cloud-based AI education tools cannot credibly tell a school principal or DET procurement officer that student data is safe, because the structural reality of cloud inference makes that claim untenable. On-premise inference makes it true by design.

The cloud economics problem for schools

Cloud AI APIs are billed for every use — per unit of text processed. For an education platform, this creates a cost structure that scales with exactly what you want to scale: student use. Every lesson, every student submission, every feedback generation is a cost event. The more the platform is used, the higher the bill.

Cloud AI platforms bill for every interaction — every student submission, every feedback generation, every lesson. That cost grows the more teachers actually use the platform, creating a perverse incentive to limit usage. Add compliance overhead, platform fees, and ongoing subscription costs, and the total is substantial.

Why the on-site server is the right hardware for school AI

The conventional assumption for AI inference hardware is a GPU server. High-VRAM NVIDIA cards, dedicated cooling, server rack, specialist installation. This is the right answer for some use cases. It is the wrong answer for a primary school's server room.

on-site hardware on-site server is the right answer for school AI for three reasons: cost, power, and form factor.

Why modern on-site hardware is now capable enough

Modern compact hardware has reached a capability threshold where on-site AI inference is now practical for schools. Hardware that previously required enterprise GPU infrastructure — expensive, power-hungry, needing specialist installation — can now run entirely on modest, quiet, affordable equipment that fits in a standard IT cabinet.

IndiLearn's architecture takes advantage of this shift. The result is AI inference that runs locally, costs nothing per query, and never touches an external network.

What on-site on-site memory runs

A on-site server with on-site on-site memory comfortably runs capable open-source models — a model that rivals cloud API quality for structured generation tasks like feedback writing and content generation. For IndiLearn's specific use cases, where the inference tasks are bounded (feedback against a rubric, decodable word generation within a grapheme set), even smaller models provide the quality required.

Cost comparison: cloud vs on-premise at scale

The economics are simple

Cloud AI bills for every query — every lesson, every student, every year. On-site AI is a one-time hardware investment with no ongoing running costs. The larger the school and the more the platform is used, the more the on-site model saves. Contact us to discuss the economics for your school.

The hardware cost does not increase with student count. The cloud cost scales linearly. For any school with more than approximately 200 students using the platform meaningfully, the on-premise model delivers cost savings within the first year.

Data sovereignty: the procurement argument cloud AI cannot make

Cost is the obvious argument. The more important one is data sovereignty.

When a school runs AI inference on a cloud API, the following things are structurally true regardless of what the vendor's privacy policy says: student data leaves the school network, is processed on servers the school does not control, is subject to the laws of the jurisdiction where those servers are located, and is handled under terms of service that can change at any time.

On-premise inference makes none of this true. The data is processed on a machine the school owns, on a network the school controls, subject to Australian law, under terms that cannot change because there is no third-party cloud provider in the loop.

The DET procurement context

Queensland DET and other state education departments require specific permission for student data to be processed by external platforms. For a cloud AI tool, every student interaction is technically a data transmission to an external processor. Schools must obtain and manage consent, assess vendor compliance, and monitor usage. For an on-premise tool, there is no external transmission — and therefore no consent barrier, no vendor assessment, and no monitoring overhead. The procurement conversation is structurally simpler.

How IndiLearn deploys on the school's on-site server

IndiLearn's on-site deployment uses an on-site inference server. our on-site inference layer exposes a standard API at the local network address, meaning IndiLearn's application code communicates with the on-site AI model in exactly the same way it would communicate with a cloud API. The privacy guarantee is architectural, not application-level — no code change is required to move from cloud to on-premise.

The CoachingProvider abstraction in IndiLearn's codebase ensures this. The application never sees whether inference is running locally or in the cloud — it calls the provider interface and receives a response. The school's on-site server configuration swaps one implementation for another without the teacher, student, or application being aware of the change.

What this means for a principal

A principal considering IndiLearn's on-premise configuration can tell the school community, genuinely and verifiably: no student data from this tool leaves our network. Not "our vendor says it doesn't." Not "we believe they comply." The data physically cannot leave — the processing happens on our hardware, on our network, under our control. That is a procurement statement ChatGPT cannot make.

Your school's AI. On your hardware. Under your control.

Register your school's interest for the on-premise pilot in 2026.

Register your school

Related articles