Module register
Each module is independently scoped, costed and delivered. The programme total is capped at A$22,000,000 — US$14,300,000 at A$1.00 = US$0.65.
Programme cap (AUD)
A$22,000,000
Programme cap (USD)
$14,300,000
Allocated across modules
$14,300,000
FX reference rate A$1.00 = US$0.65. Module prices are fixed in USD; the AUD cap moves with the prevailing rate at drawdown.
Investor deck
Ten pages covering milestones for all seven modules, per-module pricing against the A$22,000,000 / $14,300,000 programme cap, and the latest update for each module.
Fixed USD price per module, with the cumulative total tracked against the $14,300,000 programme cap. The register is fully allocated — no module can be repriced upward without reducing another.
| Module | Price (USD) | % of cap | Running total | Cap remaining |
|---|---|---|---|---|
| NEXUS CoreM01 | $3,400,000 | 23.8% | $3,400,000 | $10,900,000 |
| SENTRAM02 | $1,950,000 | 13.6% | $5,350,000 | $8,950,000 |
| ORACLEM03 | $2,250,000 | 15.7% | $7,600,000 | $6,700,000 |
| HELIXM04 | $2,100,000 | 14.7% | $9,700,000 | $4,600,000 |
| AEGISM05 | $1,500,000 | 10.5% | $11,200,000 | $3,100,000 |
| VERTEXM06 | $1,800,000 | 12.6% | $13,000,000 | $1,300,000 |
| ATLASM07 | $1,300,000 | 9.1% | $14,300,000 | $0 |
| Total allocated | $14,300,000 | 100.0% | $14,300,000 cap | $0 |
Cap utilisation
$14,300,000 of $14,300,000 · 100.0%
Equivalent to A$22,000,000 at the A$1.00 = US$0.65 reference rate.
How the seven modules differ, what each is judged on, and how they connect. Read across a row to compare one attribute; switch to the card view to read one module at a time.
| Attribute | NEXUS CoreM01 · $3,400,000 · 23.8% of cap | SENTRAM02 · $1,950,000 · 13.6% of cap | ORACLEM03 · $2,250,000 · 15.7% of cap | HELIXM04 · $2,100,000 · 14.7% of cap | AEGISM05 · $1,500,000 · 10.5% of cap | VERTEXM06 · $1,800,000 · 12.6% of cap | ATLASM07 · $1,300,000 · 9.1% of cap |
|---|---|---|---|---|---|---|---|
| Role in the platform | Reasoning substrate — the model everything else calls | Data supply chain — clean, licensed, provenance-tracked corpora | Perception — turning images, video, audio and documents into structured facts | Execution — turning a goal into a completed sequence of real actions | Control plane — safety, alignment, policy and compliance evidence | Delivery fabric — running the platform on datacentre, on-premise and edge hardware | Surface — the workspace, APIs and connectors customers actually touch |
| Key difference | The only module that produces model weights. Its capability ceiling sets the ceiling for the whole platform. | The only module that touches raw source data and owns licence provenance end to end. | Extends the platform beyond text; the only module that reads physical-world signals. | The only module that acts on external systems rather than returning an answer. | The only module with authority to block another module's output in production. | The only module concerned with where compute physically runs and what it costs. | The only module built for end users rather than for other modules. |
| Outcome measured on | Benchmark-leading reasoning at a serving cost low enough for always-on enterprise use | Every training token traceable to a licensed source with an auditable lineage record | Field-grade accuracy on document, inspection and vision tasks against expert baselines | Long-running workflows completed unattended with a full, replayable action audit trail | Audit-ready evidence packs mapped to EU AI Act and ISO/IEC 42001 controls | Identical model behaviour on edge hardware at a fraction of datacentre cost per request | Time-to-first-value under two weeks for a new enterprise deployment |
| Integration notes | Exposed through one versioned model contract (Python, TypeScript, gRPC), so checkpoint upgrades reach every module without integration work. | Publishes versioned dataset manifests; downstream trainers pin a manifest hash so any model can be reproduced exactly. | Encoders attach to NEXUS as adapters rather than a separate model, so multimodal input arrives through the same contract as text. | Runs every tool call through the AEGIS policy gate, and surfaces run state to ATLAS for human review and takeover. | Sits inline on every inference and tool call as a policy gate; nothing reaches a customer without passing through it. | Consumes compiled NEXUS artefacts and schedules them across heterogeneous fleets behind a single endpoint. | Aggregates every upstream module behind one authenticated API and workspace, with SSO, RBAC and connector packs for common enterprise systems. |
| Revenue model | Per-million-token licensing plus on-premise weight licences | Dataset and synthetic-corpus licensing to third-party model builders | Per-asset processing fees and vertical inspection packages | Per-workflow-run pricing with enterprise seat bundles | Compliance module licence, priced per regulated deployment | Managed compute margin plus edge appliance licensing | Seat-based SaaS subscription and connector add-ons |
| Principal risk | Training-run cost and schedule risk on the 70B production run | Shifting data-licensing law in key jurisdictions | Accuracy variance across unseen equipment and document classes | Reliability of multi-step autonomy in unpredictable environments | Regulatory targets moving faster than certification cycles | Hardware supply and accelerator vendor lock-in | Depends on upstream module maturity for its own delivery date |
| Depends on | M02 SENTRA | None — terminal module | M01 NEXUS CoreM02 SENTRA | M01 NEXUS CoreM03 ORACLEM05 AEGIS | M01 NEXUS CoreM02 SENTRA | M01 NEXUS CoreM05 AEGIS | M01 NEXUS CoreM03 ORACLEM04 HELIXM05 AEGISM06 VERTEX |
| Feeds into | M03 ORACLEM04 HELIXM05 AEGISM06 VERTEXM07 ATLAS | M01 NEXUS CoreM03 ORACLEM05 AEGIS | M04 HELIXM07 ATLAS | M07 ATLAS | M01 NEXUS CoreM03 ORACLEM04 HELIXM06 VERTEXM07 ATLAS | M07 ATLAS | None — terminal module |
| Status | In Build · 62%Q3 2025 — Q2 2027 | In Build · 74%Q2 2025 — Q4 2026 | In Build · 41%Q4 2025 — Q3 2027 | In Build · 35%Q1 2026 — Q4 2027 | In Build · 48%Q3 2025 — Q2 2027 | In Design · 29%Q2 2026 — Q1 2028 | In Design · 22%Q3 2026 — Q2 2028 |
Dependencies read left to right through the pipeline: SENTRA supplies data, NEXUS supplies reasoning, AEGIS gates every output, and ATLAS is the only module a customer touches directly.
Foundation model and inference engine
The reasoning substrate every other module depends on: a 70B-parameter mixture-of-experts foundation model with a custom low-latency inference runtime.
Investment (USD)
$3,400,000
23.8% of the $14,300,000 cap
Data ingestion and synthetic corpus pipeline
The industrial-scale data refinery that feeds NEXUS: licensed acquisition, provenance tracking, deduplication, and governed synthetic data generation.
Investment (USD)
$1,950,000
13.6% of the $14,300,000 cap
Multimodal reasoning and vision stack
Vision, document and sensor understanding fused into the core reasoning loop, so the platform can act on the physical and paper world, not just text.
Investment (USD)
$2,250,000
15.7% of the $14,300,000 cap
Autonomous agent orchestration layer
Long-horizon agents that plan, call tools, recover from failure, and hand work between each other with a durable execution record.
Investment (USD)
$2,100,000
14.7% of the $14,300,000 cap
Safety, alignment and compliance framework
The control plane that makes the platform sellable into regulated markets: red-teaming, guardrails, evidence generation and jurisdictional policy packs.
Investment (USD)
$1,500,000
10.5% of the $14,300,000 cap
Edge deployment and compute fabric
The distribution layer: distilled models and a scheduling fabric that run the platform on-premise, air-gapped, or at the edge with the same API.
Investment (USD)
$1,800,000
12.6% of the $14,300,000 cap
Enterprise interface and integration suite
The commercial surface customers actually touch: workspace, admin console, connectors, billing and the public API that turns the stack into revenue.
Investment (USD)
$1,300,000
9.1% of the $14,300,000 cap
