Enterprise
From one line to a global packaging network.
Standardising an autonomy layer across fabs is an organisational problem before it is a technical one. Chipira is built for the procurement, qualification and change-control realities of a manufacturing enterprise.
The enterprise problem
Twelve sites, twelve versions of the truth
A multi-site packaging organisation does not have one process. It has one nominal process and a dozen local interpretations of it, each with its own tool mix, its own tribal recipe adjustments and its own definition of a marginal package.
That divergence is expensive in a way that never appears on a single site’s P&L: a ramp learned in one fab is re-learned from scratch in the next, a defect signature solved in Kaohsiung is rediscovered in Penang, and a departing engineer takes a decade of judgement with them.
The enterprise case for Chipira is the same case as the single-line one, compounded. Every supervised correction becomes a durable, transferable asset owned by your organisation rather than by an individual — and, where you contract for it, that learning moves between your own sites under your own governance.
- One governance model across every site and module
- Site-level isolation with organisation-level visibility
- Cross-site model transfer under your contract, never by default
- Craft captured as an institutional asset, not a personal one
What enterprise adds
Beyond the Fab plan
Custom process and defect models
Models trained on your package families, your defect taxonomy and your classification standards, versioned and owned inside your tenant.
Fab-edge fleet management
Fleet-wide model versioning, staged rollout, pinning, and per-site release channels — so a model reaches your most sensitive line last, by design.
Multi-fab governance
Organisation-wide roles, envelope templates, approval hierarchies and audit consolidation across every site running Chipira.
Service levels
Contractual uptime, response and escalation commitments, with named support engineers who understand your package flow.
Deployment options
Private VPC or fully on-premise control plane, air-gapped edge configurations, and signed offline model delivery for restricted sites.
Outcome-based components
A share of measurable yield, scrap, known-good-die and ramp-time improvement, structured so both sides are paid for the same result.
Procurement
What your teams will ask for
And what we bring to the first meeting.
- Security review
- Architecture documentation, data-flow diagrams, tenancy model, encryption posture, SSO and RBAC design, and the SOC 2 programme status. On-premise and air-gapped options documented up front.
- IP protection
- Contractual guarantees on process-IP isolation, explicit terms on model training scope, and a technical architecture where the default is that nothing leaves your building.
- Qualification
- A shadow-mode evaluation protocol your quality organisation can run: measured accuracy against your existing baseline, on your packages, before any authority is granted.
- Change control
- Versioned action envelopes, pinned model releases, staged rollout, and an audit trail that satisfies a change-control board rather than annoying one.
- Commercial
- Multi-year agreements with usage ramps, annual prepay discounts, and outcome components on the highest-value workflows.
- Exit
- Documented data export, model artifact handling on termination, and a written description of what a wind-down actually looks like. Ask us for it early.
Rollout
How a multi-fab standardisation actually goes
Slowly, then quickly. The order matters more than the pace.
-
One line, one workflow, one metric
A single site proves the wedge in shadow then assist mode. Nothing is standardised on until something is proven.
-
Second module, same line
The adjacent step on the same package flow comes online, demonstrating that the compounding is real rather than theoretical.
-
Second site, same workflow
The proven workflow is replicated at a second site with its own tenant, its own envelope and its own baseline — testing transferability honestly.
-
Fleet and governance
Organisation-level roles, envelope templates and release channels are established, and the audit consolidation your quality organisation needs is stood up.
-
Network effects, under contract
Only once governance exists do cross-site learning mechanisms get switched on — explicitly, contractually, and revocably.
Quality systems directorMulti-site OSATWe can qualify a tool in six weeks. Qualifying software that writes to that tool is a different conversation, and most vendors have never had it.
Composite drawn from design-partner and industry conversations. Illustrative, not a customer endorsement.
Commercial shape
How enterprise engagements are structured
- $800k–$9M Typical enterprise landing annual contract value, depending on site count, module scope and autonomy level.
- >130% Net revenue retention target, driven by module, line and site expansion rather than price increases.
- 15–20% Indicative annual prepay discount against monthly commercial terms.
- <13 mo Target CAC payback on enterprise engagements.
Commercial figures are Chipira’s internal targets and indicative structures [ASPIRATIONAL], not published results. Every agreement is negotiated against a specific scope.
Non-negotiables
The four things we will not trade
Enterprise deals create pressure to loosen defaults. These four are not available for negotiation, in any contract, at any price.
- Process IP stays yours No training on your process data outside your tenant without an explicit, separately signed agreement. Not for “product improvement”, not for benchmarks.
- The envelope is yours Only your process owners can widen an action envelope. Chipira support cannot do it for you, even in an incident.
- The audit trail is immutable It cannot be edited, purged on request, or made selectively invisible. If that is a problem, we are the wrong vendor.
- Autonomy is revocable Any autonomy level can be withdrawn instantly by your organisation, and is withdrawn automatically when measured accuracy drifts below its gate.
Plan comparison
Line, Fab and Enterprise
| Line | Fab | Enterprise | |
|---|---|---|---|
| Scope | One tool or bonder line | Whole line, all agents | Multi-fab |
| Agents | One capability | Full seven-agent loop | Full loop, custom agents |
| Package-and-line twin | — | Included | Included |
| Model customisation | Tuned on your data | Tuned on your data | Custom architectures |
| Edge footprint | Single system | Line fleet | Multi-site fleet |
| Support | Business hours | Priority | Named engineers, SLA |
| Commercial | $18,000 / tool / mo | $120,000 / mo | Custom, $800k–$9M ACV |
Enterprise questions
The procurement conversation
-
Not yet, by the usual vendor-risk criteria, and we will not pretend otherwise. Chipira is an early-stage company. The right answers to that risk are source-code escrow, documented exit, on-premise deployment and a pilot that proves value before you depend on us — all of which we will discuss on the first call.
-
Yes. An air-gapped configuration runs the control plane on your premises with models delivered as signed offline artifacts. You give up automatic fleet updates and cross-site learning; some sites should give those up.
-
Per-tenant, per-site scoping means data residency follows the site by default. A site in Taiwan trains on Taiwanese data in Taiwan unless you contract for something else.
-
A defined share of a measured improvement — yield points, scrap reduction, known-good-die loss or ramp weeks — against a jointly agreed baseline and measurement method, capped and floored so neither side is exposed to a measurement dispute.
Start narrow, expand relentlessly
Land one workflow. Own the loop.
A Chipira engagement begins with a single wedge workflow, a shadow-mode baseline and one signed success metric. Everything after that is expansion.