How Kestrel K1 units would actually get built: contract manufacturer landscape, tooling & NRE, the volume pricing curve, and an NPI timeline gated by the silicon question in the architecture doc.
The architecture doc's cost reference says what the parts should cost. It doesn't say how those parts turn into manufactured units — who builds them, what it costs to get a production line stood up before the first unit ships, how the per-unit price actually moves with order volume, or how long that takes. This doc covers that gap: contract manufacturer (CM) landscape, tooling/NRE, the volume price curve, and a phased NPI (new product introduction) timeline.
Two paths to a manufacturable SoC, and they don't share a timeline, a budget, or a risk profile. Every schedule and cost figure later in this doc assumes Path A — Path B is the honest long-lead alternative, not a near-term option.
| Path | What it is | Lead time to samples | NRE |
|---|---|---|---|
| A — Off-the-shelf ARM SoC | Substitute an existing mobile-class part (Cortex-A73/A55-class, same tier the SteamOS POC uses reference hardware for) for KX2 at launch | 4–6 months | ~$0 (uses vendor's existing die) |
| B — Custom KX2 tape-out | Fabless design through a foundry (mask set, IP licensing, verification, packaging) for the bespoke part the architecture doc costs at $18.00 | 18–24+ months | $15M–$40M+ |
Handheld-class consumer electronics manufacturing sorts into a few recognizable tiers. These are industry classes with illustrative reference points, not vendors under contract for Kestrel — the point is knowing which tier a real RFQ round should target.
| CM class | Reference point | Typical MOQ | Fit for Kestrel |
|---|---|---|---|
| Tier-1 EMS (large-scale) | Foxconn/BYD Electronics-class contract assemblers | 100K+ units | Overbuilt for a first run; the line-setup cost only pays off at scale Kestrel doesn't have yet |
| Boutique handheld ODM | Shenzhen-based ODM class behind Anbernic/Miyoo/Retroid-style retro handhelds | 1K–5K units | Best fit — already tools for PSP-scale enclosures, dual analog sticks, and small-battery handhelds at low MOQ |
| White-label PCBA + local assembly | Contract PCB assembly house + separate final-assembly shop | 500–2K units | Lowest MOQ, but Kestrel owns integration risk the ODM would otherwise absorb |
Engineering estimate for a Path A first run at a boutique handheld ODM, before any per-unit BOM spend. Excludes Path B silicon NRE (§02).
| Item | Category | Est. Cost |
|---|---|---|
| Injection mold tooling (shell + mid-frame, 2-cavity) | Mechanical | $60,000–$110,000 |
| PCBA test jig / ICT + FCT fixtures | Test | $15,000–$30,000 |
| Display/touch module bonding fixture | Mechanical | $8,000–$15,000 |
| FCC Part 15 + Wi-Fi/BT module co-cert | Certification | $25,000–$45,000 |
| CE / RED (EU) certification | Certification | $20,000–$35,000 |
| RoHS / REACH compliance testing | Certification | $8,000–$15,000 |
| UN38.3 battery transport testing | Certification | $10,000–$18,000 |
| EVT/DVT/PVT engineering builds & NPI management | NPI | $120,000–$200,000 |
| Firmware/software compliance & QA pass | NPI | $40,000–$70,000 |
| Total est. first-run tooling & NRE | ~$306,000–$538,000 |
The $96.80 figure in the architecture doc is explicitly low-volume. SoC, display module, and battery cell — the three line items flagged there as highest pricing risk — are also the three with the steepest volume discounts.
| Order volume | Est. unit BOM | vs. low-volume baseline |
|---|---|---|
| 1,000 units | $96.80 | Baseline |
| 10,000 units | ~$84.00 | −13% |
| 50,000 units | ~$74.00 | −24% |
| 100,000+ units | ~$68.00 | −30% |
Standard four-gate hardware NPI process, sequenced so tooling commitment happens only after the SteamOS POC has validated the launcher/suspend/controller UX this whole plan is building toward.