How Long Does Hardware Product Development Take? A Phase-by-Phase Timeline
Discovery to PVT typically runs 12–24 months. Here is the week-by-week breakdown, the parallel tooling clock, and where schedules actually slip.
By Axon Labs Engineering
Ask five hardware teams how long their product took and you'll get five different answers built on five different definitions of "done." The honest range for a connected consumer product is 12 to 24 months from discovery to production validation — but the number that actually matters isn't the total. It's which weeks are compressible and which aren't. This is the phase-by-phase breakdown we use to build real schedules, not optimistic ones.
Key takeaways
A connected product runs 12–24 months discovery to PVT, with PVT alone taking 3–6 months once tooling and pilot-line validation enter the picture (OpenBOM; TechDesign).
Prototype counts scale by phase: 5–12 units at EVT, rising to 50–500 units at PVT to verify real manufacturing yield — a different kind of build each time, not just "more of the same."
Tooling runs on its own clock and often sets the critical path: aluminum prototype tools take 2–4 weeks, but domestic steel production tooling takes 8–12 weeks — 12–20 weeks offshore — and that clock has to start before DVT ends, not after.
What do EVT, DVT, and PVT actually validate?
EVT, DVT, and PVT are sequential validation gates, and each answers a different question with its own kind of evidence (OpenBOM). EVT validates engineering feasibility on intended components. DVT validates the complete design against product requirements, including regulatory compliance testing on production-grade materials and tooling. PVT validates the manufacturing line itself, running a pilot batch to confirm the factory can hit yield at scale.
That progression is the same one we walk through in our complete hardware development guide — this article is the zoomed-in schedule view: how many weeks each gate actually consumes, and where the padding needs to go.
The week-by-week schedule for a connected product
Representative phase durations for a moderate-complexity connected device. Actual durations scale with component count, radio complexity, and certification class. Sources: OpenBOM; TechDesign, 2026.
Phase
Typical duration
What ends it
Discovery & feasibility
4–6 weeks
Architecture, risk register, cost model signed off
Proof of concept (POC)
6–8 weeks
Highest-risk assumption resolved in working hardware
EVT (engineering validation)
8–12 weeks
5–12 prototypes prove the design on intended components
DVT (design validation)
10–16 weeks
Near-final units pass spec and pre-certification testing
PVT (production validation)
12–24 weeks
50–500 units confirm the factory hits yield at scale
PVT alone typically runs 3–6 months (TechDesign) — longer than most first-time teams budget for the entire back half of the program. It’s not one long test; it’s tooling bring-up, pilot-line setup, yield learning, and the paperwork a factory needs before it will commit to volume.
Why prototype counts change the schedule, not just the cost
On average, 5 to 12 prototypes complete EVT, while PVT prototype quantities typically range from 50 to 500 units to verify mass-production yield (Seeed Studio). That jump — a dozen hand-built boards to hundreds of factory-run units — is itself a schedule event: it needs a pilot line stood up, fixtures built, and operators trained before a single unit ships, none of which happens in parallel with the engineering that precedes it.
This is the same logic behind the iteration-count cost multiplier in hardware budgets: each phase isn’t just later, it’s a materially different kind of build, and treating PVT as "EVT but more units" is the single most common scheduling mistake we see in first-time programs.
Tooling runs on a separate clock — and it usually sets the critical path
Aluminum prototype tooling takes 2–4 weeks and costs $1,500–$8,000; hardened steel production tooling takes 8–12 weeks domestically, or 12–20 weeks offshore including shipping, and costs $15,000–$120,000+ (Agilian). For a product with several molded parts, the schedule is governed by the slowest tool in the package — not the average.
Typical 2026 tooling lead times by class and sourcing region. Sources: Agilian; Jaycon 2026 Injection Molding Pricing Report.
The fix is structural, not aspirational: release the PO for steel tooling as soon as DVT geometry is frozen, running it in parallel with certification testing — rather than sequencing tooling after DVT completes. That single scheduling decision recovers 6–10 weeks on most connected-product programs.
Where certification sits in the schedule
DVT is where compliance testing — FCC, CE, RoHS, and other regulatory certifications — happens, run on production-grade materials and tooling, not early prototypes. That sequencing is deliberate: pre-DVT boards rarely reflect the RF and EMC behavior of the shipping product, so testing them wastes lab time confirming problems tooling changes will fix anyway.
The scheduling risk is the same one we cover in the Cyber Resilience Act engineering guide: a failed pre-scan means a board spin plus a new lab queue slot, and test labs book out weeks in advance. Reserve the slot before DVT hardware exists — cancelling is free; missing the window costs a month.
What can actually be compressed — and what can't
Compressible: discovery and POC. Tight scoping and a decisive risk-resolution plan can pull these phases in by weeks — this is where a feasibility-first approach pays off directly.
Compressible: EVT-to-DVT handoff. Starting DVT test-plan and fixture design during EVT, instead of after it closes, overlaps work that many teams run sequentially for no real reason.
Not compressible: tooling lead time. Steel doesn’t cut faster because the schedule is tight. Order early against frozen geometry, or don’t order at all.
Not compressible: certification lab queues. You can expedite a slot at a premium; you cannot skip the physics of an EMC chamber booking calendar.
Rarely compressible: PVT yield learning. A pilot line either hits target yield or it doesn’t — rushing this phase just moves the failure downstream, into a customer’s hands.
The schedule isn't set by your fastest engineer. It's set by your slowest mold.
The bottom line
Budget 12–24 months for a connected product, with PVT alone commonly taking 3–6 months once tooling and pilot-line validation are counted.
Prototype counts jump by an order of magnitude between EVT (5–12 units) and PVT (50–500 units) — plan the pilot-line stand-up as its own milestone, not a footnote.
Tooling and certification run on physical clocks the schedule can’t compress — the only lever is starting them earlier, against frozen geometry, in parallel with the phase that follows.
If you’re building a schedule for a real program and want the padding in the right places, start with a discovery & feasibility phase — we’ll map your specific tooling and certification critical path before the calendar gets committed to stakeholders.
Frequently asked questions
How long does it take to develop a hardware product?
Typically 12–24 months from discovery to production validation for a connected consumer product. Discovery and POC run 10–14 weeks combined, EVT 8–12 weeks, DVT 10–16 weeks, and PVT alone commonly takes 3–6 months once tooling and pilot-line validation are included (OpenBOM; TechDesign).
What is the difference between EVT, DVT, and PVT timelines?
EVT (8–12 weeks) validates engineering feasibility on 5–12 prototypes. DVT (10–16 weeks) validates the complete design against requirements, including certification testing, on near-final units. PVT (12–24 weeks) validates the manufacturing line itself using 50–500 pilot units to confirm production yield.
How long does injection mold tooling take?
Aluminum prototype tooling takes 2–4 weeks. Hardened steel production tooling takes 8–12 weeks domestically or 12–20 weeks offshore including shipping (Agilian, 2026). Quoted lead times often exclude DFM sign-off and sampling iteration, which commonly add several more weeks in practice.
Why do hardware development schedules slip?
Most commonly because tooling and certification are sequenced after the phase that precedes them instead of in parallel with it. Steel tooling and EMC lab queues run on fixed physical clocks; ordering tooling only after DVT fully closes, rather than against frozen DVT geometry, is the single most common source of schedule slip.
How many prototypes are needed for hardware product development?
On average 5–12 prototypes complete EVT, while PVT requires 50–500 units to verify mass-production yield (Seeed Studio, 2025). The jump between these counts is a schedule event on its own — it requires a pilot line, fixtures, and trained operators, not just more parts.
If the product has to ship, talk to the team that builds for that outcome.
Senior engineer on the first call. NDA before technical detail. References available under NDA after qualification. Or start with a fixed-fee feasibility study.
