Ask any automotive OEM whether they practice supplier co-development and the answer is always yes. Suppliers sit in the concept review. They get the CAD model. They comment on the DFMEA. Some have engineers resident on the customer's site.
Then start of production arrives, the part comes in eight percent over its cost target, and the supplier who helped design it negotiates the first renewal like a company that knows you have nowhere else to go.
Co-development isn't failing because suppliers were brought in too late. It's failing because they were brought in early without being given any real decisions to make.
Timing Was Never the Variable
When co-development disappoints, the explanation is almost always the calendar. Procurement and suppliers get a look in only after the concept has frozen, so the one thing left to influence is price.
The evidence doesn't back that up. Research on supplier involvement in new product development — including the work of van Echtelt, Wynstra and van Weele in the Journal of Product Innovation Management — found that what drives results isn't how early a supplier is engaged, but how much of the design they're actually handed. Their distinction is the useful one: "early involvement" just moves the supplier up the calendar, while "extensive involvement" hands them real design responsibility. Bringing a supplier into the room six months earlier changed very little. Giving them a defined piece of the design to own improved cost, time-to-market and quality consistently.
Early involvement is a calendar decision. Extensive involvement is a power decision. Only one of them moves the cost curve.
This matters more in automotive than almost anywhere else. BMW Group's VP for Global Sustainability puts the figure at 70 percent of a vehicle's value coming from suppliers; Maxime Picat, formerly Stellantis' purchasing director, has put it as high as 84 percent of a car's cost. And the pattern isn't new. Analysis cited by Aberdeen found that as much as 80 percent of a product's total cost is committed by the end of the design process, even though concept and design engineering account for only 5 to 8 percent of development spending (CAM-I, cited in the Journal of Business Research).
If four-fifths of the cost is set by decisions nobody in procurement ever priced, co-development is not a collaboration initiative. It's the cost structure.
Three Moves That Actually Change the Outcome
Most programs already have the workshops, the NDA and the joint steering committee. What's missing is a clear statement — written by procurement, not engineering — of who gets to decide what. These three moves are what make it real.
-
Delegate design tasks, not design opinions
The line between a program that works and one that's just going through the motions is simple. Early involvement means the supplier sees the design sooner. Extensive involvement means the supplier designs a defined subsystem against a functional specification, with the buyer keeping architecture and interfaces.
The play. Before concept freeze, split the commodity into three buckets:
- Buyer-owned design — architecture, interfaces, anything that differentiates the vehicle or constrains other systems. Designed in-house, sourced to print.
- Co-designed — subsystems where the supplier has real process or materials expertise. Supplier proposes, buyer approves at defined gates only.
- Supplier-owned design — the classic black-box part. Buyer specifies function, envelope, interface and validation requirements. Supplier owns everything inside the boundary.
Write that split into the sourcing strategy document. Not the meeting minutes — the document that decides what gets bought and how.
The judgment. If your engineers review and redline every supplier drawing line by line, you don't have co-development. You have a supplier doing drafting work at engineering rates while your own engineers stay the bottleneck. Delegation is uncomfortable, because it means accepting a design you wouldn't have drawn yourself. That discomfort is the price of the capability you're buying.
-
Price the interface, not the part
Once design authority moves outside, unit price stops being the main lever. What decides whether the program lands on budget is who owns the interface definition, who carries validation risk, who owns the tooling, and who pays when the interface moves.
The play. Split the money into three pools and contract them separately:
- NRE and tooling — paid against defined deliverables and milestones, with tooling ownership spelled out.
- Piece price — benchmarked against a should-cost model, not against last year's quote.
- Change liability — a clause that says: if the buyer changes an interface after gate X, the buyer picks up the rework; if the supplier's design fails validation, the supplier does.
Then settle the IP position in the same document. The buyer owns integration and interface IP. The supplier owns component IP but grants a perpetual, transferable license. Joint inventions get assigned up front, not negotiated the moment they turn out to be worth something.
The judgment. Co-developed parts blow past budget on change orders, not on piece price. A program with a clean change-liability clause and a mediocre unit price will beat one with a great unit price and a vague clause — every time.
-
Keep a second option alive, or pay for the monopoly
Co-development creates a single source almost by default. By the time you reach SOP, the supplier holds the tacit knowledge, the validation data and the tooling, and your leverage is whatever your contract managed to preserve.
The play. Split the category by what's at stake. For high-value systems that genuinely differentiate the vehicle, accept single-sourcing — but design for second-sourceability from day one: documented interfaces, transferable tooling, validation data in escrow, and a pre-agreed right to nominate a second source after a set period or volume, with the licensing fee agreed now rather than when you need it. For everything below that line, don't co-develop at all. Standardize, specify, and put it out to bid.
And here's the part that stings. McKinsey's August 2025 analysis found that China's EV-focused OEMs reach start of production in roughly 24 months, against 45 months for a mass-market legacy OEM and 53 for a premium one — and some programs are now aiming at 18 months or less. Part of that gap, McKinsey notes, comes from pulling suppliers in at the concept stage, worth up to four months of schedule on its own (McKinsey, August 2025). That isn't a warmer relationship. It's fewer decisions sitting in a queue inside the OEM.
The judgment. Second-sourceability costs one to three percent up front. It's the cheapest insurance you'll ever buy, and it costs far less than the margin you hand over at the first renewal.
What Co-Development Actually Is
None of this needs a better relationship with suppliers. It needs a clearer one.
The programs that get value from co-development didn't add more workshops. They decided, before the concept froze, which decisions they were willing to stop making themselves — and then wrote down what that transfer was worth, who carried the risk, and how they'd get out. That's a sourcing decision. It belongs on procurement's desk, not in a design review.
Co-development is not about inviting suppliers in earlier. It's about deciding, in writing, which design decisions you will no longer make.
Everything else is a meeting.