How Experienced Buyers Structure Custom Battery Pack Payment Terms

Payment terms on a custom battery pack program belong in two buckets, not one: one-time money for NRE, tooling, and qualification support released against engineering gates, and per-unit production money released against shipment and acceptance. The gate is the control point. Tie every dollar to something you can inspect — a prototype that passed, a pilot lot that ran, a first-article package that closed — and the program stays funded without paying ahead of proof. The instruments, the net-day conventions, and the holdback percentage are all downstream of getting that split right.

The column I add to every quote comparison is the one nobody sends me: cash out the door, by month. Three suppliers quoting the same 48V pack can look nearly identical across unit price, MOQ, lead time, and tooling, and then diverge completely in that last column. That divergence is the actual negotiation.

The terms page is the tell

I was sitting with three quote PDFs open on a Monday, all for the same pack, and the terms pages were the only real difference. Supplier A wanted 50% on PO, 50% before shipment. Supplier B wanted 30% on PO, 30% at prototype approval, 20% at pilot, 20% at mass production release, and then net 45 on units after that. A's tooling line was a little lower. The per-unit number was a rounding error apart. On a comparison sheet with five columns, A wins.

On my sheet, B wins, and it isn't close.

With B, the largest amount I had exposed at any moment before a build I could personally inspect was less than a third of the one-time money. The tooling was being cut while my prototype payment was still sitting in my own account. If the design had come back from the DV bench with a thermal problem — and one program did exactly that — I would have been renegotiating a change order from a position where I still held most of the money. Ask me what that's worth and I'll tell you it's worth more than a slightly lower tooling number, because I've been on the other side of that conversation and it is a bad room to be in.

Put gates in the schedule, not dates

"Thirty days after PO" is a date. "On written prototype approval" is a gate. Dates move when the cell shipment slips, and once a date moves everybody argues about whether it moved for a good reason. A gate doesn't move. It either passed or it didn't, and I can look at the report.

The one-time money usually wants to break into four pieces:

• A deposit at order acceptance. Small enough that you can lose it without a committee meeting, big enough to let the supplier commit to long-lead cells and place the first tooling PO.
• A payment at prototype approval, tied to a delivered build and a completed test report you've reviewed.
• A payment at pilot run, tied to parts built on production tooling and process, not on the bench.
• A payment at mass production release, tied to whatever your final acceptance package is — a first-article approval, a PPAP-style submission, a signed-off qualification file.

Then the per-unit money runs on net terms against shipment or acceptance.

The mistake I see most is paying the entire NRE up front because the supplier asked for it and the number looked small next to the total program value. It isn't small to the program. It is the only leverage you have during the phase where engineering attention is scarcest. Once that money is gone, your change requests go into the same queue as everybody else's.

Watch the supplier's cash cycle, not just your own

Here's the thing that took me a couple of programs to internalize. A pack supplier's cash goes out before yours does. The tooling shop wants a deposit. The cells are on an eight-to-sixteen-week lead time and often want payment or a letter of credit at order. Connectors, BMS boards, busbars, the enclosure — all of that gets purchased before a single unit ships to you.

So when a buyer pushes for 100% on delivery, the supplier has three options: refuse, raise the unit price to cover the financing, or fund your program off someone else's money. The third one is how you end up as the last customer to get cells when allocations tighten. I would rather pay a deposit that buys a real, traceable deliverable than win a terms fight and lose the allocation.

Ask for their cash cycle. A supplier who can show you a straight answer — when tooling gets paid, when cells get paid, when they need money in — is a supplier who has thought about this before. The one who just repeats the deposit percentage back to you hasn't.

Wire, letter of credit, documentary collection

Three instruments cover almost everything in this category, and the choice should track the size of your exposure, not your habit.

A wire transfer is cheap, fast, and effectively irreversible. That's fine for a deposit against a supplier you've audited and shipped with before. It's a poor fit for a large one-time payment to a supplier you've never seen run a line.

A letter of credit puts a bank between you and the money, with documents as the trigger. It costs real money in fees and it is unforgiving about paperwork — a misspelled consignee on a bill of lading can hold up payment. But on a big tooling payment or a first production shipment from a new source, that rigidity is the point. It converts "I trust them" into "the documents have to match."

Documentary collection sits in between. Documents move through the banks and payment is made against them, which is cheaper than an LC and gives you more control than a bare wire, but no bank is guaranteeing payment. I use it when the relationship is established but the shipment values are getting large enough that I don't want to be wiring blind.

Net 30, net 60, net 90 — read the trigger, not the number

Net terms describe the number of days after the invoice date that payment is due. The number is the part everybody quotes in the meeting and the trigger is the part that actually costs money.

Invoice date, bill of lading date, arrival date, receipt date, and acceptance date are five different starting lines, and on an ocean shipment they can be months apart. Net 60 from invoice date sounds generous. Net 60 from invoice date when the invoice is issued at origin and the container takes eleven weeks to reach you is something closer to net 140 from the day the goods touched your dock — which is fine, if that's what you agreed to, and expensive if you modeled it as net 60 in your working capital plan.

So I read the trigger sentence twice, and I put the actual cash date in the last column, not the payment term.

Retention and holdback

Retention, or holdback, is withholding a percentage of payment until acceptance or until the end of a warranty period. Whether and how you can enforce that is squarely counsel's territory and I hand it over — I don't draft it, I don't interpret it, and I don't rely on it meaning more than the contract says.

Commercially, though, holdback has a price. Suppliers know they're waiting on that money and they price the wait. A 5% holdback out two years is not free money; it's a small financing cost you've embedded in every unit. I tend to use holdback on tooling and the first production lots, where the risk is concentrated and the exposure is short, and I'd rather shorten the holdback and lengthen the warranty than ask for both at full strength. That trade is usually available and most buyers never ask for it.

Where I've watched it go wrong

A deposit paid on a program that got canceled. The tooling was real, the cells were ordered, and the deposit did not come back. Size every deposit to the amount you're willing to write off, because that's what it is.

Payment tied to "shipment," which the supplier controls and which doesn't require anything to work. Tie it to something you receive or something that passed a test.

Milestones tied to a document instead of a result. A test report is not a passing test report.

Full NRE up front, followed by a slow quarter at the supplier's engineering group, because the revenue was already booked.

And the one that stings: hard negotiation on net terms, unit price quietly up four percent, and nobody looked at the last column. The terms improved. The program got more expensive.

The whole point

Payment terms on a custom pack are a program-risk tool that happens to live on a contract page. They decide who is exposed, when, and to how much, across a program that runs from a tooling drawing to a steady production cadence. Get the two tracks right, hang every one-time payment on a gate you can inspect, and match the instrument to the size of the bet — and you'll fund the program all the way through without ever paying in full for work that hasn't been proven yet.

The drafting is for counsel. The structure is yours.