When Battery NRE Amortization Costs More Than Paying Upfront

NRE amortization costs more than paying upfront whenever the volume that carries the per-unit recovery does not arrive on the schedule the contract assumes. An upfront NRE payment is a fixed, knowable number on the day you sign. An amortized one is a wager on your own forecast that converts into a true-up invoice at the exact moment your program is already behind. Pay upfront when you have the cash and the volume is genuinely uncertain; amortize only when you can cap total recovery, tie the schedule to accepted units rather than a calendar, and put in writing what happens to the remaining balance when the program changes shape.

I keep a folder on my desktop called amendments, and in it is a reconciliation sheet I have opened more times than I want to admit. Column D is units shipped to date. Column E is units required for full recovery. Column F is the balance. Sit next to me while I scroll right, because the first time this happened I did not scroll right. I looked at the per-unit adder, did the quick multiplication in my head, and decided we were fine.

The call came on a Tuesday afternoon. The supplier's finance contact was pleasant and unhurried, and he read the clause out loud to me — section 4.3(c), the shortfall remedy — while I pulled up the sheet. I could hear a keyboard on his end. I asked him to repeat the number, not because I had not heard it, but because I wanted to write it down twice. The printout of the agreement was already on my desk with a coffee ring on the corner, and I remember finding the clause and running my finger along the line, as if touching the paper would change what it said.

What it said was that the schedule assumed a minimum. We had not hit the minimum. The difference was now due in thirty days.

What the schedule is actually promising

An NRE number covers the design work, the tooling, the test fixtures, the first-article validation, the documentation package. Somebody pays for it. The only real question is when, and who carries the risk if the program does not go the way the forecast said.

The three structures are not complicated to describe. You can pay it upfront as a separate line item, out of capital, before a single pack ships. You can amortize it into the piece price, so the tooling and the engineering ride along with every unit for the first however-many-thousand units. Or you split it — a partial upfront payment to get the supplier moving, the remainder recovered per unit.

What is complicated is that these three options are not compared on the same axis. Upfront NRE hits a capital budget and a depreciation schedule. Amortized NRE hits the piece price and your standard cost and the number your operations team uses to decide whether the robot program is viable. Finance looks at one, manufacturing looks at the other, and the person who signed the agreement is the only one looking at both.

I have signed all three. I will tell you what actually decides it for me: not the discount rate, not the cost of capital model. It is how confident I am in the volume, and whether I can afford to be wrong.

The comparison that quietly goes bad

Here is the thing about amortization that the straightforward explanation leaves out. It does not just move money around. It destroys your ability to compare suppliers.

Take a hypothetical, round numbers, not a quote from any real program. Suppose the NRE is $150,000 and the recovery is spread over 15,000 units. That is ten dollars a unit. Supplier A quotes you a piece price with that ten dollars baked in. Supplier B, who wants to win the business, quotes an upfront NRE and a leaner piece price. On the spreadsheet, B's piece price looks better and A's upfront number looks worse. But you are not comparing two prices. You are comparing a price at 15,000 units against a price at an unknown volume, and the second one is not a number at all. It is a conditional statement with the condition hidden in column F.

The damage does not stop at the original decision. Once you are amortizing, your incumbent's piece price contains a recovery component that no other supplier can see. When you go out for a second source — and you will, because single-source pack supply into a fleet is a risk you carry on your own balance sheet — every alternative quote comes in lower by roughly the amortization amount. You cannot tell whether the second source is genuinely cheaper or whether you are looking at the same cost with the NRE stripped out and moved somewhere else. I have watched a sourcing team get excited about a twenty percent gap that was almost entirely an artifact of where the NRE sat.

There is a way out of this, and it is not clever. Ask every supplier for two numbers: the full piece price with recovery included, and the piece price after recovery is complete. Then ask what volume the recovery assumes. If a supplier cannot answer the second question in writing, you are not looking at a price. You are looking at a sales position.

The counting rules nobody reads until it matters

The clause that got me on that Tuesday was not unusual. It was standard, and that is exactly why it caught me. Standard clauses are written to be skimmed.

Which units count

Amortization schedules usually define a unit in a way that sounds obvious and is not. Shipped. Invoiced. Accepted. Paid. Four different words, four different dates, and at the end of a quarter the gap between them can be a month or more of production.

Then the counting rules get narrower. Warranty replacement packs — does a unit that ships to replace a field failure count toward recovery? Sometimes not, because it was never invoiced. Consignment stock sitting in your warehouse — does it count when it ships from the supplier or when you consume it? Engineering spares you buy to sit on a shelf? Buffer inventory built ahead of a line launch?

That last one matters more than people expect. If your manufacturing plan includes a buffer of packs to cover a potential supplier interruption, those buffer units are real money leaving your account. Whether they count toward NRE recovery is decided by one sentence in the definitions section, and I have seen that sentence go both ways. A buffer that does not count is a buffer you paid twice for.

The step-down that never happens

The other one is the tail. The schedule says recovery completes at 15,000 units. What does the piece price do at unit 15,001?

If the agreement does not say, the answer is nothing. The price stays where it is. Nobody sends you a note. The recovery line item just quietly becomes margin, and it stays margin for the rest of the program, and you find out about it two years later when someone new runs a should-cost and asks why the adder never came off.

I write the step-down into the agreement now as a specific, dated, dollar-per-unit reduction, with the trigger defined by the same counting rule that governs recovery. If the supplier resists naming a number, that tells you something about how they were planning to handle the tail.

Reconciling the commitment with the forecast you actually believe

The volume commitment attached to an amortization schedule is usually built from the demand forecast your own team produced during the sourcing phase, which is the most optimistic document your company will ever generate.

I do not sign a commitment built from that number. I build the commitment from the forecast I would defend in a room full of people who are allowed to ask hard questions — which usually means taking the ramp, pushing it out, flattening it, and then checking whether the number still works if the program slips a quarter or the customer cuts the order for the first configuration.

A few things I check every time now.

• Is the minimum a program total or an annual figure? A per-year minimum with a catch-up true-up on each anniversary can hit you more than once for the same shortfall, and the second hit lands during the recovery year when cash is tightest.
• Does the schedule follow actual receipts or the calendar? A calendar-based recovery keeps accruing while your line is idle waiting for first articles.
• What happens if a revision changes the pack? An engineering change that alters the cell, the enclosure, or the BMS board can restart NRE or reset the amortization base. Get the treatment of change orders stated before the first one happens, because the first one always happens.
• Does buffer stock count, and on what date?

When the program is cancelled or the supplier changes

The clause that matters most is the one nobody negotiates because it feels like negotiating for failure. What happens to the unpaid balance?

Many schedules accelerate it. Cancel the program and the entire remaining NRE becomes due, sometimes immediately, sometimes on a schedule that is shorter than any wind-down your finance team has planned for. That is a real exposure and it belongs in the risk register next to the tooling, not buried in an exhibit.

Supplier change is the stranger case. You leave, or they exit the program, and a balance is still outstanding. Does it survive? Is it netted against final invoices? Does the tooling transfer, and does the design package and the test fixture set transfer with it, and who pays for the fixture rework at the new supplier? I have seen a program where the recovery balance was small, the tooling transfer was straightforward, and the fixture set was not documented anywhere — which cost more to recreate than the balance was worth.

The point is not that you should expect a cancellation. The point is that the unpaid balance is a liability that lives on your books whether or not anyone at your company remembers it exists, and it gets discovered at the worst time by the person who least wants to hear about it: you.

How I structure it now

I still use amortization. It is the right instrument when the capital is genuinely constrained or when the program is large enough and long enough that the recovery is a rounding error against the volume. Paying a hundred fifty thousand dollars upfront for a program that will run for six years is a poor use of cash, and pretending otherwise is its own kind of mistake.

But the structure I ask for has changed. Split the NRE, with a meaningful upfront portion — enough that the supplier has skin in the tooling and I have a clean number to defend. Amortize the remainder against accepted units, not calendar periods. Cap the total recovery so the per-unit adder cannot run past a stated ceiling no matter how the schedule is read. Tie the step-down to the same counting rule. State the treatment of buffer stock, warranty replacements, and change orders in the definitions. And write the cancellation and transfer provisions in plain language I can read out loud to a controller without flinching.

That is more negotiation than the upfront option requires. It is also less than the cost of the phone call I took on a Tuesday afternoon, listening to someone read a clause I had already signed.