Two businesses can need their largest production runs at different times. Separately, each must provide enough capacity for its own busiest period. Together, they may be able to serve both workloads with less equipment than the sum of those separate requirements. The opportunity comes from the timing of work, not from making the work disappear. If both peaks move into the same period, the apparent saving can disappear too.
On 16 April 2026, Interfax reported, citing the Financial Times, that Nissan and Chery had discussed producing Chery vehicles at Nissan's Sunderland plant in the United Kingdom. The report described talks, not a completed production agreement.
That announcement raises a broader question about sharing productive resources. When does bringing two workloads together actually reduce the capacity needed to complete them? A spare building, an unoccupied line and transferable production time are different propositions. The arithmetic below isolates the last of these. It does not describe the Sunderland factory, either manufacturer's orders, or the technical feasibility of putting their products on a common line.
Start with a calendar, not an annual average
Consider two independent customers, A and B, using an imaginary processing service over two equally long periods. A requires 80 units of processing in the first period and 20 in the second. B requires 20 in the first and 80 in the second. All numbers in this example are invented to make the timing relationship visible. They are neither a manufacturing benchmark nor estimates for the companies in the opening report.
Assume that a unit of processing means the same amount of the same resource for both customers. Work must be completed in its specified period. There is no early production for inventory, no backlog carried forward, no downtime and no changeover loss in the initial calculation. Equipment can move freely between the two workloads. These assumptions are deliberately strong: they identify the exact conditions under which the simple comparison works.
Each customer brings 100 units of work across the two periods. Knowing only those totals would not reveal the capacity opportunity. The planner also needs to know when the work must happen.
Separate peaks add up to 160
If A has a dedicated resource, it needs capacity for 80 units per period. Providing only its average requirement of 50 would leave 30 units unfinished in the first period. The lighter second period cannot repair that missed requirement under the stated deadline rule. A therefore needs the ability to process 80 even though that ability is not fully used in both periods.
B has exactly the same standalone requirement. Its peak also reaches 80, but in the second period. Two separate resources consequently provide a combined installed capacity of 160 units per period. Across the two-period horizon, they could process 320 units if work were available and suitable. Actual required work is only 200.
The unused difference is not additional output waiting to be sold. It is unused processing capability at particular times. Whether another customer can use it depends on that customer's deadlines and compatibility. Calling the entire difference spare capacity without a calendar would conceal the central issue.
The shared peak is only 100
Now combine the workloads period by period. In the first period, 80 plus 20 requires 100. In the second, 20 plus 80 also requires 100. A fully interchangeable common resource with capacity of 100 per period can complete every specified unit on time. The total work remains 200, and neither customer receives less service.
| Workload | Period 1 | Period 2 | Peak |
|---|---|---|---|
| A | 80 | 20 | 80 |
| B | 20 | 80 | 80 |
| Combined | 100 | 100 | 100 |
The difference between 160 separately provided units and 100 jointly provided units is 60. Relative to the separate requirement, that is a reduction of 37.5%. It is not a 37.5% improvement in processing speed. Every unit still takes the same resource effort. Nor does it establish an equivalent percentage reduction in costs: buildings, integration and staffing have not yet entered the example.
Keep the work, change the alignment
To see what creates the difference, leave A unchanged and move B's busy period. B now needs 80 in the first period and 20 in the second. Each customer still requires 100 units overall, and each still has an individual peak of 80. The combined workload becomes 160 followed by 40. A common resource must now supply 160 per period if every deadline is to be met.
Nothing in the annual or horizon-wide totals distinguishes this case from the earlier one. The sum of work is still 200. Even a report listing each customer's peak would show the same figures. The missing information is the coincidence of those peaks. A capacity-sharing proposal based only on average utilisation can therefore look unchanged while its physical requirement moves substantially.
The first arrangement is complementary in time; the second is not. That is a property of the specified schedules, not a permanent characteristic of either customer's business.
A useful bound on the possible saving
For any period, A's workload cannot exceed A's own maximum, and B's workload cannot exceed B's own maximum. Adding those statements shows that the largest combined workload cannot be greater than the sum of the two individual maxima. Sharing cannot require more productive capacity than that sum in this simplified model, before any extra costs or losses associated with sharing are introduced.
But the shared resource must also be at least as large as the bigger individual peak. It must be capable of serving A when A is busiest, even if B contributes nothing then, and the same is true for B. With two individual peaks of 80, the shared requirement lies somewhere between 80 and 160.
The lower boundary would be reached by fully separated schedules such as A requiring 80 then zero and B requiring zero then 80. Our original example needs 100 because each customer's quieter period still contains 20 units. Complementarity reduces overlap; it does not mean overlap has vanished.
Unequal customers leave a different opportunity
The two customers need not be the same size. Keep A at 80 and 20, but let B require 10 and 40. Separate peak capacity is now 120, while the combined schedule requires 90 followed by 60. The shared peak is 90, giving a capacity difference of 30, or 25% of the separate requirement. The smaller customer still contributes useful work in A's quieter period, but the result is not the same as combining two equally sized profiles.
Even perfect separation could not reduce the shared requirement below A's peak of 80. Against the separate total of 120, the maximum possible difference is therefore 40. This boundary helps prevent a percentage from one partnership becoming a target for another. Adding a small customer to a much larger operation cannot remove the larger customer's own peak requirement. The achievable difference depends on both the relative sizes and the timing of the workloads, and neither can be recovered from a count of participating firms.
A longer horizon does not change the method
Two periods are enough to demonstrate the mechanism, but a longer calendar uses the same comparison. Suppose A requires 80, 20, 60 and 40 units across four periods, while B requires 20, 80, 40 and 60. Each individual maximum remains 80. The combined requirement is 100 in every period, so the shared peak remains 100.
Adding more dates does not mean averaging away a busy one. The planner must still inspect every period that the service promise treats as separate. If a weekly total hides a deadline on Tuesday, a weekly calculation may be too coarse. Conversely, splitting the calendar into artificial intervals that do not match any operating commitment can create a misleading impression of rigidity.
The time unit should follow the obligation being modelled. Changing that unit is a change in the model, not a harmless presentation choice. A result calculated for one interval should not quietly be advertised as a guarantee for another.
Co-location is not interchangeable capacity
The calculation assumes that unused processing capability assigned to A can serve B in the same period. Placing two dedicated lines in one building does not establish that ability. If each line remains restricted to one workload, the first line's quiet period still cannot relieve the second line's busy period. Shared ownership, a shared address and a shared productive resource are not equivalent.
This distinction does not make co-location worthless. Participants might share services or avoid duplicating premises. Those would be separate benefits requiring their own evidence. They cannot be counted as the 60-unit reduction demonstrated here unless the productive resource itself can be allocated across the workloads.
A commercial proposal can make its claim precise by naming the resource that becomes transferable. Without that boundary, different teams may attach the word shared to different things and calculate benefits that cannot all arise from the same arrangement.
Changeovers consume some of the opportunity
Relax the initial assumption of lossless switching. Suppose operating the shared resource consumes 10 units of its available processing capability in each period because of an agreed changeover allowance. The original workloads still require 100 productive units in each period. Physical capacity of 100 would now leave only 90 for customer work, creating a shortfall of 10.
Capacity of 110 restores the required 100 productive units after the allowance. The comparison with 160 separately provided units now shows a difference of 50, not 60. This remains a capacity comparison, not a financial saving. The allowance is an invented resource charge, not an estimate of an automotive changeover.
The accounting boundary matters. The example charges 10 once per period to the shared arrangement. Charging it once for every product, every batch or every switch would be a different assumption and could produce a different answer. An attractive total should never depend on leaving the frequency of a loss unstated.
A favourable schedule is not an unlimited guarantee
The original calculation starts with known workloads. It proves that those workloads fit within 100 shared units. It does not prove that any future combination of requests will fit. If both customers may independently request up to 80 in either period, simultaneous requests of 80 and 80 remain possible. A guarantee covering every such combination would require 160 under the same service assumptions.
This difference is easy to miss when a past calendar becomes a sales promise. Observed non-coincidence is evidence about what occurred. A contractual commitment concerns what the provider must accommodate. The commitment might preserve a fixed schedule, allow changes within limits or offer only availability when another customer releases it. Those products have different capacity implications.
No probability is needed to establish the distinction. Without assigning odds to a busy period, the planner can already identify which combinations the proposed resource can and cannot serve.
Reservations can recreate the separate peaks
Imagine that the common resource has capacity of 100, but each customer receives an unconditional reservation of 70 in the same period. The two reservations total 140. They cannot both be exercised in full at once. The fact that recent actual workloads were complementary does not make the obligations physically compatible.
An arrangement can avoid that contradiction by defining which part of access is guaranteed and which depends on released capacity. Another possibility is to specify the accepted calendar in advance. The important point is not to prescribe one commercial format. It is to prevent a provider from selling overlapping rights while evaluating the asset against a non-overlapping schedule.
Such a mismatch can exist even when neither customer behaves unusually. Each may simply exercise the access it believes it purchased. The disagreement then comes from the design of the promise, not from an unexpected failure of the arithmetic.
The physical gain does not allocate itself
If an arrangement genuinely reduces required capacity, a separate negotiation remains over who receives the benefit. The resource owner may fund adaptation, one customer may accept tighter scheduling, and another may supply the quieter-period workload that makes the combination useful. The arithmetic identifies a joint opportunity but does not determine a uniquely fair price.
Charging solely by completed units could make sense for one arrangement, while charging for reserved access could suit another. The two approaches describe different services. A customer buying guaranteed busy-period access should not be confused with a customer buying an otherwise unused window, even when both eventually receive the same number of processing units.
Similarly, a reduction in installed capacity need not lower every participant's expenditure. Integration costs, agreed compensation and the division of responsibilities may absorb part of the benefit. Those terms belong in a financial comparison after the physical requirement has been established, rather than being assumed from a utilisation percentage.
Ask which difference the proposal actually makes
A concise assessment can retain the essential distinctions without turning into a general factory audit:
- Compare workloads within the same service periods, not only their totals.
- Calculate each separate peak and the combined peak.
- Identify the resource that can genuinely move between workloads.
- Deduct the specified switching allowance before testing whether work fits.
- Check promised access against combinations customers are entitled to request.
- Keep the resulting capacity difference separate from its financial allocation.
The central result is modest but useful. Two customers can each require a peak of 80 while jointly needing only 100, provided their schedules and the resource's flexibility support that combination. Move their peaks together and the joint requirement returns to 160, without changing total work. Shared production creates a capacity opportunity when it changes the peak that must be served together. The name of the partnership, by itself, cannot do that.