Manufacturers have been talking about developing, manufacturing and delivering customer-centric products since the beginning of the Era of Manufacturing. With time, customer expectations have evolved and manufacturers have been left with no choice but to keep abreast of their facilities to address customers. However, recently, there has been a shift in customer mindset, from buying a customizable product to subscribing to a personalized service.
Customer Mind-set change
This shift leaves manufacturers with little option but to create products to meet the different needs of each consumer in the target segment. Focus on Hyper-personalization or ‘Lot size 1’ is increasing the complexity on the supply chain networks. This has put tremendous stress on developing advanced planning, scheduling systems to administer raw materials in real time to rapidly changing manufacturing priorities. Any inertia in time to market can significantly affect orders and lead to unprecedented losses.
Identifying the challenge with Status Quo
In today’s world, technologies such as AI & ML are bearing the load of such advanced planning systems and making them intelligent. With connected machines and integrated processes, AI & ML powered solutions will need more computing power to handle data with higher velocities and variability to generate real-time outputs.
Large Manufacturers govern a lattice of warehouses, logistics hubs and distribution centres spread out across the globe. In addition, they have a panel of sub-manufacturers, which bring in parts and components. These network nodes are not static in nature and change frequently because of new markets, acquisitions or partnerships, changing supply cluster or customer segments. The versatility of such networks brings in the complexity in their management.
The current inventory production and supply chain planning solutions have started to fall short because
a. Schedulers have to change the optimization schedule too frequently in real time and perform dynamic inventory allocation and production planning. Each modified, added or deleted node changes the complexity and the planning systems take longer to generate results, sometimes even a day – to put things into perspective, consider an hour’s delay for a plant producing $ 1 Mn worth products in an 8-hour shift. This would hit the throughput significantly.
b. Planners often have to run multiple “if-this-then-what” simulations to ensure that production does not stop. These simulations help with a best-fit plan to accommodate a change in order, or fulfilling a pending backlog for a priority customer or sometimes even to build a prototype to support a marketing request. The complex network simulations usually take more than an hour to churn out a plan. With decreasing cushion time between 2 job stations on a shop floor, an hour’s wait will result in nothing but losses.
Imagine a situation where a business has to ship products through 5 trucks on 20 possible routes. This means the planners have to choose the best from 205 or 32,00,000 options. Such problems can be solved with the current processing hardware in no time but if we increase the trucks to 100 (common case for a manufacturer) the possibilities become humungous and to choose the optimized route, the planners also have to consider data like traffic, weather, vendor capacity, production plans, etc. Clearly, a classical computer will take time to handle such simulations.
The Hero: Quantum Computing
Quantum computing guarantees to cater to such challenges that manufacturers will soon face. This is because Quantum computers can solve complex problems with extremely high speeds. The main idea of quantum computing is using Qubits. Qubits, unlike normal bits, can possess 0,1, or both states simultaneously.
Let us understand how quantum computing can help. Take 2 complex numbers – x & y. The property of being able to exist in multiple states is called superposition. Quantum physics doesn’t allow us to measure the value of two numbers, instead when we measure a qubit we get the state 0 with a probability |x|2and 1 with a probability of |y|2 . The sum of both the probabilities should be 1. So when a quantum operation is performed on a qubit, it must be performed on all the states simultaneously. 
Simply put, a qubit x(1) +y(0) can hold multiple states between 0 & 1, but when measured it will be either 1 or 0. This means a single operation can be carried out on 2n values simultaneously.
How it helps
Now, what does this mean for our optimization problem at hand? In classical computing, these problems have to be solved through calculating permutations and comparing them one at a time. Because of qubits and superposition property, quantum computing will apply the given operation to all possible states represented by a qubit. Instead of zillions of operations, one at a time quantum computing can reduce the number of iterations, thereby significantly reducing time to find the best-fit plan quicker than classical computing.
 Michael A. Nielsen and Isaac L. Chuang, Quantum Computation and Quantum Information: 10th Anniversary Edition, Cambridge University Press; 10th Anniversary ed. Edition