The high tech industry is riding the disruption wave to support ever-changing customer expectations. With modern customers demanding products tailored to their individual preferences, legacy systems are being migrated to faster and smarter systems that provide real-time insights to enable mass customization. But the focus on hyper-personalization or the ‘segment of one’ is increasing the complexity in the supply chain network. This creates significant strain on planning systems to allocate products and materials in real time to rapidly changing priorities, as latency could result in order cancellations.
Digital supply chains underpinned by artificial intelligence (AI) and machine learning (ML) powered cognitive assistants have emerged as the solution to the challenge. However, these intelligent assistants need higher computing power to handle enormous volumes of data coming in at high velocities and with a high degree of variability, in order to generate real-time signals that aid supply chain planners in decision making.
Why current supply chain optimization engines fall short
Large high tech and electronics OEMs have multiple warehouses and distribution centers managed by the companies themselves or by third-party logistics providers across geographies. Then there are contract manufacturers who either directly ship to OEMs’ distribution centers or to dealers or channel partners. Frequent mergers and acquisitions (M&As) increase this network complexity by adding more company-owned and third-party managed facilities. The complexity in dynamic inventory allocation to these locations exponentially increases with each distribution center added to the network.
Current optimization systems used in inventory allocation and order promising tend to lag supply chain planners’ expectations because:
- Planners execute optimization engine very frequently to promise orders in real-time and perform dynamic inventory optimization. As the supply chain network grows in complexity, the optimization engine takes longer to complete the plans, thereby delaying system availability to users.
- To create ‘what-if’ scenarios, supply chain planners perform quick simulations by changing the planning parameters. These simulations are typically required to check the feasibility of fulfilling a rush order from a high priority customer, evaluate preparedness to support a sales promotion, or support sales and operations (S&OP) meetings. In complex supply chain networks, simulations take longer to complete as well. It may take even days to provide a best-fit plan. This dilutes the very purpose of finding a quick solution using simulation.
Clearly, inadequate system performance capabilities lie at the heart of these extended planning runs. Let’s say a company plans to ship orders using 10 trucks over three possible routes. This means the company has 310 possibilities or 59,049 solutions to choose from. Any classical computer can solve this problem with little effort. Now let’s assume a situation where a transport planner wants to simulate shipments using 40 trucks over the same three routes. The possibilities in this case are approximately 12 Quintillion – a tough ask for a classical computer to handle. Add to it additional factors such as traffic and weather and the challenge of simulating different possibilities becomes even more daunting.
Quantum computing in supply chain: A game changer
Quantum computing promises to address complex supply chain optimization challenges that modern businesses face today. This is because quantum bits (qubits) unlike their on/off binary cousins, can occupy more than one state at the same time, embracing nuance and complexity - a property widely known as superposition. Their second noteworthy property i.e. entanglement makes these particles interdependent on each other and analogous to the variables of complex supply chain. According to Christoph Becher, a Professor in Experimental Physics at Saarland University, “It is possible to adjust an interaction between these qubits so that they can ‘sense’ each other. The system then naturally tries to arrange itself in such a way that it consumes as little energy as possible.” This minimized energy expenditure is exactly what organizations need to create more driverless supply chains.
Next generation DSCs leverage ML which is based on sampling and optimization methods. Sampling technology in quantum computers can provide more distributed, reliable input data for the ML algorithms. An adiabatic quantum computer (AQC), one that can write its own optimization problem can prove to be very useful in supply chain scenarios as machine learning challenges are optimization based.
Positioning digital supply chains for a quantum leap
Quantum computing has the potential to disrupt the way advanced planning and optimization systems work. It can enable planners to run plans at the flick of a button and perform scenario planning on the fly. While this is certainly a boon, use of quantum computing in supply chain could also prove to be very expensive and therefore impractical. The best way forward for organizations would be to perform optimization outside the advanced planning system (APS) using quantum computing and then feed the results back to APS for plan review. Availing as-a-service quantum computing power can also help enterprises cut back on costs. Enterprises that collaborate with partners making advancements in quantum computing to build pilot applications in supply chain planning and network optimization will get ahead in the predictive personalization race, delivering exceptional customer experiences.