The semiconductor industry is characterized by its global supply chains
These are a complex network, and achieving optimal results is a challenge, even for established industry players like Infineon. With a leading position to maintain, Infineon has its sights set on one of the key technologies for the future: quantum computing.
Global supply chains and their challenges
The semiconductor industry is based on global supply chains. This is for both supply and demand-side reasons: on the one hand, the sector serves a broad spectrum of customers worldwide, with oftentimes fluctuating demand for a wide variety of products. On the other hand, companies make use of a wide range of production sites globally – they have to, as semiconductor manufacturing is lengthy and complex. Hans Ehm, Senior Principal Supply Chain Management at Infineon, compares making semiconductors to baking a layered cake: “The entire process is strictly sequential. Individual steps cannot be done in parallel, but can only take place one after another, layer after layer, but not always in the same physical place. With some complex semiconductors, it can take half a year or more to complete a component, with over 1000 process steps. And that is despite the fact that production is running at full speed 24 hours a day, 365 days a year, using a system for ‘complex flow production’ that optimizes speed.” Semiconductor producers around the world all currently face the same challenge: chips are in short supply, while digitalization and decarbonization continue to fuel demand. This is another reason why Ehm sees a key competitive advantage in the shape of supply chain optimization.
He sees a particular challenge in how to determine an optimal “available-to-promise” (ATP) within this modern, complex supply chain. Available-to-promise refers to the inventory due to arrive at the sales warehouse at a certain point in time – in other words, the inventory for which the company can make commitments to its customers.
Demand-capacity matching and the “knapsack problem”
In the face of fluctuating demand, Infineon and its peers need to predict future inventory on a daily basis. Their response is to have alternative production lines which are optimized via a daily “demand-capacity matching” process, pairing demand and available capacities for the optimum inventory. With the ATP calculated this way, more than a million order items can be confirmed on a daily basis. The hitch: it has so far been impossible with classical computer technology to solve the underlying optimization problem as a whole. Many companies instead make do with a number of different heuristics and solvers that break the task down into subproblems and derive approximate solutions to these. For Hans Ehm, on the other hand, quantum computing offers a good chance of achieving a holistic solution to the problem. It heralds new hopes of jointly achieving solutions to tasks such as daily demand-capacity matching optimizations, resulting in better ATP and order confirmations.
“Challenges like the matching problem in global supply chains can also be understood as variants of the ‘knapsack problem’,” explains Lilly Palackal. Given her role as Infineon’s Quantum Algorithms Team Lead in Supply Chain Innovation, Palackal is also active on the issue in QUTAC. The classic ‘knapsack problem’, as she explains, is as follows: a person wants to pack their knapsack (a type of backpack) in the best way possible. They can choose from a range of items of different weights and utility values. How do they go about it and pack the combination with the highest utility value, assuming they cannot exceed the knapsack’s maximum capacity? “Demand-capacity matching is about determining the optimal ATP from about a million orders that are confirmed every day,” Palackal says. “This is obviously much more complicated than packing a backpack, but the underlying problems are very similar.” With their ability to make use of special quantum mechanical effects such as superposition and entanglement, she points out, quantum computers are able to apply specialized algorithms that are tuned for solving certain problems faster. The race is now on to harness quantum algorithms in order to solve optimization problems.
Sharing knowledge for tomorrow’s quantum computing ecosystem
It’s hard to predict exactly when the technology for demand-capacity matching will be commercially ready. Nevertheless, it is important to start developing solutions today, explain Ehm and Palackal. “Not out of concern about keeping pace,” Ehm emphasizes, “but in order to one day be a leader in this technology of the future.” The long development pipelines that quantum computing involves do not faze him. “Every time a technological innovation happened, scientists and engineers first had to spend a long time laying the intellectual groundwork for practical applications, before these ultimately found their way into the economy and commercial use became a reality.”
At the current stage, Ehm agrees that a consortium like QUTAC is the ideal vehicle. “In the semiconductor industry, it’s common to form consortia in pre-competitive phases to jointly search for solutions,” he says, explaining that it’s natural for Infineon to take the same approach to quantum computing. The company is active in the QUTAC working group on production and logistics, where cooperation has now taken on very concrete forms, as Palackal notes. “Our demand-capacity matching is one of several use cases the group is working on in a focused way. This is an area where we have gained a lot from interactions with colleagues from other companies.” QUTAC explicitly aims to share the knowledge it generates widely; as Ehm puts it, “when we share our knowledge, it not only takes us, the members, forward, but also the entire quantum computing ecosystem. This helps us develop solutions that can be of benefit to many companies and organizations across industries.”