The Age of Autonomous Supply Chains

We associate the concept of central planning with grim images of collapsing Soviet economies, but it turns out that central planning is also the main engine for how capitalist modern economies run – at least inside the enterprise, though with perhaps a slightly better track record.

It turns out, whether it’s at the national level or inside the enterprise, that people do not like being told what to do or being guided by central planning. Despite that, since the 1960s, companies have been trying to make computerized top-down planning and scheduling work so that they can operate their business networks more efficiently. 

These central planning systems are costly and complex, and with the rise of blockchain-based industrial operations, there is an opportunity to put many of them out to pasture. Just like the real world, blockchains are decentralized systems where individual actors participate alongside others, but there is no centralized coordinating entity.

In the place of top-down inventory planning and forecasting, smart contracts could be used to allow decentralized operations. Stores or locations that run out of inventory can then look around for potential suppliers – checking on the cost of buying replacement inventory from nearby stores, distributors or directly from the factory. Each local participant needs to focus only on setting their own rules to manage their replenishment plans and forecast.

Crowdsourced services like ride sharing already do something like that. None of them directly tells drivers when to drive (or riders when to ride). Instead, they allow some form of supply and demand matching to occur directly in the market. Current systems are still highly centralized, with a lot of analytics and tools being used to nudge drivers and riders toward an equilibrium with things like bonuses and surge pricing. It is less hands-on than centralized planning, but still far from an entirely free market.

These types of systems are still much more responsive than central planning systems, although that’s a low bar to hurdle. Planning has always been tough, and the gradual evolution of the process has taken decades. In the process, you take the material plan for a product and then backwards calculate how many should be ordered based on your production schedule. The result is, in theory, an orderly sequence of purchases that will refill your stocks in a timely manner. Once products are built, a separate plan, often known as the distribution plan, pushes them out into the sales channel.

The theory sounds good. In reality, most enterprise supply chains are a dumpster fire of panic drills and frantic fulfillment efforts. No plan ever survives contact with reality and just about anything can go wrong, from errors in product design to giant container ships getting stuck in the Suez Canal. I once met an executive who supervised a network of private jets focused primarily on emergency supply chain runs. While spending $100,000 to ship $10,000 in plastic door handles might seem ridiculous, it’s cheaper than idling a $1 billion factory or making several thousand cars without door handles and then having staff manually attach them later. And all that has happened before we’ve dealt with sudden changes in supply or demand.

Given all the things that can go wrong, the most successful companies make the coordination of business operations a strategic priority for their top executives and are very hands-on in managing the details of the supply chain. Early on in my career, I worked at a supply chain planning software vendor, and had the opportunity to watch the CEO of a major technology company preside over its sales and operations planning meeting.

I was invited to attend the planning meeting as an observer to inform my work in process and system design. Once there, I was amazed to watch the CEO personally engage in the process. A shortage of parts from a key supplier meant that the company’s high end (and high margin) products that were key to the holiday season would be in short supply, and while that was far from ideal, there was still available capacity to increase production of the company’s lower margin products.  

In that meeting, they agreed to ramp up production of the lower margin items and change the advertising and sales emphasis to the products they would have in stock, not the ones they originally planned. The agility they showed in the planning process probably made the very best of a bad situation. That meeting set a benchmark of CEO-level engagement in supply chain management that I have never yet seen equaled.

Although there aren’t yet any models for autonomous and decentralized supply chain planning, there is at least one centralized method that comes close, and it works very well: The Kanban system was originally developed by Toyota and executed in the early days without any software at all. Like many great ideas, the system is a perfect encapsulation of the philosophy that less can sometimes be more. Kanban systems eschew advanced planning software for physical cards. When inventory runs low (or out), a card is sent to the next level back in the supply chain requesting replenishment. Without regard to some grand plan, inventory flows through the system by a set of pull signals.  

The best feature of the Kanban system is that the “planning” signal comes all the way from the marketplace and the network responds to the demand signal directly, rather than pushing products forward based on a plan. Kanban systems are simple and efficient, and work well both in paper and digital versions. 

There remain significant limitations on these kinds of systems. They do not work well in “build-to-order” environments, where every product is customized or where there are frequent engineering changes. There are also big challenges for centralized systems. Want to make a supply chain planner cry? Tell them you have an engineering change order coming.

For many cases, though, a blockchain-based autonomous supply chain network would bring big benefits. Smart contracts would be free to look at supply not just from the warehouse or factory, but perhaps also from nearby stores and other sources. Local smart contracts could get even smarter over time, developing unique, localized historical data of demand and an understanding of how to balance difference supply sources in time and money.  

To get to this future state, we still need to make dramatic progress in our ability to tokenize digital assets as they move through the supply chain. This data exists today, but it is rarely packaged in the form of digital tokens that can be seen and managed by smart contracts. As blockchain transaction capacity rises and more companies move their procurement and traceability activities on-chain, adding planning and shifting toward self-organizing networks looks more and more feasible.

The future of planning is the rise of self-organizing systems over top-down planning. From ride sharing to vacation homes, we are already experiencing how systems that are driven by individual entities can be as responsive and effective as top-down models. It is only a matter of time before this approach moves into the industrial supply chain.

The views reflected in this article are the views of the author and do not necessarily reflect the views of the global EY organization or its member firms.