It’s no secret that I’ve been obsessed since college with how the critical infrastructures that run our society work (and how brittle they are). Already then, it was obvious to me how the Internet (and the digital infrastructure that runs it) was one of these critical systems that if it went off the society of any developed country would collapse. Unfortunately, this would be also the case for other systems like the energy grid (as we’ve recently experienced in the Iberian peninsula), or the financial system (from payment networks to institutions).

But there are other less obvious critical infrastructures present in our day-to-day that could also lead to a societal collapse and that we take for granted: like the food supply chain network.

The disruption of any of these systems would sock our modern society. We take them for granted, but their infrastructure is in many cases quite brittle, and their fragility is exacerbated by the current global uncertainty and AI advancements (if AIs can hack smart contracts audited to exhaustion, why can’t they hack into water dam written in code from the 20st century by a Junior Consultant?).

This is what pushed me to start writing this series about “Critical Infrastructures of our Society”. The goal for this series is **to analyse in depth how each of these critical infrastructures operate, what are their main vulnerabilities and bottlenecks, and how (and if) they could be made more resilient. **

And I want to kick it off with one of the systems that is more alien to me: the food supply chain.

To understand the brittleness of the food supply chain, let’s first have a look high-level at how the system works and the participants involved in the system.

Farmers are (obviously) responsible for food production. The farming process in itself is its own beast and we’ll leave it out of scope for this post.

Farmers rarely sell directly to the companies that process the food or stock the supermarkets, they rely on aggregators. Depending on the region, these aggregators can take the form of local co-ops, auction houses (for e.g. fish), country elevators (massive grain silos you see in rural areas), etc. This is the “on-ramp” to the global system.

In this stage, the raw supplies (e.g. corn, soy, wheat, pork, etc.) are processed and made into actual ingredients and industrial commodities. In the case of grain this involves crushing it, milling and refining it, into wheat flour, animal feed or corn syrup.

In the case of meat, this is the slaughterhouse. At this stage is where some of the “invisible titans” of the food supply chain operate. For instance, in the case of the global grain market, we have what they call the** “ABCD” of Global Grain**. These companies control an estimated 70-90% of the global grain trade. They are known in the industry as the ABCDs: ADM (Archer Daniels Midland), Bunge, Cargill, Dreyfus (Louis Dreyfus Company). For meat, a similar oligopoly exists: JBS, Tyson, and Cargill control the vast majority of global meat processing. These giants do not just “buy and sell.” They own the silos, the ports, the ships, and the data. They essentially control the operating system of global agriculture.

This is where the ingredients move to the Consumer Packaged Goods (CPG) giants like Nestlé, PepsiCo, Kraft Heinz. They buy the refined oil and flour from the ABCDs and turn them into the “SKUs” (Stock Keeping Units) you recognize: a frozen pizza, a box of cereal, a jar of mayo.** **Fresh foods like produce or meat often skip this stage, moving directly from the processor/packer to the distribution network to minimize spoilage time).

Before food hits the store, it enters the most complex logistical node: the Retail Distribution Center (DC). This is not a warehouse, it is a sorting machine. Giant retailers (Walmart, Tesco) use a technique called cross-docking. A truck arrives on one side with 50 pallets of toilet paper. Another truck arrives with 50 pallets of apples. They are immediately unloaded, broken down, and re-stacked onto outbound trucks destined for specific stores.

This is the stage where the product travels to the supermarket shelf. This last mile resembles that of any other logistic system with the caveat that roughly 70% of our food supply relies on artificial refrigeration (with what this involves in terms of time to delivery and energetic requirements).

On paper, the infrastructure of the system is built in a way that allows billions of tons of perishables to be moved with precision, supplying the whole world with the food they need. This infrastructure was built with efficiency and profitability in mind. And in doing so, we engineered a system that is incredibly optimized for a stable world, and catastrophically ill-suited for a chaotic and uncertain one. The very features that make the food supply chain profitable are exactly what make it fragile.

Let’s dive into why the machine breaks.

The core problem is the “Efficiency Trap”. For the last 40 years, every decision in the food supply chain has been driven by a single metric: price. To drive prices down, we stripped away all the “redundancy”, the extra warehouses, the backup suppliers, the diverse crop varieties. In engineering terms, we removed the safety factors to maximize throughput. Here are the specific structural failures that I have identified in this first stage of my research:

The supply chain is wide at the bottom (with lots of farmers distributed across the world) and wide at the top (with food consumers all around the globe), but terrifyingly narrow in the middle. This narrow waist that has benefited the Internet promoting the development of protocols in the lower and higher levels of the stack through de-facto interoperability is one of the key bottlenecks for the food supply chain network. This consolidation happens on two levels:

Corporate Oligopolies: The processing stage is a bottleneck. When just four companies (the ABCDs) control close to 90% of the global grain trade, and four companies (JBS, Tyson, Cargill, Marfrig) dominating meat processing, the system loses its ability to route around damage. In a decentralised network (like those that I love), if one node fails, traffic reroutes. In this centralized network, if one major meatpacking plant shuts down (as seen with the JBS cyberattack or COVID-19 outbreaks), there is no backup capacity. The “middle” of the hourglass is simply clogged.

Geographic Hyper-Specialization: We don’t just rely on a few companies responsible for all the food processing, we also rely on a few countries being responsible for producing it. For instance, a staggering percentage of the world’s exportable wheat comes from just Russia and Ukraine (with the corresponding consequence in the Russia-Ukraine war). Over 80% of global soybean exports come from Brazil and the USA. If a drought hits the Brazilian Cerrado or a war closes the Black Sea, there is no “Plan B.” The entire global system convulses because we put all our eggs in two or three baskets.

The retail giants (Walmart, Tesco, etc.) killed the warehouse. Holding inventory costs money (rent, electricity, insurance), so they adopted the Just-in-Time model from car manufacturing. Unfortunately what works well for manufacturing where there can be predictable demands, and manufacturing can be dynamically adapted with the volume of purchases. But without perfect information in all of the nodes of the network, this results in building a more fragile network.

Supermarkets today are not storage facilities anynmore, they are display cabinets. The food you see on the shelf often arrived just hours before. Consequently, there is no shock absorber. In a robust system, a week-long trucker strike is an annoyance, but in a JIT system, it is a crisis, as the impact amplifies down the chain. There is usually only around 3 days of food in the entire retail loop at any moment, panic buying clears the shelves instantly, and the “Bullwhip Effect” triggers chaos upstream. Any impact in the node of the chain triggers amplifying attacks down/upstream.

This is the most overlooked vulnerability. Because our food travels an average of over 1,500 miles to reach our plates (which is not great for many raasons), it must pass through specific physical bottlenecks.

A Chatham House report identified 14 critical chokepoints (straits, canals, and ports) that act as the arteries of global food security. These chokepoints in the global supply chain network don’t impact the food supply chain exclusively, and with Covid we already saw what disrupting these could cause. However the impact can be higher for perishables where any delay could lead to losing the load. A blockage in any one of these—whether by a stuck ship (Ever Given), a drought, or a geopolitical blockade—doesn’t just delay an iPhone; it causes food riots in import-dependent nations like Egypt or Lebanon.

Finally, the “Efficiency Trap” applies to biology itself. To make processing easier for machines, we demanded crops that are uniform in size, taste, and growth time.

We plant millions of acres of identical genetic strains of corn, soy, and wheat. While this is efficient for harvesting, it is a biological nightmare. If a fungus or pest evolves to kill that specific strain (like the Panama Disease threatening the Cavendish banana), it doesn’t just kill one farm, it can wipe out the entire global harvest of that crop. We have removed the genetic “firewalls” that diversity provides.

As it happens with many of the critical infrastructures that we are going to discuss throughout this series, if we really want them to become more resilient, we need to re-engineer their incentives so we build in capacity for these systems to fail safely instead of only optimising for efficiency (I had a professor that always told us that “early optimisation is the source of all evil”, and we may be facing a late flavor of this).

We currently treat food purely as a commodity, like copper or plastic. The market rewards the player who can deliver cheaper to maximise the margins, regardless of the risk hidden in that price. We must view food systems as Critical Infrastructure, similar to the energy grid or defense. This means accepting that “redundancy” is not waste.

Obviously, the first step towards this is to break the narrow waist of the hourglass. We cannot rely on four companies, two geographic locations, and three canals to feed the world. We need to move toward a mesh network of food hubs. Instead of shipping all produce to massive central processing plants (single points of failure), we need more regional processing facilities. This is where digital platforms can help. By connecting small/medium farmers directly to local retailers or co-ops (bypassing the mega-aggregators), we create thousands of smaller, independent supply chains that are harder to break simultaneously.

“Just-in-Time” works for iPhones, but it is dangerous for wheat. Governments and major retailers need to re-establish Strategic Grain Reserves. Historically, civilizations from Ancient Egypt to Cold War-era America held massive stockpiles of grain to stabilize prices during harvest failures. We dismantled these in the 90s to “save money.” Bringing them back acts as a buffer against price volatility and famine. But is this even feasible or desirable?

Total localization is impossible (London cannot grow its own coffee), but total globalization is reckless. A hybrid model may be needed:* where staples (grain/oil) *maintain global trade for efficiency but diversify the sources (e.g. don’t rely solely on Russia/Ukraine for grain) incentivising short supply chains (regional sourcing) for fresh food (mainly meat, vegetables and fish). If a region can feed itself 50% of its needs, a global shock is painful but not fatal.

Finally, monocultures are an open invitation to biological collapse. We need to promote the aggro-diversity planting different varieties of wheat, corn or bananas acts as a genetic firewall. If a new pest attacks “Wheat Variety A,” “Wheat Variety B” might be immune. It creates a biological safety net that no amount of chemical fertilizer can replace.

To back my research with numbers, I decided to have a quick chat with Baselight. I highly recommend going through it because there are a lot of interesting insights, but let me paste here a few to help me paint the picture and strengthen the arguments discussed above.

Some regions are more vulnerable than others

China is not only an manufacturing super-power, but the world’s food hub

Concentration risk in an image!

Fixing this system will not be free. A resilient supply chain is inherently less “efficient” and more expensive than a brittle one. It requires holding inventory, paying for local labor, and investing in diversity. Also, from a policy perspective we should promote a system that incentivises building a more resilient network.

But as we look at a future of climate instability, AI threats, and geopolitical friction, we have to ask ourselves: what is the cost of the alternative? The price of a resilient food system might be higher at the checkout counter, but the price of a brittle one is empty shelves (as it’s been glimpsed a few times in some regions of the world in the last decade).

This post ended up being way longer than originally expected, so if you reached this point let me just thank you! In upcoming posts of the series I’ll share my research on other critical systems of our society, and dig deeper into the problems and solutions presented here at a high-level.

Here’s a list of some of the references I used for the analysis in case you want to dig deeper:

Baselight chat!

Feeding America in a Time of Crisis

The Fragile Link: Supply Chain Disruptions and Global Food Security

The future of the food supply chain: A systematic literature review and research directions towards sustainability, resilience, and technology adoption

And a book that I was recommended (but I haven’t read at least yet): The Secret Life of Groceries: The Dark Miracle of the American Supermarket

No posts