For decades, business learned to strip out inventories, spare capacity, and duplication. Now part of that “inefficiency” is beginning to look like the price of keeping a system alive.
If a natural ecosystem were judged by the logic of a maximally optimized factory, it would contain many suspicious features. Several species can perform similar functions. Different organisms use the same resource with different levels of efficiency. Part of the system’s potential appears to sit idle much of the time. From the perspective of short-term productivity, this can look like duplication. Yet after a shock, what looked “redundant” can become critical. In 2026, researchers in Nature analyzed morphological data for 3,696 bird species across 1,281 sites worldwide. Their conclusion was more cautious than the simple claim that “more species are always better,” but it matters for this parallel: lower functional redundancy made ecosystem functions more vulnerable to further losses. If several participants can support the same function, losing one does not necessarily break the whole system.
The global economy moved for decades in the opposite direction. Where inventories could be reduced, costs fell. Where one supplier was cheaper than three, concentration looked rational. Where production could be moved to the most efficient region, duplicating it in a more expensive country looked wasteful.
The problem appears only when we change the question.
Not “how much does the system cost when everything works normally?” but “how much does it cost once rare but extremely expensive disruptions are included?”
By that measure, the most optimized system is not always the most efficient one.
The current assessment is that by the end of 2030, major industrial economies are unlikely to return to a model in which minimum unit cost is the dominant criterion for every critical supply chain. The most likely outcome is a compromise: global trade remains, but countries and companies increasingly pay for alternative suppliers, reserves, spare capacity, and the ability to switch quickly at strategically important nodes.
The probability that this structural shift will still be in place by 2030 can currently be estimated at about 80%. Confidence is 72 out of 100.
Why maximum efficiency can create fragility
Optimization itself is not the problem. It underlies a huge share of economic progress: specialization reduces costs, large producers exploit economies of scale, and global trade allows components to be purchased where they can be made best or cheapest.
The danger begins when optimization gradually removes almost all of the system’s margin for error.
Imagine two factories. The first has one supplier for a critical component, minimal inventory, and production capacity running close to full utilization. The second buys the same component in two countries, keeps a safety stock, and preserves the ability to raise output after a disruption.
In a normal year, the first factory will almost certainly look better. Its unit costs are lower. Less capital is tied up in inventory. Capacity is used more fully.
Then one supplier disappears.
Suddenly, what looked like unnecessary expense in the second factory’s accounting becomes an option to keep operating.
Economically, this resembles insurance. A homeowner does not buy insurance because they expect a fire next Thursday. They accept a small certain cost today to reduce the consequences of a rare, very large loss tomorrow.
Inventories, supplier duplication, and reserve production capacity can increasingly be viewed in the same way.
Nature does not say “duplicate everything” – it suggests a different rule
A natural parallel can easily be pushed too far. An economy is not a forest, a company is not a population, and a second factory is not a second biological species. An ecological study therefore cannot directly prescribe economic policy.
But it can help reveal the mechanism.
Resilience does not emerge simply because a system contains “a lot of everything.” What matters is whether there are alternative ways to perform a critical function.
That is fundamentally different from mindlessly accumulating stock. If a company buys the same critical component from three suppliers, but all three depend on the same raw-material plant, the visible diversity provides little protection. If a country has five importers bringing goods through one transport chokepoint, it still has a single point of failure.
So future resilience is not a fight against efficiency.
It is a fight against hidden concentration of risk.
And here comes the most interesting turn: the world has already begun to pay to reduce it.
What recently looked like waste is becoming economic policy
In 2025, the OECD modeled the opposite extreme: what would happen if countries moved production home on a large scale in the name of resilience. The result does not support the simple story that localization automatically makes economies safer. In the organization’s model, large-scale relocalization could reduce global trade by more than 18% and global real GDP by more than 5%; in more than half of the modeled economies, GDP volatility did not fall but increased.
This is an important limit on the forecast. The future of resilience is not autarky.
Closing borders, producing everything domestically, and abandoning global specialization is also a form of concentration. Instead of dependence on an external supplier, a country becomes dependent on its own production capacity, its own raw materials, and its own domestic shocks.
A more rational model looks different: global links remain, but the most dangerous dependencies are gradually spread out.
The European Union has already embedded this logic in policy for strategic raw materials. The European Commission set 2030 benchmarks that include expanding extraction, processing, and recycling inside the EU and limiting dependence on any single third country to no more than 65% of annual consumption of each strategic raw material at the relevant processing stage. The legal framework in EUR-Lex embeds the same logic in the Critical Raw Materials Act. This is no longer optimization solely for the lowest price: concentration itself becomes a factor policy tries to constrain.
A similar logic is visible in semiconductors. The U.S. CHIPS program received roughly $50 billion in federal funding to support semiconductor manufacturing, research, and supply chains in the United States. The Commerce Department explicitly links these investments to economic and national security and supply-chain resilience in its Commerce materials.
Japan stated the problem even more directly in 2026. An April METI report discusses strengthening the manufacturing base in light of geopolitical risks, diversifying critical supply chains, creating alternative production capabilities, and supporting areas where the private sector may struggle to pay for the required level of resilience on its own.
These examples do not prove that the entire world economy has changed its model. They show something narrower but more important: major industrial systems have already begun to institutionalize the price of resilience.
The world may not be abandoning optimization – it may be changing what it optimizes
This is the main trap hidden in the title of this article.
It can look as if the choice is efficiency or resilience. In reality, a competitive economy cannot afford to abandon efficiency over the long term. If a company keeps ten times more inventory, builds five factories for every component, and buys only from the most expensive “safe” suppliers, its products will stop being competitive.
Maximum resilience is therefore just as irrational as maximum efficiency.
What changes is the function being optimized.
The old simplified model looked roughly like this:
minimize unit cost under normal conditions.
The new model may increasingly look like this:
WOW:
minimize average long-term cost together with the risk of major disruptions.
The difference sounds small, but it changes almost everything.
A second supplier is no longer automatically “expensive duplication.” It can be insurance. A stock of a critical component no longer automatically means inefficient use of capital. A reserve production line can have value even if it is underused for most of the year.
Digital technology can change this arithmetic further. If a company can see several tiers into its supplier network, simulate disruptions, reroute quickly, and use AI to reorganize procurement, it may need less physical “redundancy” to achieve the same resilience.
A future system can therefore paradoxically become both more resilient and more efficient, simply because efficiency is no longer measured over one quiet year.
Globalization does not necessarily end
Another tempting mistake is to label the entire process deglobalization.
If the United States supports domestic semiconductor production, the EU limits excessive dependence on a single raw-material supplier, and Japan discusses alternative production capacity, it is easy to imagine a future of closed economic blocs.
But that is only one possible scenario.
The OECD emphasizes that simply relocating all production can be expensive and may fail to deliver the expected increase in resilience. That suggests the smartest strategy is not to make every chain short at any cost, but to make it switchable.
Two geographically independent sources may be more useful than one domestic producer. A stockpile of a critical component may be cheaper than building a second factory. A standard that allows one component to be substituted quickly can sometimes be more valuable than another warehouse.
So the real opposite of maximum optimization is not localization.
It is optionality.
A system has several ways to continue performing its function when one route stops working.
This is where the economic parallel with nature becomes strongest.
Four possible worlds by 2030
The future trajectory depends above all on whether major shocks remain frequent and costly enough to justify a permanent “resilience premium.”
| Scenario | Probability | What it means |
| Selective resilience | 60% | Global supply chains remain, but critical nodes gain alternative suppliers, reserves, local capacity, and fast-switching systems. |
| Bloc economy | 20% | Geopolitical competition intensifies enough that countries accept much higher costs for localization and production inside friendly blocs. |
| Efficiency returns | 15% | Major disruptions ease, technology improves risk management, and high costs push companies to concentrate production again where it is cheapest. |
| Digital resilience replaces physical buffers | 5% | AI, forecasting, and automatic supply-chain reconfiguration improve enough that much of physical duplication can be replaced by rapid adaptation. |
The first two scenarios together account for about 80%, which forms the current main forecast. But the difference between them is enormous. In the first, the world remains interdependent and simply insures critical points. In the second, resilience becomes political fragmentation, with duplication driven less by economic calculation than by distrust between blocs.
The first scenario currently looks the most viable.
The most important question is where to permit “inefficiency”
The hardest task of the next decade may not be finding a way to make everything maximally resilient, but identifying the nodes where a buffer is actually worth its cost.
There is little economic logic in keeping six months of inventory for every cheap product. But the calculation changes if a component worth a few dollars can stop a factory producing billions of dollars of output.
There is no reason to duplicate every facility. But it may be rational for a component produced in only one geographic area with no quick substitute.
There is no reason to abandon the most efficient supplier simply because it is large. But dependence looks very different if losing that supplier for six months would paralyze an entire industry.
Future competitive advantage may therefore come not from having the most reserves, but from understanding one’s own points of failure better than competitors do.
Just as a natural system depends not on an abstract number of species but on the functions they perform and whether they can compensate for one another’s loss, an economy does not need the maximum possible number of suppliers, factories, and warehouses. It needs enough genuinely independent alternatives at critical points.
That is a much subtler strategy than the popular phrase “just in case instead of just in time.”
What would change the forecast
The current estimate should rise if major industrial economies after 2026 continue not merely to talk about resilience but to spend money on duplicated critical production, strategic reserves, and supplier diversification even when alternative sources cost more. A particularly strong signal would be the spread of this logic from semiconductors and critical minerals into pharmaceuticals, energy equipment, digital infrastructure, and other sectors.
The estimate should fall if, after several calmer years, the economic cost of redundancy becomes politically and corporately unacceptable, major localization programs are broadly scaled back, and technology makes it possible to replace suppliers almost instantly without large physical buffers.
Another counter-signal would be concentration continuing to rise despite policy programs. That would indicate that economies of scale remain stronger than the desire to diversify.
This is also where the forecast could be wrong: the world has already recognized the problem, but recognizing risk is cheaper than paying for insurance against it for decades.
Efficiency does not lose. “Slack” simply stops being a dirty word
Perhaps the most important change will happen not in factories, warehouses, or trade policy, but in how efficiency itself is defined.
During calm periods, a system without reserves almost always looks better. That is why buffers are so easy to cut: their cost is visible every day, while their benefit becomes visible only during a crisis.
Nature operates on a different horizon. It does not optimize a quarterly report. A system that performs extraordinarily well under one combination of conditions but collapses after the first major change simply has a lower chance of persisting for long.
Economies should not copy nature. But they can borrow one useful idea: what looks like duplication in normal times can become a resource for adaptation in abnormal times.
The world is therefore unlikely to abandon optimization.
It is more likely to abandon its simplest version — the one in which the cheapest system is automatically considered the best.
The next era may favor systems that are slightly more expensive every day but much less likely to stop working altogether.
Forecast card
| Field | Recorded version |
| Forecast question | By December 31, 2030, will the United States, the European Union, and Japan still maintain institutionalized policies aimed at increasing the resilience of critical supply chains? |
| Probability | About 80% |
| Confidence | 72 out of 100 — moderately high |
| YES criterion | As of December 31, 2030, each of the three systems — the United States, the EU, and Japan — has at least one national or supranational law, program, or strategy for a critical supply chain that funds or requires at least one of the following: supplier diversification, limits on excessive dependence, strategic reserves, reserve production capacity, a local/regional production base, or an official mechanism for responding to disruptions. |
| NO criterion | By the deadline, at least one of the three systems no longer has any active instrument meeting the criterion above. |
| Horizon | Through December 31, 2030 |
| Resolution date | January 15, 2031 |
| Forecast snapshot date | September 7, 2026 |
| Main scenario | Open global economy + selective buffering of critical nodes |
| Forecast history | September 7, 2026 — initial estimate: 80%; confidence 72/100 |
Methodological limitation: there is no honest single historical base rate for a transition this broad. Energy reserves, industrial policy, logistics, and natural ecosystems are different systems, so they are used to analyze mechanisms and counterexamples rather than combined into artificial “analogue statistics.”
Disclaimer
This article contains an analytical estimate of the future, not a claim that a predetermined outcome will occur. The probability is recorded as of September 7, 2026 and may change as new data appear. Natural parallels are used to explain mechanisms and are not evidence that economic systems literally repeat the behavior of ecosystems.