K-Shaped Divergence in Markets: Three Forces Reshaping the Economy

I. The First Force: AI Supply Rigidity Stretches the Industrial Cycle

To understand K-shaped divergence, start with one change: the AI industry's boom cycle does not run on the same logic as a traditional industrial cycle.

The conventional manufacturing cycle is straightforward — demand grows, companies expand capacity, supply increases, prices fall, the boom ends. The whole thing typically plays out over one to two years, because traditional manufacturing supply is highly elastic. Where there is profit, capacity can be added within a few quarters.

AI is different. Its core constraint is not demand but supply — and more specifically, the time required to create supply.

A leading-edge fab takes three to five years from groundbreaking to volume production. High-end lithography equipment has delivery lead times of two to three years. HBM yield ramping takes one to two years. Data center construction runs about eighteen months. What these links share is that their barriers to entry are a function of time, not price. Double the price and capacity still cannot catch up in the short run.

The direct consequence: the boom lasts longer. Manufacturing booms used to fade within one to two years as supply caught up. AI supply release takes three to five years. The AI cycle may outlast any traditional industrial cycle.

But supply rigidity is not uniform across the entire AI value chain. It holds in specific bottlenecks — HBM and advanced DRAM, CoWoS and other advanced packaging, sub-3nm foundry capacity. Standard server assembly, commodity cables and connectors, and lower-complexity components respond to profitability on a more conventional timeline. A genuine supply bottleneck can extend a cycle. A compelling technology narrative cannot.

On one side, an AI supply chain where constraints keep the boom running beyond expectations. On the other, industries still governed by traditional cycle logic, growing more modestly. Different cycle lengths produce different earnings trajectories — and therefore different price trajectories. That is the first pillar of K-shaped divergence.

---

II. The Second Force: Policy Framework Transformation Changes Counter-Cyclical Logic

The second pillar comes from a deep shift in the macro policy framework.

For two decades, Chinese macro policy had a relatively stable anchor: when the economy slowed enough, policy would step in to support. Inventory cycle, property cycle, credit cycle — each round came with policy expansion and contraction.

That anchor is loosening. One observation makes it visible: finished-goods inventory growth fell to historic lows by June 2023, yet the large-scale stimulus markets expected never arrived. Take the first half of 2026 — fiscal revenue grew, but general public budget expenditure was close to zero growth year-on-year.

This is not policy failure. It is a shift in policy philosophy: from short-cycle growth stabilization to long-term structural optimization, from aggregate stimulus to targeted support, from demand-side management to supply-side reform. The old playbook of betting on a policy pivot for sector-wide beta no longer applies.

Behind it sit global macro constraints:

• Global liquidity tightening. Most advanced and emerging economies have entered a tightening cycle of rate hikes or paused cuts. Japan, Australia, the eurozone, and a number of Southeast Asian and African countries have raised rates. China is one of the few major economies still holding room to ease, but the China-US yield differential and exchange-rate stability form hard constraints.
• Narrower debt space. After several rounds of leverage expansion, local government and corporate debt capacity is not what it was. Policy must strike a finer balance between growth and risk.
• Higher priority on industrial upgrading. Policy resources increasingly flow toward technology, advanced manufacturing, and green energy, not the traditional property and infrastructure channels.

The question for investors shifts from "when will policymakers turn on the liquidity?" to "which companies are actually positioned to benefit from the direction of policy?" That is bottom-up work, not macro bet-making.

Recent data makes the point. Manufacturing PMI returned to expansion, but non-manufacturing PMI improved far less, and high-tech manufacturing growth significantly outpaced traditional industry. Policy no longer spreads evenly — and that itself amplifies K-shaped divergence. The economy can recover in aggregate while the distribution of that recovery becomes increasingly uneven. That is the essence of a K-shaped economy.

---

III. The Third Force: Global Industrial Redivision Changes Capital Allocation

K-shaped divergence is not unique to China. Pull the lens back and nearly every major economy's equity market is going through a similar structural split. Markets weighted toward technology manufacturing are performing better; markets weighted toward traditional services are lagging.

This cross-market synchronicity cannot be explained by "sentiment" or "style." Something deeper is at work: AI is reshaping the comparative advantage that underpins the global division of labor.

From Cost Arbitrage to Capability Scarcity

For two decades, the core logic of globalization was cost arbitrage. Industry moved to wherever labor was cheapest.

In the AI era, the basis of comparative advantage is shifting from cost to capability.

When comparative advantage moves from cost to capability, industrial structure becomes far more stable and persistent. A cost advantage can be challenged by a cheaper producer. A capability advantage — accumulated know-how, specialized equipment, supplier ecosystems, engineering talent, years of process optimization — cannot be copied in the short run. Capital's response follows a recognizable pattern. It is repricing global industrial competitiveness. That process breaks down into four layers.

First, Cross-Border Capital Flows

Economies with high AI weight — the US, Japan, Korea, Taiwan — continue to earn a capital premium. Countries with large service sectors and thin AI assets — much of Europe, parts of Latin America — recover more slowly. This divergence crosses national borders, a global signal: capital is not betting on individual champions but on an industrial direction.

Second, Geographic Concentration of Capacity

Leading-edge fab capacity is heavily concentrated in Taiwan and Korea; 3nm-and-below nodes sit almost entirely in those two regions. HBM shipments come entirely from SK Hynix, Samsung, and Micron, with SK Hynix alone holding roughly half. EUV lithography tools are supplied globally by a single company, ASML, with annual shipments measured in the tens of units. Capacity rebuild takes three to five years at minimum, with no shortcut. Short-term cost advantage can be bought by cutting prices; capability barriers can only be closed with time.

Third, Supply Elasticity as a Measure of Technological Scarcity

Differences in supply elasticity are the most intuitive gauge of technology barriers. Bulk chemicals like PVC are highly elastic — demand pulls price up, supply responds, and the price move fades quickly. Lithium salts for solar-grade use sit in the middle. DRAM — the memory chip central to AI compute — has the lowest supply elasticity of all. Once demand starts, price moves are sharp and persistent. On a lead-time-to-price (LP) basis, traditional manufacturing translates a price signal into capacity response in about one month, with cumulative supply gaps reaching roughly 4x. The equivalent transmission in the AI chain takes five to eight months, with cumulative gaps only about half those of traditional industry. The less elastic the supply curve, the more powerful a demand shock becomes. This is not speculation — it is a measurable industry regularity.

Fourth, Global Policy Resonance

Across major economies, industrial policy and fiscal resources keep tilting toward AI and advanced manufacturing. The US, Japan, Korea, and Germany have all launched large-scale semiconductor and AI infrastructure programs. At the same time, most advanced and emerging markets have entered a monetary tightening cycle of rate hikes or paused cuts. China alone retains room to ease, but the space is narrow, constrained by the China-US yield spread and exchange-rate stability. The combination — global monetary tightening alongside localized fiscal expansion — itself drives divergence: against a broadly tight macro backdrop, industries that receive policy support see their advantage amplified.

---

IV. An Analytical Framework: The Three-Force Model

These three forces can be organized into a simple thinking tool.

The point of this framework is to organize thinking, not to produce a forecast.

• When growth stocks keep outperforming, don't just call it "market mania." Ask: is there a real constraint on the supply side?
• When cyclical stocks refuse to move, don't just complain that "policy isn't enough." Ask whether the policy framework has structurally changed.
• When markets diverge in sync worldwide, don't look only at China. Ask whether the global division of labor is being reshuffled.

Read through this lens and many seemingly contradictory things become legible. For instance, when TMT sells off sharply, why does the discussion center on "style rebalancing" rather than "the end of the industrial trend"? Because the selloff is emotional; the three forces are structural. Short-term price movements affect volatility. Structural forces affect the direction and duration of the cycle. One creates noise, the other sets direction.

---

V. The Boundaries of This Framework

Every framework has limits. Being honest about them is a basic requirement of serious research.

• Supply rigidity is not permanent. Engineering timelines can shorten — modular data centers, AI-assisted process optimization could both change the time variable. When supply constraints break, the cycle may prove shorter than expected.
• The policy framework is not a one-way street. If the external environment shifts violently — a deep global recession, say — policy priorities may rebalance toward short-term stabilization. The framework needs to be revised as new information arrives.
• Global division rebuild may run into headwinds. Geopolitics, trade frictions, technology blockades can all alter the path and speed of industrial transfer. The reshaping will not be smooth. Some industries may become more concentrated; others may fragment as countries prioritize resilience over efficiency.

This framework is not for producing certainty. Its value is in organizing information, forming judgment, and cutting down on random and emotional decisions.

---

Closing

K-shaped divergence may not last forever. Every industrial cycle moves through boom, maturity, and adjustment. Every technology revolution moves through optimism and a return to reason.

What we need to understand today is not when K-shaped ends, but why it formed, why it persists, and what forces keep driving its evolution.

Markets do not run on our forecasts. They evolve according to economic structure. The durable investment skill is not predicting the next rise or fall — it is continually updating the way you understand the market.

The three forces discussed here may prove temporary, or they may persist for much longer than expected. What matters is not whether the framework is permanently correct. What matters is whether it helps us recognize which assumptions are changing before the market fully prices them in.

*Magic Econ · Common Sense Jack*

Magic Econ is an independent research platform focused on understanding the structural forces beneath markets. This article is based on publicly available market data and research materials, intended to build an analytical framework for understanding structural economic change. It is for educational exchange only and does not constitute investment advice.