Books / Semiconductors & Strategy

Chip War Review: What Semiconductor History Teaches About Bottlenecks and Power

An operator-focused review of Chip War that turns semiconductor history into a framework for bottlenecks, dependencies, industrial concentration, and strategic risk.

The easiest way to misunderstand Chip War is to think it is mainly a book about semiconductors.

Chris Miller certainly explains the industry. But the strategic value of the book comes from something broader: it shows how economic power accumulates around bottlenecks that are difficult to reproduce.

Modern chips require a chain of design software, intellectual property, manufacturing equipment, materials, fabrication, packaging, and specialized knowledge. No single country or company controls every layer. Yet some nodes are so concentrated that they shape the options available to everyone downstream.

That makes Chip War a book about dependency architecture.

The central lesson: markets can be global while capabilities remain concentrated

Executives often talk about supply chains as if every supplier is interchangeable given enough money. Semiconductors show why that assumption fails.

A sophisticated chip is not the output of one factory. It is the result of tightly coupled capabilities developed over decades. Design firms rely on specialized tools and manufacturing partners. Fabrication plants rely on equipment suppliers with extreme technical depth. Equipment suppliers depend on precision components and materials. Customers design products around process nodes that may be available from only a small number of producers.

The industry can therefore be large and competitive while still containing critical chokepoints.

That distinction matters far beyond chips.

If ten suppliers exist but only one can satisfy a critical specification at the required scale and yield, you do not have ten substitutes. You have one operational dependency and nine names on a spreadsheet.

Why the book remains commercially relevant

The semiconductor market has become even larger since the book appeared. The Semiconductor Industry Association reported $791.7 billion in global semiconductor sales for 2025, up 25.6% from 2024, and projected the market to approach roughly $1 trillion in 2026.

Scale amplifies the strategic importance of bottlenecks. Chips now sit underneath AI infrastructure, data centers, vehicles, industrial automation, telecom, consumer devices, and defense systems.

The implication for leaders is not that every company needs a semiconductor strategy. It is that every company should understand which external capabilities could become strategically scarce as its own product becomes more digital.

A five-layer chokepoint map

One of the best ways to apply Chip War is to map a critical product through five layers.

1. Knowledge chokepoints

Which outcomes require tacit expertise that cannot be purchased quickly?

Some capabilities live in procedures. Others live in engineers who have solved thousands of edge cases. Tacit knowledge creates a bottleneck because documentation does not fully reproduce experience.

2. Tooling chokepoints

Which pieces of equipment, software, or infrastructure have very few credible suppliers?

The more specialized the tool, the more likely the supply chain depends on long development cycles and proprietary know-how.

3. Capacity chokepoints

Even when multiple suppliers can theoretically perform a task, which ones have enough qualified capacity when demand spikes?

This is where nominal redundancy can fail. A backup supplier that cannot absorb meaningful volume is not full redundancy.

4. Coordination chokepoints

Which organizations have to work together with unusually tight tolerances?

Complexity creates its own barrier. A replacement supplier may be technically competent but still fail because integration, qualification, and learning take too long.

5. Political and jurisdictional chokepoints

Where does a commercial dependency intersect with export rules, industrial policy, sanctions, security requirements, or geopolitical risk?

This is the most politically sensitive layer, and it should be analyzed descriptively rather than as a prediction. The relevant business fact is that legal access to technology can change independently of engineering capability.

Chip War explains why “make it ourselves” is often not a real strategy

When a dependency becomes uncomfortable, executives sometimes jump to vertical integration.

The semiconductor story shows how expensive that can be.

Replicating a concentrated capability may require not only capital but process knowledge, supplier ecosystems, talent, qualification time, and years of accumulated learning. Throwing money at a bottleneck can help, but it does not erase the learning curve.

The correct strategic question is therefore not simply “Can we build this?”

It is:

• How long would capability parity take?
• What must be learned rather than purchased?
• Which complements would still remain external?
• What utilization rate would justify the fixed cost?
• How would incumbents improve while we catch up?

Vertical integration can reduce one dependency while creating another.

The book’s strongest business lesson: dependency is a portfolio

A company should not classify suppliers as simply “critical” or “noncritical.” Dependencies differ.

A useful matrix has two dimensions:

Impact if unavailable and difficulty of substitution.

| Dependency type | Impact | Substitution difficulty | Management response | | — | — | — | — | | Commodity input | Low | Low | Standard procurement | | Expensive but replaceable | Medium | Low | Cost / contract management | | Specialized component | High | Medium | Dual source, inventory, qualification | | Deep capability bottleneck | Very high | High | Executive-level risk strategy | | Regulated / geopolitical bottleneck | Very high | Variable | Scenario planning + legal monitoring |

This framework prevents teams from wasting attention on visible costs while ignoring invisible concentration.

Where Chip War is strongest

It explains technology as accumulated industrial capability

The book resists the myth that technological leadership comes from a single brilliant invention. Leadership compounds through design ecosystems, manufacturing process, capital, state investment, education, suppliers, and customer demand.

That is useful for founders because durable advantage often looks similar. One feature is easy to copy. A system of capabilities is harder.

It makes geography economically meaningful

Digital businesses often talk as if location has disappeared. Semiconductor fabrication makes the opposite point: advanced physical production is embedded in places, supplier networks, infrastructure, and skilled labor.

The lesson is not that every industry will re-localize. It is that leaders should know which parts of their “digital” business ultimately depend on physical concentration.

It separates design power from manufacturing power

A company can lead one layer while depending on another. That is strategically important because revenue leadership does not equal full-stack autonomy.

The same pattern appears in SaaS companies dependent on cloud platforms, ecommerce businesses dependent on marketplaces, and AI products dependent on model or compute providers.

Where the book requires caution

A geopolitical history naturally invites readers to treat industrial competition as a zero-sum scoreboard. Business leaders should resist turning that framing into automatic policy conclusions.

Countries can simultaneously compete, trade, specialize, regulate, and cooperate. The economic consequences of controls or subsidies depend on implementation, time horizon, substitution, and responses from other actors.

The book is most useful when it reveals dependencies. It is less useful if readers turn every dependency into a prediction of conflict.

A second caution is that bottlenecks move. Today’s scarce capability can be attacked by investment, new architectures, alternative suppliers, or demand shifts. A chokepoint is valuable precisely because others have an incentive to route around it.

A founder’s version of semiconductor strategy

You can apply the logic without touching a chip factory.

Ask:

What is our irreplaceable external dependency?

Maybe it is a cloud provider, app-store distribution, one ad platform, a payment processor, a data provider, a specialist manufacturer, or one acquisition channel.

What makes it hard to replace?

Separate contractual friction from technical integration, data history, user habit, regulation, capacity, and knowledge.

How long would substitution take under stress?

“Three vendors exist” is not enough. Estimate migration time when the primary vendor is unavailable.

What leading indicators tell us the dependency is becoming riskier?

Price changes, service quality, regulation, supplier capex, customer concentration, geopolitical exposure, or product roadmap divergence can all matter.

What option can we build cheaply now?

A secondary integration, exportable data, alternative supplier qualification, modular architecture, or extra inventory may be much cheaper before a crisis.

The strategic concept underneath the history: optionality

Resilience is not the same as redundancy.

Redundancy duplicates resources. Optionality preserves the ability to choose among paths as new information arrives.

A company may not need two complete production stacks. It may need architecture that allows a second supplier to be qualified quickly. It may not need to migrate clouds today. It may need clean abstractions, backups, and tested recovery procedures that keep migration possible.

Chip War makes optionality tangible because the cost of losing an option can be enormous when a supply chain has deep bottlenecks.

Competitive advantage and bottleneck ownership

The book also provides a lens on moats.

A bottleneck can produce economic power when four conditions hold:

1. the capability is essential to valuable downstream activity;
2. supply is limited or difficult to reproduce;
3. substitutes are weak;
4. the owner can capture some of the value without accelerating substitution too quickly.

The fourth point matters. Excessive extraction can finance the competitor trying to eliminate your moat.

For founders, this is a reminder that defensibility is not merely being unique. It is occupying a position others need and cannot cheaply route around.

A dependency audit inspired by Chip War

Run this quarterly for strategically important products:

• List the ten external inputs without which the product cannot operate.
• Estimate time-to-replace for each one.
• Identify the true technical qualifier, not merely the supplier name.
• Map second-order dependencies your suppliers rely on.
• Mark any dependency concentrated in one geography, platform, or regulatory regime.
• Define a low-cost option that reduces exposure.
• Assign an owner and trigger for escalation.

Most companies will discover that their biggest risk is not the supplier with the highest invoice. It is the supplier with the longest replacement path.

The semiconductor value chain rewards asymmetric excellence

A useful way to read Chip War is to notice how rarely one company needs to dominate the entire chain. Competitive advantage often comes from being extraordinarily difficult to replace at one layer while cooperating with specialists at the others.

This is a more precise strategy than “vertical integration is good” or “outsourcing is good.” The right question is where integration creates a learning advantage and where specialization creates a scale advantage. Semiconductor firms have repeatedly made different choices because design software, architecture, fabrication equipment, foundry operations, memory, packaging, and end-market distribution have different economics.

For a non-semiconductor company, the analogy is powerful. A business should know which layer of its value chain contains the scarce capability it must own and which layers are better purchased from partners. Owning everything can destroy focus. Owning nothing can destroy bargaining power.

A bottleneck can move

One danger of copying the current semiconductor map is assuming today’s choke point will remain tomorrow’s choke point. Technology changes where constraints sit.

Advanced packaging, memory bandwidth, energy supply, data-center construction, software ecosystems, and specialized accelerators can each become more important as architectures evolve. A strategy built around one permanent bottleneck can fail because innovation often relocates scarcity rather than eliminating it.

That suggests a more dynamic dependency review. Do not ask only “What constrains us now?” Also ask:

• What constraint is becoming easier?
• What new constraint appears if the old one is solved?
• Which suppliers gain negotiating power if architecture changes?
• Which capability is becoming standardized and therefore less defensible?
• Which complementary capability becomes more valuable as the core technology improves?

This “moving bottleneck” view is useful in almost every technology market. When distribution becomes cheap, attention becomes scarce. When compute becomes abundant, energy or data may become scarce. When software creation accelerates, trust, integration, or customer acquisition can become the limiting factor.

Resilience is not the same as duplication

After reading a book about strategic dependencies, the intuitive response is to duplicate everything: two suppliers, two regions, two factories, two technology stacks. That can create resilience, but it can also double complexity without removing the underlying correlated risk.

Two suppliers may depend on the same upstream fab. Two clouds may rely on the same identity provider. Two logistics partners may share the same port. Two teams may depend on the same undocumented expert. Redundancy is meaningful only when the failure modes are sufficiently independent.

A proper resilience audit therefore traces dependencies at least one layer deeper than the visible vendor. The goal is not maximum redundancy. It is to understand which failure would stop the system, how quickly substitution can occur, and which options are worth paying for before the crisis arrives.

That is one of Chip War’s most transferable insights: strategic resilience is built through knowledge of the chain, not through the comforting appearance of diversification.

Who should read Chip War

Founders, investors, strategy leaders, procurement teams, operations leaders, and anyone working in AI infrastructure or advanced manufacturing will find the book useful.

It is also useful for purely digital leaders because it breaks the illusion that software floats above the physical economy.

Readers should pair the historical narrative with current industry data, because the semiconductor market changes quickly and policy details can move faster than books.

Final assessment

Chip War succeeds because it turns a technical industry into a story about accumulated capability, concentration, and strategic dependence.

The durable lesson is not “chips are important.” Everyone already knows that.

The durable lesson is: find the bottleneck that the rest of the system quietly assumes will always be available. Then understand what makes it scarce, how quickly substitutes can emerge, and what option you can build before the constraint becomes a crisis.

That is a strategy lesson any company can use.

FAQ

What is Chip War mainly about?

It is a history of the semiconductor industry and the economic and geopolitical competition around critical chip capabilities.

Why is Chip War useful for business leaders?

It shows how bottlenecks, supplier concentration, tacit knowledge, and long learning curves create strategic power and operational risk.

Is the book still relevant as the industry changes?

Yes, but current market and policy data should be used alongside the book because semiconductor capacity, regulation, and technology evolve rapidly.

What is the most transferable lesson?

Map dependencies by impact and substitution difficulty rather than assuming a large supplier market automatically means true resilience.

Continue inside the Millionaire Lesson library

Competitive Advantage Switching Costs Competitive Strategy

Sources

• Simon & Schuster — Chip War
• Semiconductor Industry Association — 2025 global semiconductor sales
• Chris Miller — official publisher author profile
• NIST — AI Risk Management Framework (for downstream AI infrastructure risk context)
• Millionaire Lesson — Switching Costs (strategic dependency context)