Technological, geopolitical, and capital transformation are interconnected, with unprecedented impacts on a globally connected, directly entangled economic and geopolitical world. Choices are simultaneously investment decisions, national security decisions, and civilizational bets on which technological architecture will define the next century.
Understanding them requires insightful economic, strategic, and institutional thinking. More than ever, it requires intellectual courage and patience with complexity.
Consider three examples, each of which this book examines in depth.
The first: the United States has invested enormous political capital and policy energy in a strategy of export controls on advanced semiconductors to constrain China’s AI development. The strategic logic is straightforward and, within its own frame, coherent: control the hardware, control the capability. Then DeepSeek arrived.
A Chinese team, working with a fraction of the compute budget available to American frontier labs and under the precise export control constraints designed to make their work impossible, produced an AI system competitive with the best American offerings.
The hardware strategy had encountered an inconvenient reality. When the binding constraint on AI capability is algorithmic innovation rather than hardware access, hardware controls buy time but do not buy the outcome. The analysis that produced the strategy was technically sophisticated.
It was also wrong.
The second: the global defense industry has built a deep and largely unexamined dependence on rare-earth minerals processed almost exclusively in China. More than 90% of the world’s heavy rare-earth processing—the materials essential for permanent magnets in fighter aircraft, guided munitions, naval vessels, and advanced communications systems—runs through Chinese facilities.
The United States Navy relies on Chinese-processed materials for over 90% of its weapons systems. This dependence was not the result of a strategic decision. It was the accumulated product of individual economic decisions, each rational in isolation, that collectively produced a catastrophic strategic vulnerability.
The alarm this fact should produce is hidden by ignorance and an inability to connect the “critical mineral” dots.
The third: the hyperscalers—the handful of technology companies whose collective infrastructure spending now approaches $100 billion per quarter—are making historic capital-allocation decisions on the premise that the scaling approach to AI will continue to produce returns proportional to investment.
This premise has been correct for a decade. Whether it will remain correct is a genuinely open question, because the evidence for continued scaling returns is encountering limits that every extrapolated trend eventually encounters.
The capital markets are pricing a confident future. The evidence supports a more uncertain one. The gap between those two statements is where the most consequential investment errors of the coming decade will be made.
Each of these failures shares a common structure: confident analysis within a single domain that ignores the dynamics in adjacent domains that determine the actual outcome.
Rigorous systems thinking is essential – holding multiple levels of analysis simultaneously, recognizing that emergent properties at the system level are frequently invisible from within any single component, and maintaining the intellectual discipline to follow the analysis wherever it leads—even when it crosses the boundaries between fields that academic training, professional practice, and media coverage treat as walls.
This book asks three things of the reader.
- Question Assumptions
Most readers approach a book on artificial intelligence and geopolitics with prior views—about China, about AI risk, about the competence of American industrial policy, about whether technology is net beneficial or net threatening. Those views are the accumulated product of serious thought and genuine experience, and they deserve respect.
But they are also, in the domains this book examines, more likely than usual to be models reinforced by selective information processing until they feel less like hypotheses and more like facts. The request is not to abandon prior views. It is to hold them provisionally, to test them against the analysis rather than fitting the analysis to them, and to take seriously the possibility that the most foundational assumptions are the ones most in need of revision.
- Be Patient with Complexity
The argument of this book is not reducible to a thesis that can be stated in a sentence and then illustrated. It is an argument about a system, and systems cannot be understood through any single dimension. The chapters build on each other in ways that will sometimes not be visible until later in the reading.
The reader who skips to the sections that seem most immediately relevant will find value there, but will miss the cumulative logic that makes the whole considerably greater than the sum of its parts.
- Intellectual Courage.
Some of the conclusions in this book are uncomfortable.
- The American export control strategy cannot substitute for genuine investment in innovation is uncomfortable for American policymakers.
- China’s technological capabilities are real and, in some dimensions, rapidly advancing, is uncomfortable for those whose strategic framework requires Chinese weakness.
- Current AI infrastructure investment may follow the historical pattern of technology bubbles—creating genuine value at the system level while destroying fortunes at the investor level—is uncomfortable for those who have committed capital to the current cycle.
- The timeline for critical mineral supply chain resilience is incompatible with the political timeline on which the investments required to achieve it must be sustained is uncomfortable for anyone who prefers reassuring pronouncements to reality and the difficulty of execution.
This book is organized as a cumulative argument. Each section builds the analytical foundation for what follows. Chapters are components of an integrated whole. The final section returns to the analytical foundation that every other section has implicitly relied on:
The discipline of holding conclusions with the confidence they deserve rather than the confidence demanded by narrative convenience or institutional commitment.
Proud Certainty
Einstein’s injunction to ‘set aside your proud certainty’ is the animating principle of this book. It is applied to each of the domains examined: the proud certainty of the AI scaling optimists, the confident decouplers, the infrastructure investors, and the governance moratorium advocates. In each case, the analysis follows the same method: what does the evidence actually support, what are the foundational assumptions that the confident conclusion rests on, and what would need to be true for those assumptions to be wrong?
