Jeff Ding's analysis challenges the conventional narrative that technological dominance in hardware automatically translates to geopolitical and economic power. His book, introduced through O'Reilly Radar, examines Japan's 1980s semiconductor supremacy, a period when Japanese companies controlled global markets for chips, consumer electronics, and computing hardware. Yet despite this overwhelming hardware advantage, Japan failed to lead the broader information revolution that reshaped the global economy.
Ding, a political scientist at George Washington University, argues that the difference lay not in heroic breakthroughs but in how ordinary engineers across different contexts solved problems. Japan optimized manufacturing and hardware engineering, but the subsequent wave of value creation moved to software, services, and internet platforms where different engineering cultures and organizational structures proved decisive.
The lesson applies directly to contemporary AI debates. Observers obsess over which nation builds the biggest model or fastest chip, assuming hardware dominance guarantees economic leadership. Ding's historical lens suggests otherwise. The engineers who matter are those embedded in ecosystems that can rapidly iterate, experiment, and commercialize. These are often unglamorous teams working on integration, deployment, and adaptation rather than foundational breakthroughs.
This reframes how we should evaluate AI competition. Building an advanced language model remains important, but it tells us little about which country or company will actually capture economic value. That depends on which engineers can build better applications, integrate AI into existing workflows, and solve specific problems at scale. Russia and China might develop capable models. The question is whether their engineers can build the ecosystems that turn models into profitable services.
Japan's experience suggests that leading in components or hardware creates advantages but guarantees nothing. Today's version of that mistake would be assuming that chip manufacturing or model weights determine AI's future. Instead, the competitive advantage flows to teams of ordinary, capable engineers distributed across real organizations solving real problems faster than competitors.