Gavekal-IS

Amathia is not simply ignorance, but a refusal to learn from new information and past mistakes, says Didier. History offers plenty of examples, but the same problem can be seen today in education, politics and our response to rapid technological change. As artificial intelligence reshapes the world, he contends that the ability to learn and adapt will matter more than ever.

Asia’s extraordinary rise over the past 25 years has transformed the distribution of global wealth and gold reserves, but currencies have yet to reflect that shift. Asian currencies, says Didier, remain structurally undervalued. When this long-delayed monetary adjustment finally occurs, a stronger Asia could deliver a substantial transfer of purchasing power from the West and mark a fundamental rebalancing of the global monetary order, he argues.

The cost of financial intermediation in the US has remained remarkably stable at around 2% a year for more than a century, despite huge productivity gains and sharply falling fund management fees. For Didier, the explanation may lie in Baumol’s cost disease, as rising coordination and compliance costs have absorbed much of the savings generated by technology. This matters particularly for long-term government bonds, whose real returns increasingly struggle to cover the cost of keeping the financial system running.

In the third part of his series on the macroeconomics of AI, Didier develops a framework for deciding when investors should favor information and technology over energy. After briefly tilting toward energy during this year’s confrontation in the Strait of Hormuz, the framework is once again signaling that investors should favor information.

In the second part of his series on the macroeconomics of AI, Didier argues that the economic impact of artificial intelligence will depend not just on technological progress, but on how widely it diffuses through the economy. Countries do not need to develop the world’s best AI models to reap the productivity gains and can instead win the race by diffusing the technology across their economies. This sets up an emerging contest between the US model of proprietary AI leadership and China’s push to spread cheap, open-source technology around the world.

The economic promise of artificial intelligence depends not only on expanding computing power, but on how effectively its gains spread into the wider economy. Didier frames growth as the interaction of energy, information and network effects, arguing that AI’s impact will ultimately be determined by the productivity it delivers beyond the data center. Even a modest transfer of AI-generated knowledge into the traditional economy could materially raise global productivity, although the adjustment in employment may prove disruptive.