The Signal Problem: What the Boom Is Actually Telling Us
The investment cycle and the capability-to-capacity gap are two parts of the same AI transition. One shows where expectations have already become spending, financing and market value; the other shows why that spending may or may not become durable productivity. This checkpoint connects those mechanisms into a practical judgment: classify an AI-related economic signal by its channel, timing, distribution and evidence strength before deciding what it says about the future.
The payoff is not a better slogan about whether AI is “good” or “bad” for growth. It is a way to read mixed evidence without forcing it into a single story. Investment can be real and still be too early. Productivity can be promising and still be delayed. Valuations can support demand and still create fragility. These statements can coexist because they describe different stages of the same transition.
Four clocks moving through one boom
The first clock is the clock of commitment. Firms order chips, build data centers, borrow money and purchase infrastructure because they expect future demand. Trade in high-AI-content goods and the valuations of AI-linked firms respond to those commitments. These are leading signals in the limited sense that they reveal what firms and markets are preparing for. They are not direct measures of the output that the preparation will eventually produce.
The second clock is the clock of installation and adoption. Equipment must be connected to power, software and networks; organizations must redesign workflows; workers must learn how responsibilities change; customers and regulators must accept the new arrangement. This clock can be slowed by energy and grid constraints, skills gaps, compliance costs, managerial risk aversion and mistrust. Falling inference prices may make use cheaper while leaving the adoption process intact.
The third clock is the clock of realized productivity. Output per input may improve when tasks are automated, assisted or reorganized, but aggregate measures can lag behind early gains. Resources have to move between firms and sectors, and complementary investments have to work. The BIS's median estimate of about 0.5 percentage points of annual total-factor-productivity growth is therefore an estimate of a possible macro payoff, not a contemporaneous reading of the investment cycle. Evidence that a worker completes one task faster is closer to the mechanism than evidence that the entire economy has raised potential output.
The fourth clock is the clock of adjustment. Labor income, tax receipts, household demand, asset values and financial exposures respond to the transition at their own speeds. A rapid rise in investment may support suppliers immediately, while job displacement or the erosion of a labor tax base appears later. A valuation correction may arrive before productivity data clarify whether the long-run technology is useful. This is why a transition can be stressful even if its eventual steady state is richer.
The clocks interact rather than line up neatly. A country can be far along the commitment clock and early on the adoption clock. It can show strong market valuations while its measured productivity remains unchanged.