A recent paper published by the Federal Reserve Bank of Cleveland (one of the 12 regional Federal Reserve Banks) indicates that research has found that lowering the price of AI is more conducive to unleashing its macroeconomic effects than simply increasing the weighting of AI usage:
1. Through modeling, the Fed found that when the weighting of AI usage doubles from approximately 2.4% to 4.8%, although AI capital* ultimately increases by 81%, the scarcity of existing AI capital forces the economy to rely more heavily on a still-insufficient factor of production, creating a bottleneck and resulting in a long-term output decrease of approximately 1.9%. Conversely, if the price of AI permanently decreases by 50%, the same resources can buy more effective AI capital, increasing AI capital by approximately 89% and improving long-term output by approximately 1%.
2. The paper also found that when "expanded AI usage" and "lower AI prices" occur simultaneously, they reinforce each other: lower prices alleviate the shortage of AI capital, while increased usage weighting amplifies the output contribution of newly added AI capital. However, in the baseline model, the output effect of both occurring simultaneously is still less than that of a price reduction alone. Therefore, reducing the deployment and capital costs of AI is a crucial prerequisite for expanding AI's economic contribution.
*Note: AI capital refers to the stock of AI-related hardware capital held by enterprises for the purpose of using AI and capable of continuously participating in production. The author treats it as an independent factor of production, similar to factories and machinery.
