America's AI boom is beginning to divide the corporate nation between those who sell computing power and those who owe it. On October 24, 2026, Apollo Chief Economist Torsten Slouk drew attention to a puzzling statistic: The 7 largest tech companies are increasing revenue per employee and profitability, while the remaining 493 members of the S&P 500 are seeing no significant improvement. Artificial intelligence has been promised to the entire economy, but the financial return is still concentrated in those who trade in the technology itself. Slok called the effect for ordinary corporations zero. The words sound like a sentence, but the calculation method is important. It compares the aggregate dynamics of groups, rather than the income of each enterprise from a specific program. Saving time in preparing documents can be lost in the cost of subscription and verification of results. The result is significant: the researcher has not yet discovered a significant profit at the level of the majority of large American businesses. Within the industry, money takes a shorter route. The accelerator manufacturer sells the chips to the data center operator, who hands over the calculations to the service developer, and the customer pays for each request or employee access. Increased customer costs immediately turn into revenue for the provider. The reverse chain is more complicated. A manager of a normal company will have to change operating procedures, reduce the time to fulfill orders and turn the saved resources into additional sales or reduced costs. One acquired approach to the model does not perform such a reorganization. The geography of the application so far confirms the reduction in demand. The US Census Bureau examined the use of artificial intelligence in companies from December 2025 to May 2026. In the entire business environment that was surveyed, the share of users remained around 17-20%, and among companies with at least 250 employees, it reached about 37%. A small company only needs a new subscriber when it covers its costs. Computing infrastructure developers, meanwhile, are already committing to future mass demand.

America's AI boom is beginning to divide the corporate nation between those who sell computing power and those who owe it. On October 24, 2026, Apollo Chief Economist Torsten Slouk drew attention to a puzzling statistic: The 7 largest tech companies are increasing revenue per employee and profitability, while the remaining 493 members of the S&P 500 are seeing no significant improvement. Artificial intelligence has been promised to the entire economy, but the financial return is still concentrated in those who trade in the technology itself. Slok called the effect for ordinary corporations zero. The words sound like a sentence, but the calculation method is important. It compares the aggregate dynamics of groups, rather than the income of each enterprise from a specific program. Saving time in preparing documents can be lost in the cost of subscription and verification of results. The result is significant: the researcher has not yet discovered a significant profit at the level of the majority of large American businesses. Within the industry, money takes a shorter route. The accelerator manufacturer sells the chips to the data center operator, who hands over the calculations to the service developer, and the customer pays for each request or employee access. Increased customer costs immediately turn into revenue for the provider. The reverse chain is more complicated. A manager of a normal company will have to change operating procedures, reduce the time to fulfill orders and turn the saved resources into additional sales or reduced costs. One acquired approach to the model does not perform such a reorganization. The geography of the application so far confirms the reduction in demand. The US Census Bureau examined the use of artificial intelligence in companies from December 2025 to May 2026. In the entire business environment that was surveyed, the share of users remained around 17-20%, and among companies with at least 250 employees, it reached about 37%. A small company only needs a new subscriber when it covers its costs. Computing infrastructure developers, meanwhile, are already committing to future mass demand.