Apollo Analysis Reveals S&P 493 Seeing Little Financial Return From AI
Apollo Global Management reports that outside the Magnificent Seven tech giants, mainstream corporate America has seen flat revenues, stagnant margins, and minimal productivity gains from AI.
Despite massive enterprise investments and persistent industry excitement around artificial intelligence, financial returns remain largely restricted to the tech sector's biggest operators. According to an evaluation released by Apollo Global Management Inc., the broader corporate landscape has yet to see measurable financial gains from AI initiatives.
Apollo analyst Torsten Slok evaluated performance metrics across the "S&P 493"—representing the S&P 500 index excluding Alphabet Inc., Amazon.com Inc., Apple Inc., Meta Platforms Inc., Microsoft Corp., Nvidia Corp., and Tesla Inc. His findings show that while the so-called Magnificent Seven have converted AI investments into substantial corporate gains, the remainder of the index has experienced flat profit margins and stagnant revenue per employee, showing no concrete sign that generative technology has raised general productivity.
Divergence Between Tech Giants and the S&P 493
To understand AI's real-world footprint, Apollo tracked three key indicators: employee productivity, profit margins, and revenue per worker. The data displays a sharp divide between the seven leading tech firms and the rest of corporate America.
At the Magnificent Seven, revenue per worker rose from an average of $550,000 in 2022 to nearly $800,000 this year. Conversely, that same metric across the S&P 493 has remained unchanged at roughly $550,000 since 2019. A parallel pattern emerges in profit margins, which accelerated sharply for the seven tech leaders starting in the third quarter of 2023 while staying essentially flat for the other 493 companies. Data measuring overall productivity gains similarly indicates little evidence that average enterprises are generating more output with existing resources, challenging predictions that AI would immediately usher in an "age of abundance."
Slok noted that while venture capital-funded startups continue pouring funds into model usage, established businesses are taking a more cautious stance. Mainstream enterprises are beginning to view token consumption not as an engine for expansion, but as an operational cost center that depresses operating margins. As Slok observed, mainstream corporate America does not require the latest model and is hesitant to pay premium rates for it.
The Limits of Automated Judgment
The fundamental barrier preventing AI from driving broad corporate gains lies in the nature of business decision-making. Slok explained that while AI models perform well at calculation, data synthesis, and content generation, commercial success heavily depends on human judgment. Judgment demands contextual awareness, where appropriate answers vary depending on who is evaluated.
Current models face major structural hurdles in handling corporate culture, ethical trade-offs, accountability, fairness, and trust. Because these factors directly influence reputational risk, businesses cannot easily substitute autonomous systems for human decision-makers without taking on significant liability. Slok emphasized that the macroeconomic trajectory of AI depends entirely on whether commoditized models can help typical enterprises achieve real margin expansion, noting that turning cheap intelligence into higher revenue remains far harder than anticipated.
What it means for developers
For software engineers and system architects, Apollo's findings signal a shift in how enterprises purchase and integrate AI software. As non-tech enterprises face scrutiny over software budgets, demand is moving away from expensive, bleeding-edge frontier deployments toward cost-conscious, specialized implementations that show immediate operational value.
Because mainstream organizations increasingly treat token expenses as a margin drag, developers must design applications with strict cost optimization in mind. Teams need to carefully balance model capability against token pricing rather than defaulting to the largest models on the market. Developers looking to compare infrastructure costs can try top AI models cheaply through one API at https://apixoai.online to evaluate performance trade-offs across different model families without enterprise lock-in.
Ultimately, engineering priorities will likely center on trust, deterministic guardrails, and contextual accuracy. Because commercial adoption hinges on addressing reputational risk and institutional judgment, developers who can build transparent, auditable workflows will be better positioned to help mainstream enterprises turn their AI investments into actual financial returns.
Source: Apollo analysis shows no one besides the biggest tech giants is seeing benefits from AI — SiliconANGLE AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
