AI Returns Stay in Big Tech: Where Are the Other 493 S&P Firms?

Investment in artificial intelligence is growing the technology sector, but the same picture is hard to find in the rest of the economy. According to an analysis shared on October 24 by Torsten Slok, chief economist at Apollo Global Management, revenue per employee and profit margins are rising at the seven large tech stocks known as the Magnificent Seven, while they are moving sideways at the other 493 companies in the S&P 500.

Slok also sees no clear evidence that AI is contributing to productivity gains. Despite heavy investment and big expectations, he wrote, the technology has delivered "zero return" in financial terms to corporate America outside tech.

Token bills are growing, returns are not visible

Running large models at scale is quickly becoming expensive. Many providers charge by the token, so the bill climbs as employees send more requests or as AI agents process data and take on more complex tasks. In Slok's view, traditional companies are starting to treat token spending less as a source of growth and more as a cost line.

Venture-backed startups are increasing their token spending, but established companies have little reason to pay for the newest models if cheaper alternatives do the job, and their preferences could shift that way. That poses a problem for AI providers trying to turn rising adoption into lasting revenue. If customers do not see a payoff in productivity, revenue or costs, they may not keep spending at the same pace. And if AI costs rise without any improvement in business performance, the extra outlay could put pressure on margins.

The limits Slok points to

Slok argues that AI does not grasp accountability, trust, corporate culture, unwritten rules, ethical choices or fairness, all of which shape reputational risk and, through it, revenue growth and profit margins. His conclusion: the AI story will be decided in the S&P 493, not in the Magnificent Seven, and so far there is no evidence of revenue or margin improvement outside tech.

The analysis also notes that adoption is concentrated in a relatively small group of companies. Unless usage spreads to more sectors and produces measurable results, the technology's impact on the broader economy could remain limited.

A fintech survey in India shows a similar picture

A similar tension is visible in India's financial technology sector. According to a PwC India survey reported in September, 84 percent of participating fintech firms have yet to see a measurable financial result from their AI investments. They are continuing to invest regardless: 48 percent named building proprietary intelligence as a top strategic priority. Thirty percent expect AI agent adoption to drive customer experience over the next three to five years, and 22 percent expect cost reductions and efficiency gains to lead.

Trust is also a major obstacle in financial services. Forty-one percent of respondents cited customer trust and adoption as the main challenge to scaling agent-based payments, and 33 percent pointed to fraud, security and risk management. Rather than answering individual prompts, agents take on multi-step tasks and can broaden AI's role in business, but their access to tools, information and company systems also brings added usage costs and security concerns. Nvidia has previously introduced the Open Agent Safety Platform, a hardware- and software-based effort aimed at some of these risks, meant to oversee agents and restrict harmful actions.

Companies begin to rein in usage

Rising token consumption has also pushed companies such as Microsoft, Meta and Amazon to limit AI usage, as bills grow with employee adoption and agent-based workloads. OpenAI CEO Sam Altman has described token costs as a "big problem" for customers who want a return on what they spend.

The infrastructure bill is heavy too. Amazon, Google, Meta and Microsoft have committed enormous sums to data centers, chips and the energy needed to run AI workloads. That spending feeds the revenue of technology providers, but on the customer side, among companies outside tech, Slok says the payoff is not yet visible.