Corporate America has spent billions of dollars on artificial intelligence tools. Yet outside of a few tech giants, those bets have produced zero financial return.
That blunt verdict comes from Apollo Global Management Chief Economist Torsten Slok. In an investor note released on Saturday, Slok pointed to a growing split in the US stock market. Sales per worker keep rising for the “Magnificent 7” tech giants (Amazon, Apple, Google, Meta, Microsoft, Nvidia, and Tesla). But worker output across the other S&P 493 firms has stayed flat.
“Despite massive investments and widespread hype around artificial intelligence, the technology has delivered zero financial payoff to corporate America outside of tech,” Slok wrote.
Mainstream businesses are now pulling back. Instead of seeing AI as a growth tool, corporate leaders see token costs and cloud bills as an expense that hurts profit margins.
Technical Factsheet: The Enterprise AI Productivity Gap
| Dimension / Metric | Magnificent 7 Tech Leaders | Mainstream Corporate America (S&P 493) |
|---|---|---|
| Worker Productivity | Sales per worker accelerating continuously. | Sales per worker remains completely flat. |
| Capex Allocation | Tens of billions quarterly (Meta spent $34.9B in Q3). | Pilot budgets freezing or getting folded into IT overhead. |
| Token Spending Status | Direct revenue driver via cloud and API sales. | Unproductive cost center weighing down operating margins. |
| Free Cash Flow Impact | Absorbing capex shocks (Meta reported -$2.1B FCF). | Demanding clear ROI before approving further inference spend. |
| Data Center Valuation | Sponsoring massive debt and equity funding rounds. | Rejecting inflated valuations (Firmus Grid $30B IPO pulled). |
| Agent Autonomy Level | Building autonomous coding and reasoning workflows. | 78% read-only restrictions, only 2% full autonomy. |
The Great Split: Tech Wins While the S&P 493 Stalls
The gap between tech sellers and corporate buyers is growing fast across Wall Street, as Bloomberg News highlighted:
- Tech Giants Spend Heavily: Meta spent $34.9 billion on capital expenses in the third quarter alone. That aggressive spending pushed its free cash flow to negative $2.1 billion. CEO Mark Zuckerberg pointed to 10 million users for Meta’s Muse model. Yet real commercial sales gains remain unproven.
- Investors Push Back: Markets are rejecting sky-high data center deals. Australian firm Firmus Grid dropped its $30 billion IPO after weak demand. Data center builder Nscale also faces pushback over its planned listing valuation.
The Read-Only Bottleneck: Why AI Has Not Paid Off
Why have AI investments stalled across non-tech industries? Research from PYMNTS Intelligence shows that companies block AI agents from doing real work.
In 2025, only 7% of executives tested AI agents, and 67% barred them from company systems. By 2026, 89% of firms used agents in at least one team.
Yet most companies keep those bots locked down:
- Read-Only Limits: 78% of firms let agents read records, but block them from taking actions.
- Human Approval: Another 18% require human sign-off before a bot can finish a task.
- Full Autonomy: Only 2% allow agents to complete daily work on their own.
“Most companies have let agents in the door but haven’t let them do much yet,” PYMNTS CEO Karen Webster wrote in her analysis of the agentic edge.
Without the power to act, AI bots work like costly search tools rather than real workers.
What Lies Ahead for Corporate AI Budgets
As earnings season begins, boards want to see real returns from AI spending. Companies that cannot show clear sales gains or labor savings are cutting back their software budgets.
Until firms trust AI agents to take real actions, software costs will remain a drag on margins rather than a profit driver.
TeckScaler Editorial Analysis
The enterprise AI productivity debate highlights a critical transition from experimental token consumption to demonstrable operational ROI. Follow TeckScaler for deep dives into AI engineering, agentic architecture, and enterprise technology economics.
As hardware capex rises without immediate enterprise returns, traders turn to GPU futures to hedge exposure against potential cloud compute slowdowns.
[…] margin squeeze mirrors broader global trends. Analysts at Apollo recently reported that enterprise AI yields low immediate payoffs outside a handful of dominant chip […]
[…] recent market drops, hardware vendors felt the first blow. Similarly, Apollo analysts warned that enterprise AI yields low immediate payoffs outside leading chip suppliers. Meanwhile, AI giants like Anthropic continue to lease billions in […]
[…] fees and are cutting back. The filing landed as Apollo chief economist Torsten Slok noted that AI has delivered zero financial payoff to corporate America outside Big […]
[…] startups launched tradable GPU futures to hedge compute risks. Economists at Apollo also noted that AI has produced zero financial payoff outside a few hardware suppliers, prompting funds to re-examine balance […]
[…] November IPO filing, labs face close financial scrutiny. Economists have also questioned corporate returns on AI spending outside Big Tech. Proving that models can solve deep science problems has become a top goal for AI […]