AI Has Delivered Zero Profit for Most Companies, Apollo Economist Warns

TL;DR

• Apollo chief economist Torsten Slok says AI has produced virtually no profit uplift for corporate America outside Big Tech, despite hundreds of billions in spending on chips, cloud, and models.
• S&P 493 profit margins and productivity data show little to no AI-driven lift yet, with gains concentrated in Nvidia, Microsoft, Alphabet, Meta, and Amazon while most companies face high costs and unclear ROI.
• For investors in 2026, Slok warns the AI trade hinges on when — or if — massive capex turns into earnings, raising the risk of a pullback in AI spending and a market repricing.

The Trillion-Dollar Bet With No Payoff Yet

Wall Street has heard Torsten Slok's warning loud and clear: Corporate America has spent a fortune on artificial intelligence, and for most companies, it has so far delivered nothing to the bottom line.

Slok, the chief economist at Apollo Global Management, argues that despite two years of historic investment in AI infrastructure, data centers, chips, and enterprise software, there is little evidence of a financial return outside of a tiny cluster of Big Tech winners. The hype is massive. The profits, for almost everyone else, are missing.

In a series of recent notes and interviews that have ricocheted across markets, Slok has challenged the core narrative driving the S&P 500 to record highs — that AI will lift productivity and profits across the entire economy.

The S&P 493 vs. The Magnificent Few

At the heart of Slok's argument is a simple breakdown of earnings.

Since the launch of ChatGPT in late 2022 unleashed the current AI boom, almost all of the earnings growth in the S&P 500 has come from the mega-cap tech companies at the center of the AI buildout. Nvidia selling the picks and shovels, Microsoft and Amazon selling cloud capacity, Meta and Alphabet monetizing AI-enhanced advertising — these are the clear winners.

Strip them out, and the picture looks starkly different. Slok points to data showing the other 493 companies in the index have seen flat to minimal earnings growth and little margin expansion attributable to AI. In other words, AI has been a revenue driver for the sellers of AI, not the buyers of AI.

This is why he describes the payoff for corporate America ex-Big Tech as essentially zero. Companies are paying for Microsoft Copilots, OpenAI enterprise licenses, cloud GPU time, and internal AI projects, but those costs are not yet showing up as higher sales per worker or fatter profit margins.

Why Productivity Gains Haven't Shown Up

If AI is so transformative, why isn't it in the numbers? Slok and other economists point to several reasons.

First, adoption takes time. Like electricity and the internet before it, AI requires companies to reorganize workflows, retrain staff, and rebuild software — a multi-year process. Most firms are still in the experimentation phase, running pilots that cost money but don't scale.

Second, the costs are front-loaded. AI data centers, Nvidia Blackwell chips, and power contracts are enormously expensive. For non-tech companies, AI is currently a cost center: software fees, consulting bills, and cloud usage charges hit immediately, while efficiency gains, if they come, arrive much later.

Third, productivity statistics remain muted. U.S. labor productivity has improved from its 2022 lows, but there has been no dramatic AI-driven surge. Slok notes that broad measures of corporate profits as a share of GDP have not broken out in a way that would suggest an AI revolution is underway for the average firm.

Finally, much of AI's impact so far is defensive. Companies say they must invest in AI just to keep up with competitors, not because it will immediately boost profits. That keeps spending high even when ROI is unclear.

What It Means For AI Spending In 2026

Slok's warning lands at a critical moment. Big Tech is on track to spend well over $300 billion combined on capital expenditures in 2025, with 2026 forecasts climbing even higher as hyperscalers race to build next-generation AI data centers.

Investors have largely cheered that spending as long as cloud revenue and Nvidia orders keep growing. But Slok warns the patience may not last forever. If the S&P 493 cannot start to show tangible earnings from AI by 2026, markets will start asking harder questions about whether this capex boom is sustainable.

That doesn't mean Apollo is predicting an imminent AI crash. Slok has acknowledged that Big Tech can fund this buildout from massive free cash flow, and that AI could still deliver broad productivity gains over the next 3-5 years. His point is about timing and expectations: stocks are priced for an AI profit boom that has not yet arrived for most of the market.

For investors, that creates a two-tier market. The AI infrastructure trade — chips, power, cloud — continues to work. The AI adoption trade — buying non-tech stocks in hopes AI will supercharge their margins — remains largely unproven.

Wall Street Pushback: Is It Too Early To Judge?

Not everyone agrees with Slok's stark assessment. Tech bulls argue that calling ROI zero in 2025-2026 is like judging the internet's profits in 1998.

They point to early signs of payoff: coding assistants lifting software developer output by 20-30%, AI customer service agents cutting call center costs, pharmaceutical companies accelerating drug discovery, and banks using AI for fraud detection and efficiency. Companies like Walmart, JPMorgan, and Delta have all cited AI-driven cost savings on recent earnings calls.

Economists at Goldman Sachs and the Federal Reserve have also noted that AI may already be contributing modestly to productivity, but that it is being masked by inflation, high interest rates, and uneven adoption.

Slok's counter is simple: anecdotes are not earnings. Until AI shows up in aggregate profit margins and sustained productivity growth, investors should treat broad AI profit promises with skepticism.

The Bottom Line For Investors

The debate over AI ROI will define markets in 2026. If Slok is right that most companies are still getting zero financial return, then either corporate AI spending must slow, or stocks must adjust to lower long-term profit expectations.

If he is wrong, and 2026 is the year pilots turn into profits, then the laggards of the S&P 493 could be the next big winners as AI finally broadens out beyond Big Tech.

For now, his message to investors is caution: don't assume AI hype equals AI profits. Follow the earnings, not just the spending — because so far, only Big Tech is cashing in.

AndroGuider Team

Articles written by the AndroGuider team. We try to make them thorough and informational while being easy to read.