The Numbers Behind AI’s Limited Gains Across Industries

🔍 Read the full analysis: The Numbers Behind AI’s Limited Gains Across Industries on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get the latest gadgets delivered free with Prime

• Fast, free delivery on millions of items
• Prime Video, Amazon Music and more included
• Member-only deals all year

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

Apollo chief economist Torsten Slok said in an October 24 research note that available productivity and company financial data show no visible AI payoff outside the technology sector. The conclusion comes as major companies commit heavily to data centers, servers and cloud capacity, sharpening the question of whether the spending will translate into broad business gains.

Apollo’s note, titled “No Signs of AI Payoff Outside the Tech Sector,” compares the seven largest US technology companies—Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla—with the other 493 companies in the S&P 500. Apollo’s charts show inflation-adjusted revenue per worker rising among the seven and remaining broadly flat among the other companies. It also shows profit margins for the seven at about 30%, compared with roughly 10% for the rest.

The note points to productivity measures from the San Francisco Federal Reserve, which Apollo says show no signs of AI lifting productivity growth so far. It also cites data from Ramp, a corporate spending platform whose customers include more than 85,000 US businesses. In those data, the top 10% of firms account for nearly all reported spending on model inference and neocloud services. Ramp’s customer data are not a measure of the entire US economy, and Apollo’s broad company comparisons do not isolate AI as the cause of the observed trends.

The investment figures show the scale of the buildout. Meta’s October 22 quarterly filing lists about $355.73 billion in leases not yet started and $454.52 billion in non-cancelable contractual commitments, largely for cloud capacity, servers and network equipment, plus up to $14.72 billion in contingent cloud purchases. The combined figure is about $825 billion; it includes some commitments not specifically tied to AI. Meta reported third-quarter capital expenditures of $34.91 billion and said it expects 2026 capital spending of $135 billion to $145 billion.

The Gap Between AI Bills and Returns

The figures highlight a widening gap between measurable investment and returns that are difficult to identify across much of corporate America. If AI infrastructure costs rise faster than the revenue or productivity gains businesses can capture, companies may face pressure on margins and investors may question the pace of spending. The source reports that insurance costs for debt associated with Meta and CoreWeave have increased, adding a financing concern to the debate.

But an economy-wide or company-wide average can conceal differences among firms and job functions. Apollo’s comparison does not establish that AI caused the flat results among the 493 companies, nor does it rule out gains in particular operations. The distinction matters to workers and customers as well as investors: the evidence may look weak across a broad corporate measure while specific teams use AI to handle tasks faster or increase sales.

The source cites research reporting 14% more customer-support issues resolved per hour, with a larger reported gain for novice workers, and retail experiments ranging from no sales increase to a 16.3% increase depending on the workflow. Those results are task-specific and do not establish a typical return for all firms. They suggest that the business case may depend on how AI is integrated, what task it is applied to and how outcomes are measured.

How Apollo Built Its Case

Apollo’s October 24 note drew attention to a question already facing companies that are buying computing capacity and AI services: where are the returns? Its central evidence is a set of comparisons among large listed companies, productivity indicators and spending patterns. Slok’s description of AI as having “delivered zero financial payoff” is his interpretation of those indicators, not a direct measurement showing that AI generated no value in every company.

The note also describes AI-related spending as concentrated among a small share of businesses. According to the Ramp figures it cites, 80.4% of the top 1% of AI spenders are venture-backed, compared with 13.1% of the median spender. This points to a difference in who is paying for AI services, but the data cover Ramp’s customers and should not be treated as a complete census of business spending.

Meta’s reported commitments provide one example of how much future infrastructure can sit behind the current expansion. Its filing covers leases and contracts extending over several years, not money spent all at once. The reported total should not be read as an $825 billion AI-only bill: the source says some commitments also cover Reality Labs hardware and other capacity.

“No Signs of AI Payoff Outside the Tech Sector”

— Torsten Slok, Apollo chief economist, in the October 24 Daily Spark note

What the Averages Cannot Show

Apollo’s figures do not establish how much of the difference between the seven technology giants and other S&P 500 companies is caused by AI. The groups differ in business models and other conditions, and company-level averages can conceal gains and losses among individual firms. Productivity indicators also may not capture every effect immediately. The source gives no way to separate AI’s contribution from factors such as labor costs, interest rates or tariffs.

There is also uncertainty about how representative the spending and task-level evidence is. Ramp’s figures describe its own business customers, while the support and retail results cited in the source depend on particular tasks and workflows. The material does not provide enough detail to compare those studies on a common basis or establish a typical return on investment. Nor does the $825 billion estimate represent an AI-only payment due at once: it combines commitments over time and includes some non-AI items.

The Next Test Is Measurable Returns

The key test will be whether subsequent company filings and productivity data show gains beyond the firms already spending heavily on AI. For Meta, its future capital expenditures and the timing of its contractual commitments will help clarify how the infrastructure buildout develops. The source also notes that forecasts for spending by major technology companies are not the same as binding commitments, so actual spending and financing conditions remain important to track.

For businesses adopting AI, the evidence points to a need to measure specific outcomes—such as time per task, accuracy, revenue or operating costs—rather than assume that greater model usage means higher productivity. Apollo’s headline conclusion may change as those investments mature, but the material available as of October 26, 2026, does not show a broad, clearly attributable financial payoff across non-tech companies.

Key Questions

What does Apollo mean by “zero financial payoff”?

It refers to Apollo’s reading of aggregate trends: revenue per worker and profit margins rose among seven major technology companies but were broadly flat among the other 493 S&P 500 companies, while cited productivity measures showed no AI boost. It does not prove that every non-tech company or AI use case has zero benefit.

Does the $825 billion figure mean Meta has spent that much on AI?

No. The figure combines future leases, contractual commitments and contingent cloud purchases. The commitments extend over time, and the source says some are not specifically for AI, including items related to Reality Labs hardware.

Are companies seeing any measurable AI gains?

The source cites task-level studies reporting gains in customer support and some retail experiments, but results vary by workflow. Those examples do not establish a typical return across industries.

How reliable are the spending figures in Apollo’s note?

The figures come from different sources and have different coverage. Meta’s commitments are disclosed in a company filing; the concentration figures come from Ramp’s customer spending data, not a census of all businesses. Their scope should be kept distinct.

Source: ThorstenMeyerAI.com