🔍 Read the full analysis: A Guide To The Numbers On AI Spending And Economic Returns on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Apollo chief economist Torsten Slok says aggregate data show no clear financial payoff from AI outside large technology companies, while spending on data centers, chips and cloud capacity continues to rise. The finding describes broad trends, not proof that AI has produced no returns: some task-level studies report measurable gains, and the causes of company-wide changes remain difficult to isolate.
Apollo chief economist Torsten Slok says recent company and productivity data show no clear aggregate financial payoff from AI outside major technology firms, even as companies make enormous commitments to computing infrastructure. His October 24 note highlights a widening gap between measurable spending and returns that are harder to identify across the broader corporate economy.
Apollo’s analysis compares the Magnificent Seven technology companies with the other 493 S&P 500 firms. Its charts show inflation-adjusted revenue per worker rising for the seven and remaining broadly flat for the rest; it also shows profit margins climbing to about 30% for the seven while staying near 10% for the other companies. Those are company-level trends, not a direct measurement of how much AI caused them.
The spending side is clearer in corporate disclosures. Meta reported $34.91 billion in capital expenditures in its third quarter and narrowed its full-year 2026 capital spending forecast to $135 billion to $145 billion. Its filing lists roughly $355.73 billion in leases not yet started and $454.52 billion in non-cancelable contractual commitments, along with up to $14.72 billion in contingent cloud purchases. The combined figure is about $825 billion in future commitments, mostly tied to data centers, cloud capacity, servers and networks; the source notes that not all of it is AI-related.
Evidence of direct gains is more specific. The source cites a customer-support study finding 14% more issues resolved per hour, with a 34% increase for novice workers, and retail experiments reporting sales effects ranging from zero to 16.3%, depending on the workflow. By contrast, a survey of nearly 9,000 executives found 83% reported no productivity impact so far, while the reported average gain was 0.2% a year. These figures describe different studies and measures, not one unified estimate of AI’s economic return.
Infographic 01 · The “zero” in numbers
What “zero financial payoff” actually measures.
Apollo’s headline rests on five charts. Here are their numbers, and what they can and cannot prove.
Who pays for AI? Share of spending by the top 10% of firms
Ramp customer data via Apollo · non-AI categories in grey for comparison
Spending is concentrated in every category, not only AI; AI is the most extreme.
What the five charts do and do not show
shown not shown
- Shown: revenue per worker and margins rising for the 7 tech giants, flat for the other 493
- Shown: no AI lift yet in US productivity statistics (San Francisco Fed TFP, output per hour)
- Shown: AI spending sits in few, mostly venture-funded firms (Ramp, 85,000+ US businesses)
- Not shown: that AI caused the gap; rates, tariffs and wages move the same lines
- Not shown: gains inside single functions, which can cancel out in an average
- Not shown: the S&P 493 itself in the Ramp data; those are Ramp’s customers
“The AI story will be decided in the S&P 493, not the Mag 7.”Torsten Slok, Apollo chief economist, 24 Oct 2026 (opinion)
Sources: Apollo Daily Spark (24 Oct 2026) with charts by Bloomberg, Macrobond, SF Fed, BLS and Ramp; Apollo (20 Oct 2026); Ramp AI Index. Values with ≈ are read off Apollo’s charts. Assessment: Thorsten Meyer.
Powered by Thorsten Meyer AI
The Gap Between AI Bills and Returns
The issue matters because companies, investors and workers are trying to judge whether large AI infrastructure costs will be matched by durable gains. Capital spending and contractual commitments are visible now; economy-wide productivity effects are slower to measure and may vary sharply by task, industry and company. That mismatch can make it difficult to tell whether firms are investing ahead of future demand or paying for capacity that will not generate sufficient returns.
Meta’s figures illustrate the scale without proving that the company’s AI spending has failed. Its third-quarter operating cash flow was $32.79 billion and capital expenditures were $34.91 billion, yielding negative free cash flow of $2.12 billion, compared with positive $10.6 billion a year earlier. The company also reported quarterly revenue of $64.60 billion, up 26%. Those figures cover the business as a whole; they do not identify the contribution of AI investments separately.
For readers, the practical takeaway is that claims about AI’s economic impact depend on what is being measured. A productivity gain in one support team does not establish broad gains across a company, while flat averages do not rule out substantial benefits for particular users. The evidence supports uneven adoption and uncertain aggregate returns, not a simple verdict that AI either pays off everywhere or nowhere.
Products Worth Considering
Apple MacBook Pro 16-inch Laptop with M5 Pro, 18-Core CPU, 20-Core GPU, 48GB Memory, 2TB SSD, Space Black
Apple MacBook Pro 14-inch with M5 Max Chip (18-Core CPU, 20-Core GPU), 48GB RAM, 2TB SSD, Space Black
Apple MacBook Pro 14.2-inch with M5 Pro, 15-core CPU, 16-core GPU, 48GB Memory, 1TB SSD, Silver
As an affiliate, we earn on qualifying purchases.
What Apollo’s Charts Measure
Apollo’s October 24 “Daily Spark,” titled “No Signs of AI Payoff Outside the Tech Sector,” brought together company revenue and margin trends, productivity measures and data on business AI spending. The note cites San Francisco Federal Reserve measures of utilization-adjusted total factor productivity and output per hour, which it says show no visible AI-driven productivity acceleration. Such national and sector-level indicators can lag changes at individual companies and cannot, on their own, establish why productivity moved.
The note also uses spending data from Ramp, a corporate card and spending platform with more than 85,000 U.S. business customers. Ramp’s figures show that the top 10% of firms account for 99.5% of spending on model serving and inference, 99% of neocloud spending and 95.6% of spending on OpenAI and Anthropic. These are Ramp customer data, not a census of American companies. The source further reports that 80.4% of the top 1% of AI spenders are venture-backed, compared with 13.1% of median spenders.
Slok’s “zero financial payoff” wording is an interpretation of these aggregate patterns. A flat result across hundreds of companies may conceal gains and losses that offset each other, and the charts do not isolate AI from factors such as labor costs, interest rates or tariffs. The supporting evidence is stronger for the narrower statement that a broad payoff is not yet visible in these measures.
“AI has delivered zero financial payoff to corporate America outside of tech.”
— Torsten Slok, Apollo chief economist, in the October 24 Daily Spark note
What the Return Figures Cannot Show
The figures do not establish how much of the technology giants’ rising revenue per worker or margins comes from AI rather than their other businesses or broader market conditions. Nor does the absence of a clear economy-wide productivity signal establish that AI has had no effect: national indicators may not capture early or concentrated gains, and company averages can obscure different outcomes across teams.
Meta’s roughly $825 billion in commitments also should not be read as an AI-only bill or as cash already spent. The source says the total includes leases and contractual obligations extending into the future, and some spending relates to non-AI activity. Forecasts for capital expenditures by major technology firms are likewise projections, not completed spending. It remains unclear how much infrastructure will be used, how quickly it may produce revenue, and whether returns will cover operating and financing costs.
The source material also points to a changing financing environment, including a reported $40 billion borrowing by SpaceX to buy AI chips and higher insurance costs for some technology-related debt. Those examples suggest scrutiny of funding needs, but they do not by themselves establish a sector-wide credit crisis. Future demand, financing terms and returns remain uncertain.
Evidence to Watch in Coming Results
The next useful evidence will come from company earnings, spending disclosures and productivity data that show whether infrastructure commitments are turning into revenue, lower costs or measurable output gains. Investors and analysts will also watch whether companies maintain or revise capital spending plans as construction and computing needs grow. Reported returns should be distinguished from forecasts, and from broad claims that do not specify a time period or comparison baseline.
For Meta, later filings and earnings reports can clarify how its commitments change and how its capital spending affects cash flow. Across the wider economy, repeated measurements of productivity and company-level results will be needed to determine whether today’s task-specific gains spread beyond early adopters. Until those measures provide a clearer picture, the central finding remains limited: the investment is substantial, but broad economic returns are not yet evident in the cited data.
Key Questions
What did Apollo conclude about AI’s economic payoff?
Apollo chief economist Torsten Slok said aggregate data show no clear financial payoff outside major technology firms. The conclusion describes broad trends and does not prove that every non-tech company has received no benefit.
What does Meta’s $825 billion figure represent?
It combines about $355.73 billion in leases not yet started, $454.52 billion in non-cancelable contractual commitments and up to $14.72 billion in contingent cloud purchases. The commitments extend into future years, and the source says not all are related to AI.
Are there measured examples of AI improving work?
Yes. The source cites a customer-support study reporting 14% more issues resolved per hour, with a larger gain for novice workers, and retail experiments with sales effects ranging from zero to 16.3%. Results vary by task and study.
Does the evidence prove AI has no productivity effect?
No. The cited aggregate measures show no broad boost yet, but they cannot isolate AI’s effect from other economic forces or rule out gains at individual companies and in specific tasks.
Source: ThorstenMeyerAI.com
