🔍 Read the full analysis: The Numbers Tell A Cautious Story About AI’s Payoff on ThorstenMeyerAI.com
Get audio and creator gear 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 economist Torsten Slok says corporate America outside technology shows no visible financial payoff from AI in aggregate, based on trends in revenue per worker, margins and productivity. The finding does not prove AI has produced no gains at individual companies: targeted trials show improvements in some workflows, while spending on computing infrastructure and financing keeps rising.
Apollo Global Management chief economist Torsten Slok said an October 24 analysis found no visible aggregate financial payoff from artificial intelligence for U.S. companies outside technology, even as spending on data centers, computing and cloud capacity climbs. The conclusion is a reading of broad economic and company data, not proof that AI has produced no gains at individual businesses.
Apollo’s note, titled “No Signs of AI Payoff Outside the Tech Sector,” compares the seven largest technology companies with the other 493 firms in the S&P 500. Its charts show revenue per worker and profit margins rising for the seven, while the corresponding measures for the other companies have remained broadly flat. Apollo also cited productivity measures that, it said, show no clear boost from AI so far.
The gap is substantial in Apollo’s presentation: the seven technology companies were near $780,000 in inflation-adjusted revenue per employee in the third quarter of 2026, compared with about $525,000 for the other 493. The seven’s margins were around 30%, against roughly 10% for the rest. Those figures describe group-level trends; they do not isolate AI’s contribution from other business or economic factors.
The spending picture is clearer. 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, along with up to $14.72 billion in contingent cloud purchases. Together, the amounts are roughly $825 billion. The commitments cover data centers, energy and network infrastructure, servers and cloud capacity; they are not all exclusively for AI. Meta also reported third-quarter capital expenditures of $34.91 billion and narrowed its 2026 spending forecast to $135 billion-$145 billion.
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
Returns Lag Behind AI Infrastructure Bills
The contrast between large, measurable infrastructure costs and harder-to-identify returns matters to companies deciding how much to invest, and to investors assessing whether that spending can earn a return. If AI costs arrive before productivity or revenue gains, the investment can weigh on margins and cash flow. Meta reported $32.79 billion in operating cash flow in the third quarter, but capital expenditures were higher, leaving free cash flow at negative $2.12 billion, compared with positive $10.6 billion a year earlier.
That does not establish that Meta’s spending will fail to pay off. It does show why investors are watching the gap between current outlays and future benefits. Apollo’s Slok has argued that companies outside tech may increasingly treat spending on AI model use as a cost center rather than a growth driver. That is his interpretation of the available data, not a confirmed outcome for every company or use of AI.
Financing adds another pressure point. The source material reports that SpaceX borrowed $40 billion to buy AI chips, while credit protection for Meta and cloud provider CoreWeave became more expensive. These developments point to the scale of capital needed to expand computing capacity; they do not, by themselves, show whether the investments will produce adequate returns.
Products Worth Considering
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
Samsung 32-Inch Odyssey OLED G8 (G80SH) 4K Gaming Monitor
LG UltraGear 27-Inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDMI 2.1
As an affiliate, we earn on qualifying purchases.
The Data Behind Apollo’s Claim
Apollo’s conclusion draws on several kinds of evidence, including company financial trends, productivity measures and corporate spending data. Its analysis cites Ramp, a corporate card and spending platform with more than 85,000 U.S. business customers. In that dataset, 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 with OpenAI and Anthropic. The figures indicate concentrated spending among Ramp’s customers, not across every U.S. company.
Apollo also cited data showing that 80.4% of the top 1% of AI spenders are venture-backed, compared with 13.1% of median spenders. Slok has described such venture-funded spending as “circularity” and argued that compute costs may exceed the value delivered to users at the margin. Those are his assessments of the investment model, rather than independently established outcomes for all AI providers.
Evidence from specific workplace uses is more mixed than the broad averages suggest. The source material cites customer-support research finding 14% more issues resolved per hour, with a 34% increase for novice workers, and retail experiments in which sales effects ranged from no increase to 16.3%, depending on the workflow. It also reports that 83% of nearly 9,000 executives said AI had not affected productivity so far, with an average estimated productivity effect of 0.2% a year. These figures come from different studies and measures; they should not be treated as one comparable trial.
“No Signs of AI Payoff Outside the Tech Sector”
— Torsten Slok, Apollo chief economist, in the October 24 Daily Spark note
What Aggregate Numbers Cannot Show
A flat average for 493 S&P 500 companies cannot show whether AI is helping some firms while other results offset those gains. Apollo’s measures also cannot separate AI’s effects from factors such as interest rates, tariffs, labor costs or changes in demand. Its Ramp spending data cover Ramp customers, not the S&P 493 as a whole, and the note does not measure returns for each business function.
There is also no direct accounting figure in the cited evidence that attributes Meta’s commitments, cash flow or capital spending solely to AI. The company’s listed commitments include infrastructure and other purposes, including Reality Labs hardware. The source material does not establish when those commitments will turn into cash payments, how much revenue they will support, or what returns they may generate.
The productivity estimates and workplace examples use different populations, time periods and definitions of impact. The reported 0.2% annual average is an estimate from the cited executive survey, not a measured economy-wide AI productivity gain. It remains unclear how quickly isolated workflow improvements will spread into company-wide results.
Spending and Returns Under Watch
The next evidence will come from company results and filings showing whether investment in computing capacity is followed by sustained revenue, productivity or margin gains. Meta’s 2026 capital-spending forecast of $135 billion to $145 billion sets a near-term benchmark, while its future commitments show that infrastructure obligations extend well beyond this year.
Investors and executives will also watch financing costs and whether spending remains concentrated among a small group of firms. Apollo’s finding may change as more businesses adopt AI, but the cited analysis does not specify when a broad financial payoff should appear. For now, the central question is whether measurable gains from particular workflows can spread far enough to match the rising cost of AI infrastructure.
Key Questions
What did Apollo conclude about AI’s financial payoff?
Apollo chief economist Torsten Slok said broad trends show no visible aggregate financial payoff outside technology so far. That is Apollo’s interpretation of group-level data, not proof that no individual company or workflow has benefited.
Does the analysis show that AI has no benefits?
No. The analysis concerns aggregate trends and does not rule out gains at particular firms or in specific tasks. The source material cites examples including higher customer-support issue resolution and mixed retail sales results.
What does Meta’s roughly $825 billion figure include?
Meta’s filing lists future lease obligations, non-cancelable contractual commitments and contingent cloud purchases. The amounts largely relate to data centers, servers, cloud capacity and related infrastructure, but not all are exclusively for AI.
Why is AI spending concentrated among a small number of firms?
Apollo cited Ramp data indicating that a small share of its business customers accounts for most recorded spending on model serving, inference and neocloud services. The data describe Ramp’s customer base, not every U.S. business.
What would show whether AI is paying off more broadly?
Future company results and productivity data could show whether AI-related spending is followed by durable gains in revenue, output or margins. The timing and scale of any broader payoff remain uncertain.
Source: ThorstenMeyerAI.com
