← AI Native Landscape 9 min read

Fortune 500 Earnings Calls: Everyone Talks About AI, Almost Nobody Quantifies It

Executive Summary

  • 68% of S&P 500 companies (331 of 485) cited “AI” on Q4 2025 earnings calls — the highest in a decade, up from 86 as the 10-year average. The conversation has gone from IT departments to boardrooms. (FactSet, March 2026)
  • Only 10% of companies that discussed AI quantified its impact on specific use cases. Only 1% quantified impact on earnings. The gap between AI enthusiasm and measurable results is the dominant story of this earnings season. (Goldman Sachs Economic Research, March 2026)
  • Companies that mentioned AI on Q4 calls underperformed those that did not — gaining 1.5% vs. 5.6% from Dec 31, 2025 through March 10, 2026. The market is done rewarding AI talk; it now punishes unsubstantiated claims. (FactSet, March 2026)
  • A landmark NBER study of 6,000 executives across four countries found nearly 90% report zero impact from AI on employment or productivity over the past three years. The Solow Paradox — “you can see the computer age everywhere but in the productivity statistics” — is back. (NBER, February 2026)
  • The two areas where AI is actually working: software development and customer service, both showing ~30% productivity gains among companies that measured. Everything else remains aspirational. (Goldman Sachs, March 2026)

The Numbers: A Decade of Acceleration

FactSet has tracked AI mentions across S&P 500 earnings calls since 2015. The trajectory:

  • 10-year average (2015-2025): 94 calls per quarter
  • 5-year average (2020-2025): 149 calls per quarter
  • Q2 2025: 292 calls (then-record)
  • Q3 2025: 306 calls
  • Q4 2025: 331 calls — 68% of all S&P 500 earnings calls

The term “AI” has appeared on more S&P 500 earnings calls than “inflation,” “recession,” or “interest rates” for the first time in financial history. “Uncertainty” mentions declined from 415 in Q1 2025 to 201 in Q3 2025 — executives are replacing anxiety language with AI language. Whether the substance has changed is another question.

Sector Breakdown: Financials Surge, Tech Plateaus

Q4 2025 sector percentages of calls citing AI:

Sector % Citing AI # of Calls Change vs. Q3
Information Technology 94% 67 Stable
Financials 91% 67 +13 calls
Communication Services 89% — Stable
Health Care — — +8 calls

The real story is Financials. The sector added 13 AI-citing calls quarter-over-quarter — the largest increase of any sector. Health Care added 8. Technology was already saturated at 94-95%. The AI conversation is migrating from builders to buyers, which is where it matters for most companies.

What CEOs Actually Said: The Specificity Spectrum

The quality of AI claims on earnings calls varies enormously. Goldman Sachs senior economist Ronnie Walker analyzed Q4 earnings season and found a clear hierarchy:

Tier 1: Quantified Impact (10% of companies)

These companies attached numbers to AI outcomes:

  • Meta — CFO Susan Li: engineer output up 30%, “power users” up 80% YoY. No methodology disclosed for how “output” was measured. Simultaneously announced $115-135B in 2026 capex, nearly double 2025. (Q4 2025 earnings call, January 2026)
  • Google — CEO Sundar Pichai: 30%+ of new code AI-generated, up from 25% six months prior. 750M monthly Gemini App users. Cloud revenue up 48% YoY to $17.7B in Q4. (Q4 2025 earnings call, February 2026)
  • Walmart — CEO Doug McMillon: 40%+ of new code AI-generated or AI-assisted. Rolling out ChatGPT Enterprise licenses and OpenAI certifications across workforce. (Q3 FY2026 earnings call, November 2025)
  • JPMorgan — Marianne Lake (Consumer Banking CEO): productivity in AI-deployed areas rose to 6%, up from 3% pre-deployment. Operations roles could “eventually” see 40-50% gains. 200,000+ employees have access to internal GenAI platform. (Goldman Sachs conference, December 2025)
  • General Mills — CFO Kofi Bruce: AI models analyze 5,000+ daily shipments, saving $20M+ since FY2024. Predicts $50M in manufacturing waste reduction. (Earnings call, 2025)
  • Goldman Sachs — CFO Denis Coleman: Devin (AI coding agent) delivering 3-4x productivity in software development. OneGS 3.0 positions AI as foundational operating capability. (Q4 2025 earnings call)

Tier 2: Directional Claims (54% of companies)

These companies discussed AI in productivity and efficiency terms but provided no metrics:

  • Bank of America — CEO Brian Moynihan: $4B of $13B technology budget allocated to AI and related initiatives. Erica virtual assistant absorbs high-volume service interactions. No quantified productivity gains disclosed. (Earnings call, 2025)
  • CVS Health — AI deployed in call centers, pharmacy workflows, and clinical operations. CFO cited “mid-teens adjusted EPS growth” supported by “technology-enabled productivity gains.” No AI-specific figures. (Q4 2025)
  • Thermo Fisher Scientific — CEO expressed optimism about AI in wet lab research. $300M in cost reductions using PPI Business System, which “incorporates AI for process improvements.” AI contribution not isolated. (Q4 2025)

Tier 3: Name-Drop Only (36% of AI-citing companies)

These companies mentioned AI in passing — investor expectation management without substance. FactSet’s analysis shows this is the largest group.

The Productivity Paradox Returns

The most important finding this earnings season comes not from any single company but from a February 2026 NBER study surveying 6,000 CEOs, CFOs, and executives across the U.S., U.K., Germany, and Australia:

  • ~66% of executives reported using AI
  • Average weekly usage: approximately 1.5 hours
  • 25% reported no workplace AI usage at all
  • Nearly 90% reported zero impact on employment or productivity over the past three years

Executives forecast just 1.4% productivity increase from AI over the next three years and a 0.8% output increase — a far cry from the transformation rhetoric on earnings calls.

Goldman Sachs’ Walker concluded there is “no meaningful relationship between productivity and AI adoption at the economy-wide level.” The two exceptions: customer support and software development, both showing median productivity boosts of approximately 30% among companies that actually measured.

This is Robert Solow’s 1987 paradox updated for 2026: “You can see the computer age everywhere but in the productivity statistics.” Despite $250B+ in corporate AI investment in 2024 alone, macroeconomic data shows minimal AI impact outside two specific functions.

The Market’s Verdict: AI Talk Now Hurts Stock Performance

FactSet tracks stock performance of AI-citing companies vs. non-citing companies. The correlation has flipped:

Q3 2025 (through Dec 4, 2025):

  • AI-citing companies: +13.9% since Dec 31, 2024
  • Non-citing companies: +5.7%
  • AI premium: +8.2 percentage points

Q4 2025 (through Mar 10, 2026):

  • AI-citing companies: +1.5% since Dec 31, 2025
  • Non-citing companies: +5.6%
  • AI penalty: -4.1 percentage points

The market rewarded AI talk through 2025. In 2026, it reversed. Investors are now separating companies that deliver AI results from those that merely discuss them. This is a significant signal for C-suite executives still relying on AI as a narrative crutch.

The Capex Reckoning: $690B with No Clear Payback

The four hyperscalers announced combined 2026 capex approaching $690B:

Company 2025 Capex 2026 Capex (Projected) Change
Amazon $125B $200B +60%
Google/Alphabet $57B* $175-185B +200%+
Microsoft ~$80B ~$140B+ +75%
Meta $72B $115-135B +60-88%

*Google 2025 figure based on earlier guidance; actual spend may differ.

Analyst projections warn that Big Tech free cash flow could drop up to 90% in 2026. Amazon’s shares fell 12% on its $200B capex forecast. One-third of total U.S. GDP growth is now directly linked to AI infrastructure spending — a concentration that raises systemic questions.

The CFO Reality Check

RGP surveyed 200+ U.S. CFOs and found:

  • 66% expect significant AI ROI within 2 years
  • Only 14% report meaningful value today
  • 86% cite legacy systems limiting AI readiness
  • Only 10% fully trust their enterprise data

ManpowerGroup’s 2026 Global Talent Barometer (12,000+ workers surveyed) adds the workforce perspective: regular AI usage jumped 13% to 45% of workers, but confidence in using technology fell 18%. Baby Boomers reported a 35% drop in tech confidence. 43% of workers fear automation will replace their job within two years — up 5 points from 2025.

The picture: executives promise AI returns they cannot yet measure, while workers use AI tools they do not trust, in organizations with data they know is unreliable.

Key Data Points

Metric Value Source
S&P 500 Q4 calls citing AI 331 (68%) FactSet, Mar 2026
10-year average AI citations 94 per quarter FactSet, Mar 2026
Companies quantifying AI impact 10% Goldman Sachs, Mar 2026
Companies quantifying earnings impact 1% Goldman Sachs, Mar 2026
Executives reporting zero AI productivity impact ~90% NBER (n=6,000), Feb 2026
AI-citing stock underperformance (Q4) -4.1pp vs. non-citing FactSet, Mar 2026
Productivity gain in customer service/dev ~30% median Goldman Sachs, Mar 2026
CFOs expecting ROI in 2 years 66% RGP (n=200+), Dec 2025
CFOs reporting ROI today 14% RGP (n=200+), Dec 2025
Worker AI confidence decline -18% ManpowerGroup (n=12,000+), 2026
2026 hyperscaler capex (projected) ~$690B Company earnings calls

What This Means for Your Organization

The earnings call data reveals a dangerous gap. Executives feel compelled to discuss AI because their peers do — 68% of S&P 500 companies now mention it. But the substance behind those mentions is thin. Only 10% can point to measured outcomes. Only 1% can tie AI to earnings impact. Meanwhile, the market has stopped rewarding talk and started punishing it.

For mid-market companies watching Fortune 500 peers make bold AI claims, three realities matter:

First, the two proven use cases — software development and customer service — deserve investment. Goldman Sachs found ~30% median productivity gains in both areas among companies that actually measured. These are not aspirational; they are repeatable. If your organization has not deployed AI coding tools for your development team or AI-assisted customer interaction, you are behind on the one area where evidence is clear.

Second, everything beyond those two functions remains experimental. The NBER study of 6,000 executives — the most rigorous survey to date — found near-zero economy-wide productivity impact from AI. General Mills’ $20M logistics savings and JPMorgan’s 6% productivity gains are real but narrow. Most companies claiming broad AI transformation cannot substantiate it. The honest answer for most organizations: AI is working in pockets, not across the enterprise.

Third, the capex arms race is irrelevant to you. The hyperscalers spending $690B are building infrastructure. Mid-market companies are buying services. Your question is not “how much should we invest in GPU clusters” but “which of the proven use cases apply to our specific workflows, and can we implement them without the data readiness problems that 86% of CFOs report?” The answer requires an honest assessment of your data infrastructure before committing to AI tools — not after.

Sources

Created by Brandon Sneider | brandon@brandonsneider.com March 2026