AI delivers 'zero financial payoff' to most companies, Apollo economist says
An Apollo economist’s claim that AI has delivered “zero financial payoff” to most large companies has triggered debate over whether current adoption is creating real economic value or just matching competitors’ gains while enriching AI vendors. Many engineers and knowledge workers report substantial productivity boosts, but others note that faster coding or content creation often fails to move key business metrics like revenue per employee, especially in bureaucratic organizations where bottlenecks lie elsewhere. Commenters frame this as a classic technology hype cycle: AI is clearly useful in specific niches and small firms, but its broad impact on profitability, jobs, and competitive dynamics remains uncertain and uneven.
What the Apollo analysis actually measures
- Thread notes the report mainly tracks revenue per worker for large firms, not direct ROI on AI projects.
- Many argue this can’t show AI payoff yet, since AI spend is still a small fraction of revenue and benefits are swamped by other noise.
- Some see the headline as a corrective to extreme AI-boosted GDP predictions, but misleading for judging current economic value.
Productivity gains vs. business impact
- Numerous practitioners report large personal/team productivity gains, especially in coding, data work, and content creation.
- Others warn people often misjudge their own productivity and confuse activity with value creation.
- Even where efficiency gains are real, they may not change company finances if:
- Bottlenecks are elsewhere (governance, sales, legal, deployment cadence).
- Competitors get similar gains, turning AI into a “tax” or Red Queen race.
- Extra capacity is not turned into new revenue or cost cuts.
Quality, process, and misuse concerns
- Some see AI encouraging bad engineering habits: less code review, less testing, more low‑quality features.
- Several argue AI mostly accelerates the easiest 5–10% of software work; the rest is coordination, integration, and risk management.
- Others describe using AI carefully (as a drafting or discovery tool) to improve reliability, refactors, and large migrations.
Where AI is clearly helping (anecdata)
- Examples include: faster chemical formulation experiments, infra migrations that unlock big infra savings, consulting report automation, expense management, marketing content, customer support, and creative businesses automating lead gen and service.
- These often show clear time or cost savings, but translation into higher margins or revenue per employee is mixed or unmeasured.
Competition, moats, and distribution of gains
- Several compare AI to past tech in commoditized industries: everyone adopts, margins stay flat, vendors and some consumers capture most gains.
- Some expect AI to enable tiny teams or individuals to compete with large orgs; others think incumbents will mainly use it for cost cutting.
Adoption stage and hype
- Many say mainstream usage is still shallow (treated as “better Google”), with deeper workflows and agents barely adopted.
- Others warn of a classic buzzword cycle: exec-driven pilots with poor alignment, lots of waste, eventual sorting into niches where AI truly matters.