{"slug": "harness-survey-estimates-26-of-ai-spending-is-wasted", "title": "Harness Survey Estimates 26% of AI Spending Is Wasted", "summary": "Harness released its 2026 State of AI in FinOps survey on July 29, estimating that enterprises waste 26% of AI spending. The survey of 700 engineering leaders across five countries found that 52% lacked a clear AI-cost owner, 72% had encountered a surprise AI bill in the past year, and only 20% could diagnose a doubled bill within hours.", "body_md": "# Harness Survey Estimates 26% of AI Spending Is Wasted\n\nHarness released its 2026 State of AI in FinOps survey on July 29, estimating that enterprises waste 26% of AI spending. The 700-person survey across five countries found that 52% of respondents lacked a clear AI-cost owner, 72% had encountered a surprise AI bill in the past year, and only 20% could diagnose a doubled bill within hours.\n\nHarness released its 2026 State of AI in FinOps survey on July 29, estimating that enterprises waste **26% of AI spending**. The company surveyed 700 engineering leaders and practitioners at organizations that use AI or large-language-model services.\n\nThe online survey was conducted by Sapio Research in May and June 2026. It included 300 respondents in the United States and 100 each in the United Kingdom, France, Germany, and India. Every respondent worked at an organization with at least 1,000 employees and recurring AI spending.\n\nThe results are self-reported estimates from a vendor-sponsored survey, not audited financial data. They are most useful as a benchmark for how respondents describe ownership, visibility, and governance rather than as a measured waste rate for every enterprise.\n\n### Ownership and visibility gaps\n\nHarness reported that **52%** of respondents had no clear owner for AI costs, with responsibility divided among engineering, platform teams, FinOps, finance, and IT. It also found that 72% had experienced an unexpected AI cost spike or bill in the prior year, including 33% that had been surprised more than once.\n\nOnly **20%** said they could identify the cause within hours if AI spending doubled. Forty percent said diagnosis would take a full day, 32% said a week, and 8% said longer or never.\n\nThe survey found that 67% of respondents' organizations spend more than **$250,000 per month** on AI and 20% spend more than **$1 million per month**. Applying the survey's 26% waste estimate to a $1 million monthly bill yields $260,000, but that calculation remains an extrapolation from respondents' estimates rather than a verified loss.\n\n### Cost signals often arrive after deployment\n\nHarness reported that 45% of engineers understood the cost of the AI features they build, while 56% of respondents described AI-spend forecasting as guesswork. It also found that 57% said their organizations encourage maximizing AI usage regardless of demonstrated value, a pattern the report calls \"tokenmaxxing.\"\n\nThe governance gap was also visible in policy enforcement. Although 73% reported having an AI-cost policy, 47% said it was fully enforced, and 26% reported a robust way to measure the business value of AI spending.\n\nFor ML platform, FinOps, and engineering leaders, the practical distinction is between monitoring aggregate provider bills and attributing costs to a workload, feature, team, or business outcome. Model choice, token volume, retries, retrieval pipelines, and agent execution paths can all change cost. The survey does not prove that unclear ownership causes every instance of waste, but it gives teams concrete questions to test: who owns the bill, how quickly anomalies can be traced, and whether usage data connects spending to delivered value.\n\n## Key Points\n\n- 1Harness's vendor-sponsored survey estimates that 26% of enterprise AI spending is wasted; the figure is self-reported rather than audited.\n- 2More than half of respondents lacked a clear AI-cost owner, and only 20% said they could diagnose a doubled bill within hours.\n- 3Workload-level attribution matters because model, token, retrieval, retry, and agent choices can change costs before they appear on an aggregate invoice.\n\n## Scoring Rationale\n\nThe survey offers timely enterprise benchmarks on AI cost ownership, waste estimates, and spend visibility for ML platform and FinOps teams. Its evidence is a vendor-sponsored, self-reported survey rather than audited financial data, which limits the strength and generalizability of its quantitative conclusions.\n\n## Sources\n\nPrimary source and supporting public references used for this report.\n\nPractice interview problems based on real data\n\n1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.\n\n[Try 250 free problems](/problems)", "url": "https://wpnews.pro/news/harness-survey-estimates-26-of-ai-spending-is-wasted", "canonical_source": "https://letsdatascience.com/news/harness-report-finds-ai-spend-waste-f0a5fac5", "published_at": "2026-07-30 16:46:00+00:00", "updated_at": "2026-07-30 18:28:17.298974+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-policy", "ai-infrastructure"], "entities": ["Harness", "Sapio Research"], "alternates": {"html": "https://wpnews.pro/news/harness-survey-estimates-26-of-ai-spending-is-wasted", "markdown": "https://wpnews.pro/news/harness-survey-estimates-26-of-ai-spending-is-wasted.md", "text": "https://wpnews.pro/news/harness-survey-estimates-26-of-ai-spending-is-wasted.txt", "jsonld": "https://wpnews.pro/news/harness-survey-estimates-26-of-ai-spending-is-wasted.jsonld"}}