Skip to content
RedRock Systems CAIO
say “ciao”
Insights

The numbers behind the seat, with their sources

Every figure on this site comes from a named source, with its qualifiers kept: predictions are labelled as predictions, vendor data as vendor data. Full list at the bottom of the page.

Proof of concept

Many GenAI projects stall after the pilot

Three different measures. Gartner’s 2024 figure was a prediction; its 2026 figure is an assessment of hundreds of implementations. S&P counts organisations, not projects.

Share abandoned or abandoning
Gartner prediction (2024): GenAI projects abandoned after PoC by end-2025≥30%
Gartner assessment (2026): GenAI projects abandoned after PoC≥50%
S&P Global: organisations abandoning most AI initiatives, before17%
S&P Global: organisations abandoning most AI initiatives, now42%

Source: Gartner press release, Jul 2024 (prediction); Gartner, “Why 50% of GenAI Projects Fail”, Jan 2026; S&P Global Voice of the Enterprise, via CIO Dive, Mar 2025.

Agentic AI
40%+

of agentic AI projects will be cancelled by the end of 2027, Gartner predicts. A prediction, not a measured result.

Source: Gartner press release, Jun 2025 (prediction).

The value gap

Most AI initiatives have not paid back

McKinsey finds high performers are about three times more likely to redesign workflows: a planning and architecture job, not a tool purchase.

Expected ROI
25%

of AI initiatives delivered the expected ROI (IBM, 2,000 CEOs).

Source: IBM IBV + Oxford Economics CEO study, May 2025.

Scaled
16%

scaled enterprise-wide. 64% invest in some technologies before they understand the value.

Source: IBM IBV + Oxford Economics CEO study, May 2025.

Any EBIT
37%

attribute at least some EBIT to AI, flat versus 2025.

Source: McKinsey, The state of AI in 2026, Aug 2026.

High performers
~6%

qualify as AI high performers. BCG finds only 5% of companies are “future-built”.

Source: McKinsey, The state of AI in 2026, Aug 2026; BCG Build for the Future 2025, via PR Newswire, Sep 2025.

Cost

AI spend now behaves like cloud spend

98%

of FinOps teams now manage AI spend, up from 31% two years earlier

Source: FinOps Foundation, State of FinOps 2026.

Constraint
~20%

say AI operating costs, including tokens, constrained their AI use.

Source: McKinsey, The state of AI in 2026, Aug 2026.

Explaining the bill
87 → 22%

of finance leaders CloudZero surveyed need to tie AI spend to outcomes this year; 22% can. Vendor data.

Source: CloudZero blog (vendor survey, n=260), Sep 2026.

Agent sprawl

More agents, more tokens

Anthropic says multi-agent setups are viable only when the task’s value is high. First-party figures, not an independent cost study.

Tokens relative to a chat interaction
1×Chat
~4×Single agent
~15×Multi-agent system

Source: Anthropic Engineering, multi-agent research system, Jun 2025.

Failure modes
14

distinct failure modes in the MAST study of more than 1,600 traces across 7 multi-agent frameworks.

Source: Berkeley et al., Why Do Multi-Agent LLM Systems Fail? (arXiv:2503.13657).

People and security

Use is rising faster than training

Workers (PwC, n=49,364)
Used AI at work64%
Could access the learning they needed, before59%
Could access the learning they needed, now51%

Source: PwC Global Workforce Hopes and Fears 2026 (n=49,364).

Shadow AI
$670K

average added to breach costs; 1 in 5 organisations reported a shadow-AI breach.

Source: IBM Cost of a Data Breach 2025 (Ponemon), Jul 2025.

Controls
97%

of organisations reporting AI model or app breaches lacked proper AI access controls. 63% had no AI governance policy or were still developing one.

Source: IBM Cost of a Data Breach 2025 (Ponemon), Jul 2025.

The role

Few companies have a CAIO yet

Large US companies
10.7%

of Fortune 500 and S&P 500 companies had a Chief AI Officer or equivalent title as of mid-2026 (71 of 665).

Source: Crist|Kolder 2026 Mid-Year Volatility Report, via CFO Dive, Oct 2026.

Remote
60%

of fractional Chief AI Officer roles on GoFractional’s marketplace are remote (20% hybrid, 20% in-office). Small sample: 5 job posts and 302 profiles.

Source: GoFractional, Fractional CAIO rates 2026 (marketplace data).

Supporting point: energy
+17%

growth in data-centre electricity use in 2025, to about 485 TWh. Fewer tokens on right-sized models usually means less energy too.

Source: IEA, Key Questions on Energy and AI, 2026.

Sources

Every number, every source

Last checked 8 Oct 2026 [placeholder: re-verify sources and update this date before launch]. Excluded on purpose: MIT’s “95% of pilots fail”, “RAND found 80% fail”, and any fixed savings percentage.

  1. S1Gartner press release, Jul 2024 (prediction). https://www.gartner.com/en/newsroom/press-releases/2024-07-29-gartner-predicts-30-percent-of-generative-ai-projects-will-be-abandoned-after-proof-of-concept-by-end-of-2025
  2. S2Gartner, “Why 50% of GenAI Projects Fail”, Jan 2026. https://www.gartner.com/en/articles/genai-project-failure
  3. S3Gartner press release, Jun 2025 (prediction). https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027
  4. S4S&P Global Voice of the Enterprise, via CIO Dive, Mar 2025. https://www.ciodive.com/news/AI-project-fail-data-SPGlobal/742590/
  5. S5IBM IBV + Oxford Economics CEO study, May 2025. https://www.prnewswire.com/news-releases/ibm-study-ceos-double-down-on-ai-while-navigating-enterprise-hurdles-302446435.html
  6. S6IBM IBV + Oxford Economics CEO study, May 2025. https://www.prnewswire.com/news-releases/ibm-study-ceos-double-down-on-ai-while-navigating-enterprise-hurdles-302446435.html
  7. S7IBM IBV + Oxford Economics CEO study, May 2025. https://www.prnewswire.com/news-releases/ibm-study-ceos-double-down-on-ai-while-navigating-enterprise-hurdles-302446435.html
  8. S8McKinsey, The state of AI in 2026, Aug 2026. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
  9. S9McKinsey, The state of AI in 2026, Aug 2026. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
  10. S10McKinsey, The state of AI in 2026, Aug 2026. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
  11. S11McKinsey, The state of AI in 2026, Aug 2026. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
  12. S12McKinsey, The state of AI in 2026, Aug 2026. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
  13. S13BCG Build for the Future 2025, via PR Newswire, Sep 2025. https://www.prnewswire.com/news-releases/ai-leaders-outpace-laggards-with-double-the-revenue-growth-and-40-more-cost-savings-302570218.html
  14. S14RAND, Root Causes of Failure for AI Projects, Aug 2024. https://www.rand.org/pubs/research_reports/RRA2680-1.html
  15. S15FinOps Foundation, State of FinOps 2026. https://data.finops.org/
  16. S16CloudZero blog (vendor survey, n=260), Sep 2026. https://www.cloudzero.com/blog/explain-the-ai-bill-cfos/
  17. S17Anthropic, Prompt caching with Claude. https://claude.com/blog/prompt-caching
  18. S18OpenAI API docs, Batch API (retrieved Oct 2026). https://developers.openai.com/api/docs/guides/batch
  19. S19LMSYS, RouteLLM, Jul 2024 (one benchmark: MT-Bench). https://www.lmsys.org/blog/2024-07-01-routellm
  20. S20Anthropic Engineering, multi-agent research system, Jun 2025. https://www.anthropic.com/engineering/built-multi-agent-research-system
  21. S21Berkeley et al., Why Do Multi-Agent LLM Systems Fail? (arXiv:2503.13657). https://arxiv.org/abs/2503.13657
  22. S22IBM Cost of a Data Breach 2025 (Ponemon), Jul 2025. https://newsroom.ibm.com/2025-07-30-ibm-report-13-of-organizations-reported-breaches-of-ai-models-or-applications,-97-of-which-reported-lacking-proper-ai-access-controls
  23. S23IBM Cost of a Data Breach 2025 (Ponemon), Jul 2025. https://newsroom.ibm.com/2025-07-30-ibm-report-13-of-organizations-reported-breaches-of-ai-models-or-applications,-97-of-which-reported-lacking-proper-ai-access-controls
  24. S24IBM Cost of a Data Breach 2025 (Ponemon), Jul 2025. https://newsroom.ibm.com/2025-07-30-ibm-report-13-of-organizations-reported-breaches-of-ai-models-or-applications,-97-of-which-reported-lacking-proper-ai-access-controls
  25. S25PwC Global Workforce Hopes and Fears 2026 (n=49,364). https://www.pwc.com/gx/en/issues/workforce/hopes-and-fears.html
  26. S26GoFractional, Fractional CAIO rates 2026 (marketplace data, small sample: 5 job posts and 302 profiles; average $207/hour, about $10.7K/month; page updated 9 Oct 2026). https://www.gofractional.com/insights/rates/chief-ai-officer
  27. S27GoFractional, Fractional CAIO rates 2026 (marketplace data; 60% remote, 20% hybrid, 20% in-office as of 9 Oct 2026). https://www.gofractional.com/insights/rates/chief-ai-officer
  28. S29Heidrick & Struggles, 2025 Data, Analytics & AI Officers Compensation Survey. https://www.heidrick.com/-/media/heidrickcom/publications-and-reports/2025-data-analytics-and-artificial-intelligence-officers-compensation-survey.pdf?rev=aa52cdc0b0e64e4d81d09858f178146d
  29. S30Crist|Kolder 2026 Mid-Year Volatility Report, via CFO Dive, Oct 2026. https://www.cfodive.com/news/chief-ai-officers-arrive-c-suites-cristkolder/832014/
  30. S31IEA, Key Questions on Energy and AI, 2026. https://www.iea.org/reports/key-questions-on-energy-and-ai/executive-summary
  31. S32IEA, Key Questions on Energy and AI, 2026 (projection). https://www.iea.org/reports/key-questions-on-energy-and-ai/executive-summary
  32. F1European Commission, AI Literacy Q&A; Regulation (EU) 2026/1744. https://digital-strategy.ec.europa.eu/en/faqs/ai-literacy-questions-answers

You don’t need more AI. You need AI you can explain

What it’s for, what it costs, and who owns it. Start with a diagnostic: in a few weeks you’ll have a costed, business-aligned plan and a clear view of your AI spend and risk.

Not ready to talk?
Start free with the workshop and scorecard →

Book a diagnostic call →