IBM Study: 75% of AI Projects Fail. Here's How to Avoid the Same Mistakes (2025)

# IBM CEO Study 2025: Why 75% of AI Initiatives Fail to Deliver Expected ROI

A new IBM study asked 2,000 CEOs around the world about their AI projects, and the results are pretty wild. Only 25% of AI initiatives actually delivered the return on investment they expected over the past three years. Even worse? Just 16% managed to scale up across their whole company. That's a big gap between what companies want to do with AI and what actually happens.

The study came out in May 2025, right when boards and investors are putting more and more pressure on CEOs to show that AI is actually working. A Forbes article about the same topic found that 74% of CEOs are worried they'll lose their jobs within two years if they don't prove AI is making money.

Why are three-quarters of AI initiatives failing to meet expectations? And more importantly, what can startups and growth-stage companies learn from these failures to avoid the same costly mistakes?

The IBM CEO Study: Key Findings

Survey Methodology

The IBM Institute for Business Value, in cooperation with Oxford Economics, conducted a global survey in the first quarter of 2025 (February–April 2025). The study surveyed 2,000 CEOs from:

33 countries across all major regions
24 industries spanning technology, finance, healthcare, manufacturing, retail, and more
Organizations ranging from mid-market to enterprise-level

The full survey looked at how companies perform, what they're focused on, innovation challenges, how they use new tech, leadership styles, hiring plans, and rules they have to follow.

Critical Statistics: The AI ROI Reality Check

AI Initiative Success Rates:

25% of AI initiatives delivered expected ROI over the past three years
16% successfully scaled enterprise-wide
68% of CEOs report having clear metrics to measure innovation ROI effectively
65% prioritize AI use cases based on ROI

AI Investment Intentions:

CEOs expect AI investment growth rates to more than double over the next two years
61% of CEOs are actively adopting AI agents today and preparing to implement them at scale
85% expect positive ROI from scaled AI efficiency investments by 2027
77% expect positive ROI from scaled AI growth and expansion investments by 2027

Organizational Challenges:

50% of CEOs report disconnected technology due to rapid recent investments
64% invest in technologies before fully understanding their value (FOMO-driven decisions)
59% struggle to balance funding for existing operations and innovation investments

Why 75% of AI Initiatives Fail: The Root Causes

If you're thinking about using AI at your company, you need to know why most projects fail. The IBM study and Forbes article both found the same problems happening over and over.

1. The Speed vs. Strategy Tension

The Problem: CEOs feel like they need to move fast, but they don't have a clear plan.

Here's the thing: two-thirds of CEOs say they pick AI projects based on ROI, but only 52% actually see their AI investments making money beyond just cutting costs. So there's a gap between what they want to do and what actually happens.

Why This Happens:

FOMO-driven decisions: 64% of CEOs invest in technologies before fully understanding their value
Lack of clear AI strategy: Only 37% say it's better to be "fast and wrong" than "right and slow"
Insufficient planning: Only 12% of CEOs have a formal AI deployment roadmap extending beyond one year

The Startup Parallel:

Startups deal with the same stuff, but it's way worse because they don't have the money to bounce back. Big companies can afford to mess up with AI, but if a startup's AI project fails, it can stop their growth or burn through all their cash.

2. Disconnected Technology and Data Architecture

The Problem: When companies rush to buy AI tools, they end up with a mess of systems that don't talk to each other.

Half of the CEOs surveyed said their companies moved too fast and now have tech that doesn't work together. When everything's broken up like this, AI can't do its job.

Key Findings:

68% identify integrated enterprise-wide data architecture as critical for cross-functional collaboration
72% view proprietary data as key to unlocking generative AI value
50% report disconnected technology due to rapid investments

Why This Matters:

AI needs your data to be organized and connected to work. When everything's broken up:

AI can't see all your data
It gets way harder to connect everything
It costs way more to keep running
You basically can't scale up

3. The "AI Commodity Trap"

The Problem: CEOs think they can just buy any AI tool and it'll work as well as something built specifically for them.

According to Forbes, 87% of CEOs admit they fell into the "AI commodity trap"—basically thinking generic AI tools will fix their specific problems.

This Trap Leads To:

AI that doesn't work well
Wasting money on tools that don't fit
Missing chances to beat competitors
Getting frustrated because AI doesn't do what you thought it would

The Reality:

Different industries need different AI solutions. A basic customer service chatbot won't work the same way for a fintech company's compliance tool or a healthcare company's diagnostic system. Each industry needs its own custom stuff.

4. Governance Gaps and Shadow AI

The Problem: When employees use AI tools without permission, it creates problems with rules, security, and quality.

Forbes found something wild: 94% of CEOs think their employees are secretly using AI tools that the company hasn't approved. This creates big problems with management and oversight.

Risks Include:

Compliance violations: Using AI tools that don't meet industry regulations
Data security breaches: Sensitive data exposed through unapproved platforms
Quality inconsistencies: AI outputs that don't meet company standards
Attribution issues: Accountability problems when AI decisions go wrong

5. Regulatory Uncertainty and Compliance Paralysis

The Problem: CEOs put off or cancel AI projects because they're worried about breaking rules.

Statistics from the Forbes analysis:

37% of CEOs have delayed AI projects due to unclear compliance requirements
32% have abandoned AI initiatives due to regulatory concerns
79% of global executives are concerned about potential slowdowns from EU AI Act and similar regulations

Impact on Startups:

Startups usually have fewer rules to worry about at first, but if you're in healthcare, finance, or other regulated industries, you're dealing with the same problems. The uncertainty stops a lot of companies from even trying AI.

6. Insufficient Strategic Planning and Roadmaps

The Problem: Most CEOs don't have long-term plans for AI.

Only 12% of CEOs have a real plan for AI that goes beyond one year. That's pretty surprising, since most CEOs expect to be judged on their AI results within two years.

Why This Matters:

Without a plan:

AI projects end up being reactive instead of strategic
You use your resources wrong
You don't know how to measure success
Scaling up is really hard

7. Talent and Expertise Gaps

The Problem: Companies don't have the right skills to make AI projects work.

Key Findings:

54% of CEOs are hiring for AI roles that didn't exist a year ago
31% of the workforce will require retraining/reskilling over the next three years
65% plan to use automation to address skill gaps
67% say differentiation depends on having the right expertise in the right positions

The Startup Challenge:

Startups usually can't afford to hire full-time AI experts, but they need that same level of skill to succeed. It's a catch-22: you need AI experts to make AI work, but you can't pay for them until AI starts making you money.

What Successful CEOs Are Doing Differently

The IBM study found "leading CEOs" who do better than their competitors in revenue growth and profit margins. These successful CEOs handle AI differently:

1. Focus on ROI-Focused Innovation

Successful CEOs focus on AI projects that will clearly make money instead of jumping on every new AI trend. They:

Connect AI performance to company results
Figure out what success looks like before starting
Measure ROI in bigger ways (not just cost cutting)

2. Build Integrated Data Environments

Top CEOs build connected data systems that make AI work:

One view of all your data across systems
Self-service ways to access data
Real-time collaboration across teams
Cloud platforms that can grow with you

3. Adopt Strategic, Not FOMO-Driven Approaches

Top CEOs don't jump on every new tech trend. Instead, they:

Take smart risks based on data and strategy
Balance moving fast with being careful
Focus on what they can actually control
Avoid the "move fast and break things" mindset

4. Invest in Talent and Partnerships

Successful companies don't try to do everything themselves. They:

Hire for jobs that didn't even exist a year ago
Train their current employees to learn new skills
Work with outside experts for specialized help
Create AI teams that mix tech and business skills

Key Takeaways for Startups and Growth-Stage Companies

1. Start with Strategy, Not Technology

Before investing in AI tools or platforms, develop a clear AI strategy:

Identify specific business problems AI can solve
Define success metrics upfront
Create a roadmap that extends beyond initial pilots
Align AI initiatives with business objectives

2. Invest in Data Architecture Early

Don't wait until AI initiatives fail to fix your data architecture:

Create integrated data views before scaling AI
Establish data quality frameworks
Implement proper governance from the start
Use cloud-native platforms that scale

3. Avoid the Commodity Trap

Resist the temptation to use generic AI tools for everything:

Evaluate whether off-the-shelf solutions fit your specific needs
Consider custom solutions for competitive advantages
Build industry-specific AI capabilities
Partner with consultants who understand your domain

4. Establish Governance and Policies

Don't let shadow AI create compliance and security risks:

Create approved AI tool policies
Provide training on responsible AI use
Implement usage dashboards and monitoring
Develop clear guidelines for AI in sensitive areas

5. Balance Innovation with Operations

Don't sacrifice existing operations for AI experiments:

Allocate separate budgets for innovation
Measure ROI broadly (not just cost reduction)
Create "innovation liquidity" for new opportunities
Balance short-term ROI with long-term transformation

6. Plan for the Long Term

AI transformation is a multi-year journey, not a one-time project:

Create roadmaps extending 12+ months
Plan for scaling from the beginning
Invest in talent development
Build partnerships for ongoing support

The Path Forward: Learning from the 25% Who Succeed

The IBM study shows that while 75% of AI projects fail to make money, 25% actually succeed. Knowing what makes the difference between success and failure is super important.

Success Factors Identified:

Clear ROI Focus: Successful companies focus on projects that make real money
Connected Systems: Data that works together makes AI successful
Smart Patience: Balance moving fast with being careful
Right Skills: Access to experts and partners who know what they're doing
Long-Term Planning: Plans that go beyond just testing things out
Good Rules: Clear policies and oversight from the start

What This Means for AI Consulting

The IBM study findings show why getting help from AI consultants is more valuable than ever, especially for startups and growing companies:

Why Startups Need Expert Guidance

1. Avoid Costly Mistakes

The study shows that 75% of AI projects fail—often after spending a ton of money. Expert consultants help companies avoid common mistakes and boost their chances of success.

2. Access Specialized Expertise

With 54% of CEOs hiring for AI jobs that didn't even exist a year ago, most companies don't have the skills they need. Consultants give you access to experts without having to hire them full-time.

3. Strategic Roadmap Development

Only 12% of CEOs have formal AI plans beyond one year. Consultants help you build real strategies that stop you from making random, rushed AI investments.

4. Navigate Complexity

From data architecture to rules to compliance, AI transformation involves a lot of hard areas. Experienced consultants help you figure out all of it.

What to Look for in an AI Consultant

Based on the IBM study findings, choose AI consultants who:

Focus on ROI: Focus on projects that will clearly make money
Understand Your Industry: Skip generic solutions; they know your specific business
Build Connected Systems: Help you create data systems that work together, not separate silos
Plan Long-Term: Make plans that go beyond just testing things out
Provide Ongoing Support: Offer part-time leadership or monthly support for real success

Conclusion: Turning AI Challenges into Opportunities

The IBM CEO study shows a tough truth: 75% of AI projects fail to make the money companies expected. But it also shows what makes the difference between success and failure.

For startups and growing companies, these findings show why it matters to:

Plan before you start buying tech
Build connected data systems from the start
Avoid making decisions just because you're afraid of missing out
Set up clear rules and policies
Balance trying new things with keeping operations running
Work with experts who get both AI and your business

The pressure on CEOs is real—74% are worried they'll lose their jobs if they don't deliver AI results. But instead of rushing into AI investments, successful companies take a careful, planned approach that helps them join the 25% who succeed.

The question isn't whether to invest in AI; it's how to invest smart so you don't end up in the 75% who fail. By learning from the IBM study and doing what successful companies do, startups can boost their chances of AI success a lot.

Sources and References

Primary Sources:

IBM Institute for Business Value (2025). "2025 CEO Study: 5 mindshifts to supercharge business growth." Survey conducted February–April 2025 with 2,000 CEOs globally. Available at: https://www.ibm.com/thought-leadership/institute-business-value/en-us/report/2025-ceo

IBM Newsroom (May 6, 2025). "IBM Study: CEOs Double Down on AI While Navigating Enterprise Hurdles." Press release summarizing key findings. Available at: https://newsroom.ibm.com/2025-05-06-ibm-study-ceos-double-down-on-ai-while-navigating-enterprise-hurdles

Forbes (May 22, 2025). "Why AI Demands Have 74% Of CEOs Fearing For Their Jobs" by Caroline Castrillon. Analysis of CEO pressure and AI implementation challenges. Available at: https://www.forbes.com/sites/carolinecastrillon/2025/05/22/why-ai-demands-have-74-of-ceos-fearing-for-their-jobs/

Related Research:

McKinsey Global Survey on AI: "The State of AI" - Cited in Forbes article regarding top-down AI implementation necessity
Dataiku/Harris Poll (2025): Survey of 500 global CEOs regarding AI and job security concerns

Related Resources

About Lighthouse AI

Lighthouse AI is a fractional AI operations partner for growth-stage tech companies. We help Series A–C startups design, implement, and scale AI-powered operations in 90 days. Our team combines strategic AI consulting with hands-on implementation to help startups avoid the common pitfalls that cause 75% of AI initiatives to fail.

Last Updated: May 7, 2025

Next Review: August 2025

📬 Want More AI Insights?

Join 500+ tech founders getting weekly AI implementation strategies, case studies, and tools delivered to their inbox.

No spam. Unsubscribe anytime. We respect your privacy.