Boards are still waiting for AI returns

Awaiting AI returns

Most enterprises have embraced AI, but few are seeing returns. With 95 percent of pilots failing to deliver measurable impact, boards must rethink how they fund, govern, and scale AI.

The reality is unforgiving. Enterprise AI usage is widespread; impact is not. The evidence is coming thick and fast: 95 percent of organisations report no measurable return from AI pilots (MIT 2025). Less than a quarter of firms have advanced beyond the proof-of-concept stage (McKinsey & BCG 2025)—and only 4 percent are creating substantial business value (BCG 2025). What is more, 42 percent of companies reported abandoning the majority of their AI initiatives, up sharply from 17 percent the year before (S&P Global Market Intelligence 2025). The list goes on.

So, why does this keep happening despite the heightened focus on AI at Board level? How do you spot if your AI investment isn’t effective? And what must boards do to stop funding illusion and start demanding impact?

Why is this happening?

Boards possess robust financial and deep industry experience, but often lack the digital, data, and AI perspective required to unlock the next wave of growth.

These blind spots leads to, amongst others, two distinct outcomes:

1. Over-reliance on external voices. Vendors chasing sales quotas, consultants on day rates, and influencers with little real-world experience. Left unchecked, they overpromise and underdeliver. Yet when managed well and incentivised for long term success — through risk/reward sharing and co-investment — the right external partners can be powerful accelerators. The key is retaining strategic ownership, ensuring they have skin in the game, and leveraging their expertise to upskill your organisation, not blind trust.
2. Impaired ability to perceive and respond to long-term risks. Without digital and AI literacy, Boards struggle to see the full implications of emerging technologies, including potentially existential threats.

In the absence of critical challenge, Boards often approve AI investments without sufficiently scrutinising answers to critical questions: Who owns the outcome? Is the right leader accountable? Do we even have the data—and is it good enough?

The result is predictable: enthusiasm for AI, proliferation of pilots, and mounting investment followed by inevitable disappointment when nothing shows up in the P&L. 

Here’s what most Boards don’t yet appreciate:

• In the short term, your data is your moat. You will not outrun rivals on model accuracy, context window length, or whatever jargon your friendly salesperson throws at you. You will win on the quality, liquidity, and governance of your data—if focused on the right use cases. Even in industries with stronger grasp of data quality—such as life sciences—52 percent of R&D leaders cite poor data quality as the number one barrier.
• In the long term, re-imagining your business with AI at its core will be the winning strategy.

How to spot this if your AI investment is going to waste

Warning signs are obvious if you know where to look:

1. The pilot graveyard. Easy-to-demo tools thrive in experimentation. But enterprise deployments stall at the coalface: brittle workflows, poor integration, no contextual learning, and no clear path to benefit realisation. Fewer than one in five pilots actually scale.
2. Generic chatbots rolled out to all. LLM-powered copilots are useful productivity aids but hardly transformational.
3. Misplaced bets. Too much AI spend flows into superficial front-office pilots — sales, marketing, customer service. They create activity and dashboards but little lasting ROI. Finance, procurement, and operations remain underfunded despite being where margins are lost and leakage measurable. 
4. Finance in particular is a missed opportunity: automating reconciliation, forecasting, and controls can cut cycle times by 70–80 percent, slash errors, and tighten compliance. Unlike front-office pilots, these initiatives deliver ROI with clear metrics and fast feedback loops. Boards that overlook them are buying visibility, not value.
5. Timeframe confusion. Boards overestimate what AI can deliver in a year and underestimate what it can do in three to five. The result: money goes in, value never arrives, and attention shifts to the next shiny thing—sometimes yet another AI demo.

This is not a strategy. This is trend-following with a budget.

What the Board must do

I’ve seen this pattern too many times. But I’ve also worked with leaders who break it. They share a clear discipline:

• Invest more in people and processes than in algorithms.
• Focus on core business processes.
• Prioritise use cases can be pushed into production quickly, while building the infrastructure and culture to scale.
• Use AI not just to automate—but to reimagine the business: transform what you do, not just how fast you do it.

Recent studies by MIT CISR and McKinsey back this up: Organisations who are able to scale and embed AI, financially outperform their peers significantly. 

Boards must act differently. 

They need to embed AI expertise where decisions are made—guest speakers and one-off briefings no longer cut it. Companies need board and executive directors with scars from real AI transformations; that is, people who can interrogate both status quo and proposed investments, and deliver the outcomes.

I have seen outcome ownership assigned to the P&L with the Chief Data Officer (CDO) as co-owner. ROI comes only from a P&L, so the accountable executive must sign up to deliver it. The CDO’s role is to co-own outcomes by quantifying opportunities, ensuring the data foundations are fit for purpose and driving adoption of the emerging solutions. A CDO confined to reporting and governance is a wasted seat. (CDO here is shorthand for any Data, Analytics, or AI-focused executive.)

It is also time to call PoCs what they are: a distraction. Fund fewer, bigger, better projects. Kill the pilot zoo. Back a small portfolio of high impact initiatives that deliver measurable ROI while building reusable capabilities.

Finally, build a scalable platform. 

Every project must leave behind stronger foundations: reusable components, higher data and AI literacy, better data governance, and continuous performance evaluation. Each cycle should compound capability—making the next initiative faster, cheaper, and more impactful.

Closing thoughts

Boards that follow this discipline will stop funding pet projects and build sustained advantage. 

The final takeaway is simple: automation preserves today, but only reimagination secures tomorrow. AI is already reshaping product design—from drugs to consumer goods—enabling discovery, customisation, and new revenue streams. Boards that see AI as process improvement will survive. Boards that redesign their business around it will lead. The choice is stark: build the future, or watch your model erode.

Join the HotTopics Contributing Editors Network

Ako Sabir is a transformation leader with 25 years of experience driving change through data, analytics, AI, and technology. As Chief Data, Analytics and AI Officer and Chief Technology Officer, Ako has led major programmes across global financial services, telecoms, and technology organisations. He advises Boards and Executive teams in PLC and PE-backed companies on turning AI ambition into measurable business value.