Most boardroom conversations about AI start in one of two places: productivity or risk. Both matter, but neither is how a board, an investor or an acquirer actually values a business. Having built a business and taken it through an acquisition by a private equity-backed group, I've come to think about AI in terms of the levers that drive shareholder value, and it changes the conversation considerably.
Bottom up and top down
Employees use AI to make their jobs easier. Organisations need AI to change how the business performs. Those aren't the same thing, and most organisations currently have far more of the first than the second. The latest ONS figures show that by mid-2026, 55% of workers in Great Britain said they used AI for work or education, while only around 35% of businesses with 10 or more staff reported using AI themselves. The ONS notes the two measures aren't directly comparable, but the gap is hard to ignore.
That bottom-up adoption is genuinely valuable, but on its own it tends to produce individual time savings rather than business outcomes, and it carries risk nobody has formally accepted. The job of leadership is to connect the two: to decide which value levers AI should move, and to make sure the use already happening across the business is pointed at them.
The shelfware lesson
In IT asset management, one of the most consistent findings was that people use a small fraction of the capability they are given. Organisations bought suites for everyone and paid for features that were never touched. I see no reason AI will be different, and the early evidence suggests it isn't, with one large US survey of workers with paid AI subscriptions finding that, where Copilot and ChatGPT were both available, only 18% made Copilot their main tool.
The other lesson from software is that utilisation is uneven. Some roles get a great deal of value and others very little, which is exactly why licensing AI for everyone rarely moves the numbers on its own.
The value levers
Private equity tends to look at a business through a small number of value drivers: top-line revenue growth, the proportion of recurring revenue, margin, customer retention, intellectual property, market position and people. AI can strengthen almost all of them, but it can also quietly erode them, and it is worth looking at both sides.
| Value lever | How AI can enhance it | How AI can quietly erode it |
|---|---|---|
| Revenue growth | Faster proposals, new AI-enabled services | Inaccurate outputs reaching customers |
| Recurring revenue | New subscription or managed offerings | Customers replacing part of your service with AI they run themselves |
| Margin | Efficiency, less reliance on outsourcing | Metered AI spend and unused licences absorbing the saving |
| Customer retention | Faster, more consistent service | AI errors or data incidents damaging trust |
| Intellectual property | Capturing and scaling know-how | Leakage into external tools, unclear ownership of AI-generated work |
| Market position | Speed and scale | Commoditisation, as competitors use the same models |
| People | Capability and capacity | Deskilling, dependency, and knowledge held in personal AI accounts |
The pattern is that the upside tends to come from deliberate, top-down use aligned to a specific lever, while the downside tends to come from unmanaged, bottom-up use that nobody has visibility of.
IP: the lever most at risk
For many businesses the most significant exposure is intellectual property, because it is often where much of the value sits and its erosion is largely invisible. Three things concern me most.
The first is leakage: proprietary designs, pricing models, source code and client information pasted into AI tools outside the organisation's control. The second is ownership. The legal position on who owns AI-generated work is still developing in many jurisdictions (this is a general point, not legal advice), and relying heavily on AI-generated material without understanding that position is a risk in itself. The third, and the least discussed, is know-how. Much of a business's value sits in how its people do things, and that knowledge is increasingly being captured in prompts, workflows and personal AI accounts rather than in the business. When someone leaves, it can leave with them.
For any business expecting to raise investment, be acquired or sell, this matters directly. AI-specific warranties and indemnities are becoming increasingly common in M&A as buyers try to bridge the information gap on how AI is used. A business that can't evidence how its IP is protected is likely to find that reflected in the deal.
In my experience, true IP has always been hard to value unless a business owns something genuinely innovative. AI adds a further question: proving that the business, rather than a tool or a former employee's personal account, owns the innovation in the first place. I expect that to become a routine part of due diligence.
The obvious objection
An obvious objection is that this is a private sector lens and doesn't fit public sector or not-for-profit organisations. I don't think that holds. The use of AI is a business decision whatever the sector, and every organisation has its own equivalents of these levers: service outcomes rather than revenue, cost to serve rather than margin, public trust rather than market position, and the expertise of its people in every case. The questions are the same; only the language changes.
Questions for the board
- Revenue growth: which revenue opportunities could AI genuinely open up, and who owns them?
- Recurring revenue: could AI allow customers to replace part of what we sell them, and how are we responding?
- Margin: are we measuring AI savings net of what we spend on AI?
- Customer retention: where does AI touch our customers, and how would we know if it got something wrong?
- Intellectual property: do we know what proprietary information is going into AI tools, and who owns what AI produces for us?
- Market position: if our competitors use the same AI, what still differentiates us?
- People: is our know-how staying in the business, or moving into personal AI accounts?
Where to start
Most of these questions can't be answered without knowing how AI is actually being used across the business. That evidence is what tells you which levers are genuinely moving, which are at risk, and where to invest next, which is why I'd always start there rather than with a strategy document or a licence purchase.
Which value lever do you think AI will affect most in your business, for better or worse?
Sources
- Office for National Statistics, Artificial intelligence in UK businesses: 2023 to 2026 (July 2026)
- Lexology, the pitfalls of AI warranties in M&A and commercial transactions
- Recon Analytics, AI Choice 2026: Why Licenses Don't Equal Adoption (US survey of paid AI users, published February 2026)