Value before velocity: Why AI makes investment discipline more important, not less
AI is making it easier to build. That does not make it easier to know what’s worth building.
Worldwide AI spending is forecast to reach $2.7T in 2026, up 49.5% year over year, according to Gartner. Yet only 37% of respondents to McKinsey’s 2026 State of AI report say AI has had any impact on EBIT, almost unchanged from 2025.
Investment is accelerating; measurable value is not. The challenge is less about technology than about how organizations decide which AI initiatives to fund.
Many still fund AI as an annual budget line, with no criteria for when it should be reassessed, redirected, or stopped. As 2027 plans take shape, the question is whether organizations are investing in AI to solve clearly defined business problems, or for its own sake.
The noise problem
AI is the loudest story in business, and it seems that every day there’s a new model, tool, or company that claims to have completely automated another workflow. In the Q2 earnings season, 331 companies in the S&P 500 discussed AI on their calls, nearly double the five-year average. Every vendor has an AI roadmap, every conference an AI keynote, every board an AI question.
That volume clouds judgment, making organizations reactive to competitors and the market. When everyone is talking about AI, having AI starts to look like the goal. The story becomes “we’re doing AI,” not “here’s what AI is doing for our customers.”
The pattern is familiar. A vendor announces a model that promises higher productivity. An executive sponsor or a few eager employees turn it into a pilot, which becomes a cost in usage fees and, even more, in time. Before long, the portfolio is full of initiatives with no clear link to strategy and no explicit definition of value for customers or the business.
Emergn’s research reflects this challenge: across 700 senior leaders in the US and UK, organizations were running an average of 6.6 transformation and AI initiatives at once, and only 16% deliberately limited that number.
AI is here to stay, and it will change how value is created across the business. But to find that value, organizations need to stop treating AI as the story; what it does for customers is the story. Organizations that get the most from it filter the noise. They decide early how they’ll evaluate which opportunities are most likely to create value, and treat everything else as something to watch, not a pilot to fund.
From budget to value
A budget is a statement of what you will spend. It says nothing about what you will get. Most AI lines in 2027 plans are exactly that: a budget with no value hypothesis attached. Where they do carry a number, it is usually a savings target. Public announcements follow the same pattern: how much AI will save, rarely what it will make.
Even where revenue is named, savings dominate. At its AI investor day in October, Deutsche Telekom laid out €2.5B in savings by 2030, against €800M in AI revenue. These are unproven projections, and they point straight back at cost: an old mindset applied to new technology.
This isn’t because finance only cares about cost. CFOs are accountable for growth as well as financial performance. So why does cost dominate the plan?
The answer is uncertainty. If you are accountable for the numbers, you want certainty, and cost feels controllable. Growth doesn’t. Every route to growth carries real uncertainty, yet organizations often manage big ideas as if they were certain: detailed estimates, fixed scope, committed dates. As a result, most large organizations are far better at managing cost than at growing.
Annual planning focuses on technical risk: can we build it, on time and on budget? What decides the return is market risk: will customers value it? AI can help answer both, yet process, mindset, and measures all point toward cost and speed. As building gets faster, the capability that matters is discovery: reducing market risk with evidence from customers.
So the framing of the budget question has to change. Cost-based funding asks: Can we afford this? Value-based funding asks: What does this return, by when, and what are we not doing instead? AI makes the second question urgent, because it speeds up building without reducing the risk of building the wrong thing. Put simply, the question stops being “What can we afford?” and becomes “What can we afford to delay?”
The shift is not simply about spending less or spending more carefully. It is from funding activities to funding outcomes, with the value visible before the commitment is made. And once you fund for value, the decision to stop is no longer a failure. It is a result.
The final quarter problem: When budgets overshadow value
For many organizations, the final quarter of the year is budget season, and planning discussions often focus on justifying costs: what is in, what is out, what can be approved. Too often, the question that actually matters gets overlooked: what happens if we don’t do this? When trade-offs happen, they can take place several levels above the work, usually with a spreadsheet and not much else.
The risk is that existing initiatives roll into the following year, new work, much of it now AI-related, gets added on top, and difficult prioritization decisions are postponed. By January, teams may still be finishing last year’s work. By March, pilots may be struggling to scale, new initiatives falling behind, and questions about ROI starting to emerge.
What often gets overlooked is that the budget is only part of the equation. The bigger question is what the work is worth to your customers. Spending the right money on the right things creates value on several fronts at once: your customers, your people, your business. Yet weighing cost against value can get lost in planning discussions when the conversation focuses on expenditure rather than outcomes.
Cost of Delay makes those trade-offs visible. Strip the jargon out and it’s a simple question: what is gained, what is lost, and what does it cost to not do this today? Expressing that cost as a weekly amount can make the urgency tangible. Budget requests often arrive with a cost figure and a hope. A number per week helps show what waiting is costing, and what opportunities are sitting in the queue behind the initiative you chose not to stop.
For leaders finalizing their 2027 plans, the challenge is simple. Question everything. For every line in the plan, is this valuable? If we didn’t do it, so what? Before you name the ten things you’re doing next year, name the three you’re not. Say them out loud in the planning meeting.
If you can’t name three, that’s your answer.
Making value-based decisions work
Cost of Delay tells you what waiting costs, which points to where the highest economic value may sit. The point is not a perfect number. It is naming what you believe waiting costs, writing that assumption down with the team that owns the decision, and keeping it open to challenge when new evidence appears.
Most large organizations prioritize in silos. Each team or business area ranks its own backlog, which works when the value you care about lives inside that one team. It does not work when value depends on several teams delivering together.
If you are improving an existing product or experience, value shows up across a customer journey that spans many teams, platforms, and channels. If each team ranks only its own work, you get local improvements that do not add up: lots of change, little realized value.
If you are creating something new, the risk is different: backing the wrong bet, or backing the right one too slowly while a better use of capital waits. No single team’s backlog can show that trade-off.
Within a team, Cost of Delay ranks the problems most worth solving. Across the portfolio, alongside evidence for your value assumptions, it shows which opportunities to fund and which to stop, so capital moves to those whose assumptions still hold.
You need three things for that to work:
- One shared view of the portfolio, so every investment is visible against the same strategic choices.
- Value made visible before you commit, so nothing enters the plan without a value hypothesis and evidence that justifies its existence.
- Stopping initiatives as a normal outcome, not something people have to defend. If the value of solving a problem does not justify the effort, park or stop it and redirect the resources to problems worth solving.
At the portfolio level, revisit, at least quarterly, the Cost of Delay estimates for the opportunities you are funding. A portfolio Cost of Delay figure that has not moved in six months is almost certainly out of date.
Better investment decisions need more than a prioritization framework. They need a way to connect problems, evidence, and funding decisions across the portfolio, so organizations can see where value is being created, where assumptions no longer hold, and when to change direction: decisions based on outcomes, not output.
The decision that matters most
Budget season ends the same way every year. Someone signs off on a number, the plan gets locked, and everyone gets back to work. That part’s fine. The concern is how much ends up in the plan without anyone stopping to ask what it is actually for.
AI lets organizations start faster, but it doesn’t tell them any sooner what’s worth doing. Somebody still has to make that call, and it needs to be someone who can see across the whole portfolio and has the authority to stop things.
So, as your plan comes together this year, be honest with yourself: do you know what’s worth investing in, and what needs to stop, before the plan takes on a life of its own?
Additional insights
Emergn Survey Report: The Value Gap
Efficiency won’t turn your AI investment into growth
From AI strategy to impact: a product-led playbook
