The AI Budget Trap Why Easy Approvals Create Dangerous Spending Habits

Brianna Monsanto's recent IT Brew article reveals a statistic that should alert every CFO and board member: 62% of business leaders have tied a software purchase to an AI initiative specifically to secure budget approval. The implication is clear: executives have discovered that uttering the letters 'A' and 'I' functions like an open sesame for corporate spending.

This isn't just interesting industry gossip. It's a red flag signaling that companies are making technology investments based on label manipulation rather than strategic value. I've watched this scenario unfold before, and it doesn't end well.

The Real Challenge of AI Investment

Let me be direct: AI deserves significant investment. According to McKinsey's 2024 Global Survey on AI, organizations that invested heavily in AI capabilities reported 20% higher profit margins than their peers. PwC's 2024 AI Business Survey found that companies with mature AI implementations achieved average cost reductions of 15-20% in operational areas. These are real benefits worth pursuing.

The issue isn't that companies are investing in AI. The problem is that they've created a two-tiered budget approval system where projects labeled 'AI' receive preferential treatment regardless of their actual merit. This distorts capital allocation in ways that will take years to unwind.

Michele Shepard, CRO at Emburse, told IT Brew that AI projects now receive easy approvals because of their 'long-term benefits' and 'real efficiency savings.' But this explanation doesn't align with the 62% figure. If business leaders felt confident that their AI projects would receive approval on merit alone, why would nearly two-thirds of them feel compelled to artificially attach AI labels to software purchases?

The answer is uncomfortable: many of these projects probably wouldn't survive rigorous scrutiny without the AI label.

Historical Lessons on Investment Trends

Shepard compared the current AI investment wave to the shift from on-premise to SaaS solutions. This comparison actually illustrates the problem more than it justifies the spending. During the SaaS transition of the mid-2010s, Forrester Research documented that companies wasted an estimated $17 billion on redundant or underutilized SaaS subscriptions between 2014 and 2017. Gartner found in 2018 that the average organization used only 58% of its licensed SaaS capacity.

We made those mistakes because SaaS became a magic word that bypassed normal procurement discipline. Individual departments could spin up subscriptions without IT oversight. The result was Shadow IT chaos that companies are still cleaning up today.

The AI spending environment feels worse. At least SaaS solutions had clear, measurable functionality. You could test whether Salesforce actually managed your customer relationships or whether Slack improved team communication. AI systems are often black boxes where causation is harder to establish and results are easier to manipulate.

Consider the blockchain parallel. Between 2017 and 2019, Deloitte's 2020 Global Blockchain Survey found that 43% of blockchain projects initiated during the hype period were abandoned without ever reaching production. Companies spent billions attaching 'blockchain' labels to projects that didn't need distributed ledger technology. We called it 'blockchain-washing,' and it represented one of the most spectacular misallocations of corporate capital in recent memory.

We're now doing the same thing with AI, except the dollar figures are larger. IDC projects that worldwide spending on AI systems will reach $632 billion by 2028. If even 20% of that spending is misdirected due to budget manipulation, we're talking about $125 billion in wasted capital.

Understanding the Dysfunction

Ed Frederici, CTO at Appfire, offers an important counter-perspective in the IT Brew article. Despite his company tripling its AI budget in 2026, Frederici argues that AI projects actually face more scrutiny than conventional purchases. The 'AI' label jumps projects up the priority queue, but they undergo 'the same amount of rigor' during evaluation.

This should reassure us, but it actually highlights how dysfunctional the current system has become. If AI projects truly received the same scrutiny as other investments, business leaders wouldn't feel the need to artificially attach AI labels to their proposals. The fact that 62% do so anyway suggests one of two scenarios:

• Most companies don't have Appfire's discipline, and the AI label really does create a shortcut around normal approval processes. This means billions in questionable spending.

• Even companies with rigorous processes have created a de facto priority system where AI projects consume evaluation resources and executive attention at the expense of non-AI initiatives. This means companies are systematically under-investing in valuable non-AI projects because they can't compete for attention.

Neither scenario is good. In the first case, we're wasting money. In the second case, we're making suboptimal portfolio decisions.

Frederici notes that AI projects 'jump the priority chain' and 'get looked at sooner.' This is capital allocation malpractice. Priority should be determined by expected return on investment, strategic alignment, and risk-adjusted value, not by whether a project uses a fashionable technology. A company that evaluates a mediocre AI project before an excellent non-AI project has reversed the proper order of decision-making.

The Rise of AI-Washing

The 62% figure represents the mainstreaming of AI-washing, a practice where companies exaggerate or fabricate AI capabilities to attract investment or attention. Until now, AI-washing was primarily associated with startups misleading investors. A 2019 study by London venture capital firm MMC Ventures found that 40% of European startups classified as AI companies didn't actually use AI in a meaningful way.

Now we're seeing the same behavior inside established corporations, except the deception is internal. Business leaders are misleading their own CFOs and budget committees. This creates multiple problems:

• It corrupts the strategic planning process. If your capital allocation is based on mislabeled projects, your actual technology portfolio will diverge from your intended technology strategy.

• It makes performance measurement impossible. When projects are approved based on AI benefits they don't actually deliver, how do you evaluate success?

• It crowds out honest conversation about where AI actually creates value.

Integration and Strategic AI Use

Shepard makes one critical point that deserves more attention: 'Using AI in a silo is going to get you far, but not as far as aggregating that intelligence across multiple applications.'

This observation cuts to the heart of why the current approval environment is so problematic. Easy approval for AI projects creates exactly the siloed implementation pattern Shepard warns against. When individual business units can secure AI budgets by attaching AI labels to their wish lists, you get a fragmented landscape of disconnected tools.

MIT's 2024 research on AI implementation found that companies with integrated AI strategies achieved 3.5 times higher returns than those with fragmented, department-level AI deployments. The reason is straightforward: AI systems become more valuable as they access more data and connect more processes. A marketing AI that shares insights with a sales AI that informs a customer service AI creates compounding value. Three separate AIs that don't talk to each other create three small improvements.

A Framework for Better AI Investments

Companies need to replace their current ad hoc approach with structured evaluation frameworks. Based on my work with organizations navigating technology investments, here's what actually works:

• Establish AI-agnostic evaluation criteria. Every technology project, whether it uses AI or not, should be evaluated on the same dimensions: expected ROI, strategic alignment, implementation risk, and operational impact.
• Separate exploration from implementation budgets. Companies should maintain a dedicated AI experimentation budget for testing new capabilities and building organizational learning.
• Require proof of concept before scale. Companies should mandate that AI initiatives demonstrate value at small scale before receiving implementation budgets.
• Create portfolio-level AI governance. Companies should develop portfolio strategies that specify where AI can create value and how different initiatives should integrate.
• Track AI-specific benefits separately. This forces intellectual honesty about whether the AI components actually matter.

The CFO's Challenge

CFOs face a genuine challenge. They know their companies need AI capabilities to remain competitive. They've read the same McKinsey and PwC reports showing that AI leaders outperform AI laggards. They've heard board members ask about AI strategy in every quarterly meeting.

But they also sense something is amiss when 62% of business leaders admit to gaming the approval process. They see budget requests that would have been titled 'Customer Service Platform Upgrade' two years ago now called 'AI-Powered Customer Experience Transformation.' They notice that every proposal now includes the word 'intelligent' or 'smart' or 'cognitive.'

The temptation is to simply approve these requests and hope that some of them deliver value. After all, if 58% of leaders believe AI purchases are easiest to approve, maybe there's collective wisdom in that assessment. Maybe the organization's immune system is correctly identifying AI as beneficial and accelerating its adoption.

This would be a mistake. The historical record is clear: when capital allocation becomes disconnected from rigorous evaluation, bad things happen. The dot-com bubble, the blockchain boom, the SaaS sprawl—every instance where a technology category received preferential budget treatment produced massive waste alongside genuine innovation.

CFOs should instead use this moment to impose more discipline, not less. Here's what that looks like in practice:

• Institute a moratorium on the word 'AI' in budget requests. For the next budget cycle, require that proposals articulate their value proposition without using the terms 'AI,' 'artificial intelligence,' or 'machine learning.' If a project can't explain its benefits without those labels, it doesn't understand its own value proposition.
• Demand technical architecture reviews. Every project claiming AI capabilities should undergo a technical review that validates whether AI is actually core to the solution.
• Benchmark against non-AI alternatives. Before approving any AI project, require a comparison against conventional solutions to the same problem.
• Create accountability for AI-specific claims. When projects receive approval based on AI capabilities, tie compensation and performance reviews to the delivery of those specific capabilities.

Rethinking Business Leadership Tactics

For the 62% of business leaders who have attached AI labels to budget requests, a message: you're winning the battle but losing the war.

• You're training your organization to make decisions based on labels rather than value.
• You're creating an AI portfolio that looks impressive on paper but delivers mediocre results.
• You're building technical debt as fragmented AI tools proliferate without integration.
• You're undermining your own credibility when the promised AI benefits don't materialize.
• You're making it harder for genuine AI initiatives to compete for attention and resources.

The smarter approach is to be scrupulously honest about where AI creates value and where it doesn't. When you have a legitimate AI opportunity—a problem where machine learning genuinely offers advantages over conventional solutions—make that case explicitly. Provide technical details. Explain the data requirements. Specify the expected lift over non-AI alternatives. Show proof-of-concept results.

When you have a good project that doesn't require AI, present it as exactly that. A well-designed conventional software solution that doesn't use AI is vastly superior to a mediocre solution that does. The goal isn't to maximize AI deployment; it's to maximize business value.

Companies that figure this out will outperform their peers. According to Boston Consulting Group's 2024 AI Maturity Research, organizations with 'selective AI deployment'—meaning they carefully choose which problems benefit from AI rather than trying to apply it everywhere—achieved 27% higher ROI on their technology investments than companies with 'aggressive AI adoption' strategies.

The Path Forward

The current budget approval environment for AI projects is unsustainable. It's creating misaligned incentives, fragmenting technology portfolios, and wasting capital on a scale that will become obvious when we look back from 2030.

But the solution isn't to stop investing in AI. The solution is to invest more strategically.

Companies should redirect their energy from chasing AI labels to pursuing specific capabilities that happen to use AI. Instead of asking 'How can we deploy more AI?', ask 'What business problems do we have where the best solution involves machine learning?' Instead of measuring success by the number of AI projects approved, measure the business outcomes those projects deliver.

This requires courage from both budget requesters and budget approvers. Requesters need to stop gaming the system even though gaming currently works. Approvers need to impose discipline even though saying yes feels easier than saying no.

The organizations that make this shift will build AI capabilities that create sustained competitive advantage. The organizations that continue down the current path will build impressive-sounding AI portfolios that don't actually do much.

Five years from now, we'll be able to identify which companies took which path by looking at their financial performance. The AI washing will have washed out, and the real AI value will be obvious.

The choice is whether your organization will be in the outperformer category or the underperformer category. That decision is being made right now, in every budget approval meeting where someone utters the words 'AI.' Choose carefully.

For more insights on the implications of AI on budget allocations, explore this thought-provoking article on IT Brew.