Stop looking at AI as an IT project. Start seeing it as a Capital Allocation Strategy by Richard Taveras
- sgiddens8
- 1 day ago
- 4 min read
In the world of corporate finance, every dollar has a job to do. For decades, the most
common application for dollars was to acquire a business on capital expenditures – either buying a competitor’s cash flow or investing in growth. But in 2026, the conversation among executives has changed. We are now in a capital allocation super-cycle where CEOs must decide: do we buy growth through M&A, or do we increase our ability to reduce margins/drive growth (while leveraging AI)?
This capital allocation decision is not a technical debate around technology for technology’s sake – it has real world valuation implications. Capex (buying hard assets), AI (technology R&D), and M&A are simply three different levers for the same goal: Total Shareholder Return (TSR). However, the risks and rewards of these paths are diverging in a way we’ve never seen before.
The New Math of Capital Allocation
The scale of investment in AI is staggering: global corporate spending on AI infrastructure and integration is expected to top $500 billion by 2026. To put that in perspective, that is capital that otherwise would have fueled a massive wave of mid-market consolidation in prior cycles.
This sea change is indicative that the market is beginning to reward “home-grown” competitive advantages over "purchased" scale. A company that successfully implements AI to lower its operating costs by 10% isn't just saving money - it is structurally changing its operational effectiveness and creating an advantage that competitors cannot easily bridge by simply buying more assets. It changes pricing dynamics and can reshape entire markets.
Living with a 95% Failure Rate: The High Price of Entry
Despite the hype, the numbers on the ground are sobering. Recent data from MIT’s Project NANDA indicates a brutal reality: 95% of enterprise AI projects fail to break even or show a measurable P&L impact within their first year.
From a financial lens, many would see this as "value leakage" or, put plainly, yielding a low/negative ROI. Most companies are treating AI like a software rollout, when they should be approaching it how they would an acquisition: when you buy a company, you perform months of due diligence on their assets, their people, and their data. Yet, many firms are buying AI solutions off the shelf or launching full-scale implementations of proprietary software without checking if their own "house" is in order.
The primary culprit for this 95% failure rate isn't the technology - it’s technical debt. What is technical debt? It can take many forms, but is often silo-ed software, disparate databases, unstructured data, legacy codebases (“spaghetti code”), etc.
Trying to run a modern AI model on fragmented, siloed, or "dirty" legacy data is like putting a high-octane racing engine into a car with a broken transmission. You might spend millions, but the car isn't going anywhere.
The Winner-Take-All Dynamic
This brings us to the most significant trend of 2026: market polarization. We are seeing a "K-shaped" outcomes in corporate performance. On one side, the "AI High Performers" (representing roughly 6% of respondents on McKinsey’s 2025 Global Survey on the state of AI) are seeing their profit margins expand and their growth increased, with a report from BCG’s Build for the Future 2025 Global Study showing that the initial advantage of these investments is expected to compound over time.
These dynamics give the highest performers increasing pricing flexibility, improved purchasing power, and the ability to raise capital at a higher valuation to invest more – a virtuous cycle for those who execute their AI and M&A strategies well.
On the other side are the laggards who are stuck in "AI pilot purgatory." They are spending the same amount of capital but seeing zero-to-negative returns. If a market leader can use AI to handle twice the volume with the same headcount, they can outprice and outbid their competitors for every future deal. This creates a winner-take-all dynamic that leaves three distinct options for those who can’t keep up: evolve, get bought out, or die. The moat isn't just the tech - it's the capital efficiency the tech provides.
Transform as You Transact: The AI/M&A Hybrid
The most sophisticated players aren't simply choosing between AI and M&A; they are using AI to de-risk their deals. Historically, the "synergy" phase of a merger - where you actually save money by combining operations – is the hardest part of a business plan to realize. It usually takes 18 to 24 months to merge ERP systems and customer databases, to rationalize headcount, and to consolidate shared services – and that’s if those synergies are realized, at all.
Today’s winners (the elite 5%) are using AI to compress that timeline. They use autonomous agents to map data and integrate back-office functions in weeks, not years. This accelerates the "Speed to Value," allowing the combined entity to show margin improvement in months, not years.
Another tangible path in an M&A strategy is capability-driven M&A: larger firms are no longer just buying competitors for their customers; they are buying smaller "AI-native" firms just to acquire their data structures and talent (the latter has historically been known as “acqui-hires”). In this world, an acquisition is often the fastest way to fix the "95% failure" problem - it’s cheaper to buy a success than to build one on top of a broken legacy system.
The Mandate for Executives and Directors
For any CEO or Board member reading this, the message is clear: Stop looking at AI as an IT project, but more as a capital allocation strategy.
Every dollar you sink into an AI pilot that doesn't have a clear path to a positive impact on the P&L is a dollar that could have been used for a strategic acquisition. Conversely, every acquisition you make without an AI-driven integration playbook is likely to leave value on the table.
The goal for 2026 isn't to have the most AI for having AI’s sake - it’s to drive the most value for each dollar spent.