Organizations are at an inflection point with artificial intelligence. McKinsey found that four in five organizations see no impact on their bottom line despite rolling out AI across their companies, while 85% of leaders in Deloitte’s 2025 survey increased their AI investment in the past 12 months, even as payback periods stretch to 2-4 years instead of the typical 7-12 months.

These studies represent snapshots in time, showing that AI technology is not the problem. The real problem is how companies are bolting AI onto legacy systems that were never built to support it. Without proper foundations, most organizations cannot even measure what success looks like. They are spending because they are afraid of falling behind, not because they have seen proof it works.

For procurement and supply chain leaders, 2026 will be the year that separates those who can prove ROI from those who cannot. Organizations that show faster cycle times, documented cost savings, and business impact outputs that their CFOs trust will gain executive support. Everyone else will watch their AI budgets be reallocated and, possibly, even their roles.

Why AI Projects Fail in Procurement

McKinsey's research reveals fundamental mismatches between how AI works and how companies are deploying it. The "GenAI paradox" emerges from an imbalance between horizontal enterprise-wide copilots, which have scaled quickly but deliver diffuse, hard-to-measure gains, and vertical function-specific use cases, about 90% of which remain stuck in pilot mode. 

Generic AI tools like ChatGPT excel at individual tasks precisely because they are flexible and general-purpose. In enterprise settings, this generality is a liability, as these bolt-on tools often fail, because they do not systemize your specific workflows, adapt to your processes, or capture institutional knowledge. Organizations get the same generic outputs, whether one is sourcing office supplies or negotiating a multi-million-dollar software contract.

Most enterprise GenAI budgets still flow to sales and marketing tools despite back-office automation delivering bigger ROI potential. Procurement should be getting the lion's share of AI investment. Instead, what little investment does reach procurement gets wasted on legacy systems never designed for AI in the first place.

These legacy systems cannot handle the constant data flows and real-time analysis AI requires, as they were built for rigid workflows, not adaptive intelligence. The companies getting real returns on AI in procurement are using platforms where AI is natively built into the architecture from day one.

The high failure rates today reflect implementation problems, not AI limitations. Organizations that mistake current struggles for permanent constraints will miss the window to build competitive advantage

What Successful AI Implementations Have in Common

Organizations getting results from AI let their teams drive adoption rather than mandating it from a central AI lab. When the people doing the work have ownership, adoption happens faster, and tools get shaped by real needs.

The most advanced procurement and supply chain organizations are experimenting with agentic AI systems. These tools learn from past sourcing events, remember supplier performance data, and execute multi-step processes within set boundaries. 

Early adopters are already seeing returns. BNY reports having 117 agentic solutions that touch everything at the bank with tangible bottom-line impact. As Vinod Bidarkoppa, CTO at Walmart International, put it, “AI is a force multiplier that lifts productivity across all lines of business.” These solutions handle end-to-end workflows like supplier onboarding or contract renewals without needing a professional to babysit every step.

The biggest difference between organizations that prove ROI and those that do not comes down to what they measure. Successful teams track metrics their CFOs care about: How fast are you moving from request to contract signature? How accurate are your supplier risk assessments? What cost savings can you document and defend in a board meeting? Those questions determine whether your AI budget grows, or gets cut.

How Procurement Leaders Should Approach AI ROI in 2026

The industry needs to rethink how it measures AI returns. Traditional ROI frameworks do not capture what AI actually delivers in procurement. You need metrics that account for both quick wins and longer-term changes. 

Quick wins include productivity gains and cost reduction in the next quarter. Longer-term value includes process redesign and the shift from reactive buying to strategic supplier relationships. Both require different measurement approaches.

Procurement leaders should focus AI spending where the value is quantifiable and immediate. Intelligent front door intake routing cuts days off request processing. Automated RFP analysis turns week-long bid reviews into same-day decisions. And these are not theoretical benefits. They show up in system cycle time reports and team calendars.

AI pilot purgatory has to end. Too many organizations run experiments that never scale or conclude. Define success metrics up front, and kill projects that cannot show measurable returns within 18 months. Then, redirect that budget to what is actually working.

Next year, procurement will divide into two distinct paths. One group will treat AI as foundational infrastructure, embedding it into operations the way finance embedded ERP systems 20 years ago. These organizations invest in purpose-built platforms designed for AI from the architecture up. 

The other path leads nowhere. Those organizations will keep bolting generic solutions onto incompatible systems and wondering why nothing sticks. 

The gap between these two groups will widen rapidly. Organizations in the first camp will have CPOs at the executive strategy table, armed with documented savings, faster cycle times, and CFO-trusted impact metrics. Those in the second will spend 2026 defending budget cuts and explaining to boards and executives why AI still has not delivered. 

Leaders who use 2026 to set hard metrics, eliminate unproductive pilots, and build on AI-native infrastructure will shape procurement's next decade.

Stan Garber serves as co-founder and president of Levelpath, an AI-native procurement platform.