In football, even the top quarterback in the league fails if they aren’t protected. AI has become that quarterback, but its success is determined by whether enterprises have built the fundamentals required to execute consistently at scale.
According to NTT Data, 70-85% of AI initiatives are failing to meet expectations, largely due to weak governance, poor data quality and lack of operational readiness. The failure rate is a signal that execution maturity hasn’t kept pace with adoption.
As organizations look ahead, the following 2026 predictions explore what will influence the ability to turn AI investments into measurable business results.
Prediction 1: Governance will become the primary competitive driver
In football terms, governance is protection. It rarely gets credit, but without it, the quarterback never gets time to throw.
AI adoption is already widespread. McKinsey’s State of AI Research report shows that 71% of organizations use generative AI in at least one business function, yet more than 80% report that it hasn’t produced a tangible impact on enterprise-level EBIT. This gap between adoption and results is where the real challenge lies.
In earlier stages of AI adoption, the challenge was a lack of access to AI or a willingness to experiment. There’s been a fundamental shift here, and now the challenge is the absence of governance that connects AI activity to business outcomes.
NTT Data reinforces this conclusion, citing poor governance and weak data hygiene as leading causes of AI initiatives that fail to deliver ROI. AI efforts become isolated plays rather than a coordinated drive when there’s no clear ownership, standards or oversight.
This year, the clearest signal that an organization is serious about AI will be how well they govern its use. If they continue to treat AI as a collection of separate pilots, they will struggle to scale. Those who build clear rules for how AI is approved, used and monitored will be able to apply it more broadly.
Prediction 2: Data quality will decide whether AI scales or stalls
A strong offensive line creates momentum and consistency, while a weak one stalls drives before they develop.
As AI moves from experimentation into routine use, organizations will feel the effects of their data foundations more directly. Data that is incomplete, inconsistent or poorly governed will slow AI down and change the way it behaves.
When data is weak, outputs become unreliable, automation behaves unpredictably and users hesitate to rely on recommendations. The result looks like a broken play at the line of scrimmage.
The Stanford HAI 2025 AI Index Report shows continued growth in AI investment, reaching $109 billion in U.S. private AI investment in 2024. However, the report also highlights that data readiness will be a persistent barrier to value realization.
This finding matches what delivery teams see every day. PWR Teams reports that poor data quality is a common reason AI projects fail, and that up to 80% of AI project time is spent cleaning and preparing data rather than delivering business value.
Data quality will determine how far AI can realistically be applied inside the enterprise.
Prediction 3: Trust will shape how AI is used
In football, players have to trust the playbook. If they don’t believe the play will work, execution breaks down.
As AI becomes more visible across enterprise workflows, trust will become a central factor in how it’s used day to day.
According to Gartner’s 2025 Hype Cycle, AI trust, risk and security management (TRiSM) will play a major role in how AI will be adopted over the next five years. As organizations turn to AI to scale operations and make faster, better decisions, the emphasis is shifting away from what is novel to what actually works. Gartner notes that AI agents and AI-ready data deliver value only when they are applied deliberately and governed well, not when they’re rolled out broadly without a clear purpose.
This will be seen in how employees interact with AI systems. When AI use is clearly defined, governed and supported by strong data management, people are more likely to trust and use its outputs. When those protections are missing, usage will become cautious and selective, regardless of how advanced the technology appears.
Prediction 4: AI success will be about reliability over time
The path for AI won’t look like a highlight reel of flashy plays; it will look like long, controlled drives that wear down the defense.
The way organizations talk about AI success will change in 2026. Early phases of AI adoption emphasized experimentation and deployment milestones. Attention will shift to whether AI produces stable, repeatable outcomes in real operating conditions.
The Stanford report found that 78% of organizations use AI in some form, yet value creation remains uneven. This imbalance will be impossible to ignore. Gartner’s analysis reinforces this pattern, noting that many generative AI initiatives stall before delivering the value expected.
As a result, conversations about AI will focus less on rollout and more on sustained performance. AI systems will be judged on whether they continue to improve decision quality, efficiency or risk management over time. Reliability will matter more than novelty.
A league divided by fundamentals
In 2026, the field will be clearly defined. Some enterprises will still have a lot of tools, but little coordination, running pilots that never mature. Others will operate with less noise but more impact, applying AI in ways that are repeatable and trusted.
After decades of enterprise transformation cycles, one lesson holds: hype creates motion, but fundamentals create results.
Some firms have already oriented themselves around this reality. At Columbus Global, we’re focusing our AI and data strategies on strong data foundations, disciplined governance and enterprise-level alignment rather than chasing isolated use cases or experimental tools.
AI may be the quarterback, but in this league, the winner is still decided in the trenches.
