Executives say artificial intelligence will upend their organizations? Most still can�t get it out of pilot mode?
In a new Readiness Report from IT services provider Kyndryl, 87% of 3,700 senior executives across 21 countries say AI will completely transform roles and responsibilities within the next year? Yet only 29% believe their workforce has the skills to use it, and a mere 13% qualify as �pacesetters� pairing bold vision with the investments and operating changes to execute? If you�re wondering why AI feels both everywhere and nowhere inside large companies, this is your answer?
The widening execution gap
The report�s paradoxes are stark? While 90% of leaders say their tools and processes can rapidly test and scale new ideas, 57% admit innovation stalls on foundational issues in their tech stack? More than half (54%) report measurable ROI from AI�but 62% say their AI work remains stuck in pilots?
�A readiness gap exists as enterprises grapple with the promise of transformative value from AI,� Kyndryl CEO Martin Schroeter said? �Closing that gap is the challenge and opportunity ahead?�
Viewed against market dynamics, that gap looks less like hesitation and more like structural friction: legacy data plumbing, integration debt, unclear ownership for process redesign, and escalating compute costs? The stakes are rising fast?
Follow the money�and the risk
Venture groups have poured $161 billion into AI this year, with the bulk flowing to just 10 companies whose combined valuations swelled by nearly $1 trillion, according to the Financial Times� Due Diligence analysis? OpenAI, the sector�s bellwether, reportedly has $13 billion in annual recurring revenue yet posted an estimated operating loss of $8 billion in the first half of 2025 (FT)? Other analysis puts the first-half loss closer to $9?7 billion (Ars Technica), underscoring how expensive these systems remain to build and run?
Concentration cuts both ways? The FT warns of �cargo cult� behavior�businesses mimicking AI strategies without proven revenue gains�with one cited estimate that 95% of companies have yet to see AI drive revenue growth? Bain projects roughly $2 trillion in revenue will be needed to fund data centers by 2030, highlighting the capital intensity required to scale AI infrastructure? That�s not a pilot problem; that�s a balance-sheet problem?
Investors aren�t just chasing the boom�they�re hedging
Blackstone president Jonathan Gray says Wall Street is underestimating how quickly AI could upend rules-based white-collar work�from claims processing to parts of legal and accounting�and has pushed AI risk to the front pages of the firm�s investment memos? �People say, �This smells like a bubble,� but they�re not asking: �What about legacy businesses that could be massively disrupted?�� Gray said? Blackstone has avoided some software and call-center acquisitions over AI vulnerability concerns while doubling down on data center and utility infrastructure exposed to AI demand?
That split view�froth atop real disruption�captures why most companies are stuck? As Ars Technica�s conversation with AI critic Ed Zitron emphasizes, costs are rising �unilaterally across the board� and grand promises like �autonomous agents� remain unproven? Yet even skeptics acknowledge utility for well-scoped tasks? Meanwhile, tech luminaries such as Jeff Bezos have called today�s frenzy an �industrial� bubble that could still be �good� by accelerating digital infrastructure buildout, paving the way for durable future uses?
What the 13% pacesetters actually do
Both Kyndryl�s pacesetters and the FT�s investor lens suggest a common playbook:
- Fix the foundation first? Address data quality, integration, and security bottlenecks before scaling? The 57% citing tech-stack delays aren�t wrong; they�re early?
- Target rules-based workflows? Prioritize processes with clear inputs, constraints, and measurable outcomes (e?g?, claims triage, invoice matching, QA summarization) where AI can cut cycle times and error rates?
- Track unit economics, not demos? Require model costs and failure modes to be visible at the work-step level; tie savings to P&L owners?
- Avoid vendor sprawl? Consolidate on a small set of platforms and model providers, but maintain exit ramps to manage cost and quality drift?
- Upskill with purpose? The 29% skills readiness finding is a wake-up call; pair training with redesigned processes and new KPIs, not generic courses?
Notably, Kyndryl�s segmentation mirrors a recent Cisco study that also identified a roughly 13�14% cohort of leaders, suggesting the �execution minority� is real, not an artifact of one survey?
Why this matters now
For boards and CFOs, the question isn�t whether to �do AI,� but whether your operating model can support it at production scale? The market is rewarding infrastructure builders and punishing exposed middlemen? A year from now, the advantage may belong to organizations that turned pilots into durable workflows with clear economics�not just viral demos?
Is 13% a ceiling or a starting line? The answer will depend less on model benchmarks than on whether companies are willing to invest in the boring, expensive parts of transformation�and to measure results with the same rigor they use for any other capital project?