Wharton’s third annual Gen AI adoption study reveals enterprises moving from experimentation to accountability and struggling with the transition. Here are the headline numbers.
Usage is mainstream: 82% of senior decision-makers use Gen AI weekly, up from 37% in 2023. Nearly half use it daily.
Budgets are growing: 88% expect increases in the next 12 months, with two-thirds investing $5M+ annually. About 30% of tech budgets now go to internal R&D.
ROI is growing rapidly: 74% report positive returns. Only 72% formally measure ROI, suggesting many assessments are subjective.
The skills gap is widening: 49% struggle to recruit advanced AI talent. Training investment is actually declining (-8pp YoY) while expectations rise.
The study tracks three phases:
- 2023: Exploration (37% weekly usage, fascination and caution)
- 2024: Experimentation (72% weekly usage, pilots spreading, scrutiny increasing)
- 2025: Accountable Acceleration (82% weekly usage, ROI measurement standard, guardrails tightening)
The research predicts 2026 as the inflection point where organizations move from controlled experiments to performance at scale.
Interestingly, a striking pattern emerges between executives and mid-managers:
- 56% of VP+ say their organization is adopting “much quicker” vs. 28% of managers.
- 45% of VP+ report significantly positive ROI vs. 27% of managers
- Mid-managers are 18 percentage points less likely to see Gen AI causing skill proficiency declines
Those closest to implementation are more cautious than those in the C-suite. Does this reflect out-of-touch executives, or does it more reflect The fear for mid-level managers of displacement by AI?
As expected, adoption is uneven across a variety of dimensions:
By function: IT and Procurement lead (90%+ weekly usage). Marketing/Sales and Operations lag behind, despite obvious use cases.
By industry: Tech/Telecom, Banking/Finance, and Professional Services report 90%+ weekly usage and 80%+ positive ROI. Retail (63% weekly usage, 54% positive ROI) and Manufacturing (80% usage, 75% ROI) trail significantly.
By company size: Large enterprises (Tier 1, $2B+ revenue) closed last year’s adoption gap but are more likely to report ROI as “too early to tell” (34% neutral/too early vs. ~10% for smaller firms).
When it comes to what’s not working, it seems that the top challenges aren’t technical but tend to be cultural:
- Recruiting talent with advanced Gen AI skills (49%)
- Providing effective training for current employees (46%)
- Maintaining employee morale in impacted roles (43%)
- Leadership that can manage organizational change (41%)
This very much aligns with what we are seeing. Security risks, operational complexity, and data inaccuracy rank as top concerns—but these are implementation problems, not adoption barriers.
What It All Means
Organizations have moved past “should we adopt Gen AI?” to “how do we make this work?”
The constraint is no longer technology or budget—it’s human capital. Companies can’t hire fast enough, training programs aren’t keeping pace, and 43% worry about skill proficiency declining even as 89% believe Gen AI enhances employee capabilities.
This creates a paradox: organizations are accelerating AI adoption while simultaneously struggling to build the capabilities needed to use it effectively.
The study reveals a disconnect between investment and capability:
- 60% of enterprises now have Chief AI Officers, but most are “new responsibilities for existing roles” rather than dedicated positions
- 48% invest in employee training programs, but confidence in training as a path to fluency dropped 14 points YoY
- Internal R&D spending is substantial (30% of tech budgets), but most companies still rely on general-purpose tools rather than custom solutions
In short, most organizations are betting on building capability internally, but the gap between ambition and execution is growing.
Gen AI adoption is real and accelerating. But “adoption” means daily use of tools like ChatGPT, not necessarily strategic transformation.
The organizations pulling ahead aren’t just investing more—they’re solving the capability problem. The ones falling behind have budgets and enthusiasm but lack the talent, training, and organizational design to execute.
As the study’s authors frame it: this is the shift from “accountable acceleration” to “performance at scale.”
The question isn’t whether enterprises will adopt Gen AI. It’s whether they can build the capabilities to use it effectively before their competitors do.
Source: “Accountable Acceleration: Gen AI Fast-Tracks Into the Enterprise” - Wharton Human-AI Research and GBK Collective, October 2025. Study of ~800 senior decision-makers at US enterprises with 1000+ employees and >$50M revenue.