Editorial illustration for Data Issues Plague Disjointed AI Initiatives, Says Kask
AI Initiatives Underperform Due to Data Issues: SAP
Data Issues Plague Disjointed AI Initiatives, Says Kask
SAP put a number on something a lot of executives have suspected for a while: the AI they've rolled out is doing less than it could. The SAP Value of AI Report 2026, built with Oxford Economics from a survey of 2,600 business leaders across 13 countries, found AI now handles close to 30% of tasks in the average organization, up from 25% a year ago. ROI expectations for agentic AI nearly doubled too, climbing from 10% to 17%.
Those are real gains. But the report's authors argue the bigger story is the gap between what AI is producing now and what companies believe it should be producing.
Sean Kask, SAP's chief AI strategy officer, points to a familiar culprit: not the models themselves, but the mess underneath them. Data that doesn't talk to other data. Governance built after the fact, if at all.
Strategy assembled piecemeal rather than set from the top. More than half of organizations surveyed are still investing in AI in an ad hoc way, and only 17% describe their approach as strategic and holistic, even though that figure has nearly doubled since last year.
ROI expectations for agentic AI have jumped from 10% last year to 17% this year, but many organizations believe AI could be delivering far more value. The report reveals that the gap comes down to strategy, data, and governance, rather than access to the newest model, says Sean Kask, chief AI strategy officer at SAP.
Why this matters
The SAP numbers look good on a slide: AI touching 30% of tasks, ROI expectations for agentic AI climbing from 10% to 17%. But Kask's quote is the part worth sitting with. Silos don't disappear because a pilot works.
A team gets clean data for one use case, ships it, celebrates the ROI bump, then hits a wall trying to scale that same model to a neighboring department with different data hygiene. That's the pattern we keep seeing across enterprise AI reporting this year: success at the point of contact, breakdown at the point of expansion. For founders selling into enterprise, this is your actual buying signal, not the ROI headline.
Buyers aren't asking "does it work," they're asking "does it work everywhere we'd need it to." For researchers and platform teams, it's a reminder that agentic AI's ceiling isn't model capability right now, it's data plumbing nobody wants to fund. Anyone building AI infrastructure should treat "consistent data across use cases" as the actual product spec, because that's where 2,600 business leaders say the value is stuck.
Related Reading
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
From Archives Dili raises USD 21.7M to automate AI compliance for infrastructure projects
From Archives Microsoft Confirms Copilot 'Super App' for This Year
Common Questions Answered
What percentage of tasks does AI currently handle in the average organization according to the SAP Value of AI Report 2026?
According to the SAP Value of AI Report 2026, AI now handles close to 30% of tasks in the average organization, which represents an increase from 25% a year ago. This survey was conducted with Oxford Economics and included responses from 2,600 business leaders across 13 countries.
How much have ROI expectations for agentic AI increased according to the report?
ROI expectations for agentic AI have nearly doubled, climbing from 10% last year to 17% this year. Despite this significant increase, many organizations believe AI could be delivering far more value than it currently is.
What does Sean Kask identify as the main barriers preventing organizations from maximizing AI value?
According to Sean Kask, chief AI strategy officer at SAP, the gap between current and potential AI value comes down to strategy, data, and governance, rather than access to the newest AI models. He suggests that organizations need to focus on these foundational elements rather than simply upgrading their technology.
Why do data silos prevent AI initiatives from scaling across departments?
Data silos prevent scaling because teams often achieve success with clean data for one specific use case, but encounter obstacles when attempting to apply the same model to neighboring departments that have different data hygiene standards. This pattern of isolated pilot successes followed by scaling failures is a recurring issue in enterprise AI implementation.
Further Reading
- The AI Data Problem: Bridging The ‘AI Activation Gap’ - CRN
- The Data Problems Undermining Midmarket AI Projects In 2026 - MES Computing
- Why AI Stumbles Without a Solid Data Strategy - Bain & Company
- The Data Problem Stalling AI - MIT Sloan Management Review
- 4 data gaps that cause enterprise AI projects to fail - Quartz
