Stay updated with Data & AI News!
Join The Ravit Show Newsletter
“The Evolution of AI — The State of Enterprise AI and Data Architecture Report”
Ravit Jain
September 25, 2025
Cloudera’s latest installment of its annual enterprise AI survey lands clear signals from over 1,500 IT leaders across the U.S., EMEA, and APAC. Here are the takeaways I think matter.
Adoption and value
96% have AI at least somewhat integrated in core processes.
21% say it is fully integrated, 54% significant, 21% somewhat.
52% report significant, measurable value from AI; only 1% report no success.
Models in play
Generative 60%, deep learning 53%, predictive 50%.
Agentic models show up at 36%, and readiness to manage “new forms” of AI rises, with 67% feeling more prepared.
Classic methods are not dead. Regression jumps from 14% to 24%. Supervised and deep learning each climb 8 points year over year.
Data reality
Culture: 86% say they are at least moderately data-driven; “extremely” rises to 24% from 17%. Trust in data is also up, with 24% “much more” and 41% “somewhat more” than last year.
Where data lives: private cloud 63%, public cloud 52%, data warehouse 42%, on-prem mainframe 38%, on-prem distributed 32%, data lake 25%, lakehouse 24%. Hybrid is the norm.
What leaders want from architecture: integrated MLOps 52%, automated pipeline orchestration 51%, granular governance 44%, unified access 41%.
Roadblocks
Biggest technical limits: data integration 37%, storage and compute 17% each, lack of automation 17%, latency 12%.
Only 9% have 100% of data available to AI; 38% say “most” is available.
Cost to access compute for model training spikes from 8% to 42% year over year.
Security worries are steady: data leakage during training 50%, unauthorized access 48%, insecure third-party tools 43%. Confidence is improving, and the share calling security the “biggest challenge” drops from 66% to 54%.
ROI focus for the next 12 months
Operational efficiency leads at 29%, followed by customer experience 18%, product innovation 15%, revenue 14%, risk 13%, and talent productivity 11%.
My take
Most enterprises are past the AI “if” and deep into the “how.” The pattern is clear: value shows up where teams marry model choice with disciplined data practices. Hybrid data is the default, so the winning move is to bring AI to the data with integrated MLOps, strong access controls, and automation across pipelines.
Two gaps still mute ROI: partial data availability and rising compute costs. Solving both needs tighter data integration and governance and smarter workload placement. The momentum around agentic AI is real, but it will only scale on trusted, well-governed data with consistent controls across cloud and on-prem.
🔍 Stay Ahead in AI & Data! Join 137K+ Data & AI professionals who stay updated with the latest trends, insights, and innovations.
📢 Want to sponsor or support this newsletter? Reach out and let's collaborate! 🚀
Best,
Ravit Jain
Founder & Host of The Ravit Show
