If data there is no artificial intelligence. The one that is supposed to be the engine of the companies in the coming years, it is impossible for it to do so correctly because the companies do not have the data for the AI sufficiently prepared. At least it is what follows from the study The Evolution of Ai: The State of Enterprise ai and data architecture carried out by Cloudera.
The report, which examines how AI is being adopted in all types of companies, what architectures of Data for AI They are being used and what are the challenges that have emerged in 2025, it is shown that companies are still at the beginning of taking out all the potential to this disruptive technology.
Compared to the past report, it stands out how the priorities, challenges and objectives have changed in just one year. In this sense, 24% of respondents say that their company has a completely culture Data-Drivencompared to 17% of 2024. However, the accessibility of the Data for AI It is still an obstacle: only 9% of organizations affirm that all their Data for AI They are prepared and accessible, while 38% say that most are. The highest technical limitations identified include the integration of Data for AI (37%), storage performance (17%) and computer power (17%).
The AI integrated into the processes
At the general level, 96% of those responsible said that AI is integrated, at least to some extent, in their basic business processes. This is an increase with respect to 2024, when the figure was 88%. In fact, 70% of respondents claimed to have achieved significant success with their initiatives based on Data for AIand only 1% have not seen results yet.
The study also analyzes what types of artificial intelligence are being used, with the generative in the lead (60%), followed by the Deep Learning (53%) and the predictive (50%). He also asked about the confidence in these models, and 67% of those responsible for you feel more prepared than a year ago to manage new forms of AI, especially those that require large volume of Data for AI.
“In just one year, AI has gone from being a strategic priority to becoming an urgent need, redefining the rules of competition,” said Juan Carlos Sánchez, regional vice president of Cloudera. However, the manager stressed that “our study reveals that companies continue to face great challenges in security, regulatory compliance and exploitation of Data for AI. Many of them are stagnant in the concept test phase ». The clouding mission is to take AI to the Data for AI Wherever they are (public cloud, private or local environments), guaranteeing governance, lineage and trust at all times ».
Data architecture for AI
Other conclusions of the study include that the hybrid approach in the architecture of Data for AI It has become the norm, offering organizations the necessary flexibility to manage loads in local and cloud environments. Among their main advantages, respondents mentioned security (62%), the improvement of the management of Data for AI (55%) and analytics optimization (54%).
The hybrid approach in the architecture of Data for AI It has become the norm, offering organizations the necessary flexibility to manage loads in local and cloud environments
Regarding storage, 63% claimed to do so in the private cloud, 52% in the public cloud and 42% in a Data Warehousehighlighting the need for them Data for AI are always available and safe.
The integration of AI continues to worry in security matters: exactly half of the respondents said the escape of Data for AI During models training it is an important concern, 48% mentioned unauthorized access and 43% The use of non -safe third -party tools.
Despite these concerns, organizations show confidence: almost a quarter (24%) said they have full security in the protection of their Data for AI53% said they have a lot of confidence and 19% expressed enough confidence in the maturity of their systems.
