AI and Hybrid Cloud – IBM Doubles Down On Core Strategy at IBM Think 2025

Focusing on IBM's core AI platform, watsonx, the growing portfolio is noteworthy. IBM has integrated much of its developing data fabric capability into watsonx Data Intelligence. Furthermore, the new watsonx Data Integration brings data pipeline capabilities directly within the platform. This also includes the bundling of functionalities from DataStage for batch processing and StreamSets for streaming data integration.

A significant emphasis is being placed on making unstructured data a first class element within watsonx.data. Leading this effort are new entity extraction capabilities designed to populate metadata, thereby enabling more deterministic queries of unstructured data. While the discussion around governance of unstructured data is ongoing, IBM is providing support for Retrieval Augmented Generation (RAG) through Milvus, with DataStax vector stores on the horizon. My analysis suggests that a knowledge graph would be a valuable addition here. Let's also acknowledge the inclusion of streaming analytics. Perhaps a reevaluation of the watsonx.data branding is in order, given its expanding scope.

A particularly interesting development is IBM's response to Google Cloud's AI Agent initiatives. IBM presented a well considered approach to AI Agents under the watsonx Orchestrate product suite. IBM clearly understands the complexities involved in AI agents from the detailed AMA with Dinesh Nirmal. It's not just about the AI itself. Orchestration and observability are equally critical. The most significant aspect, in my estimation, is the announced integration of these AI agents with seven major enterprise software players. These include Salesforce, Microsoft, Adobe, ServiceNow, and Oracle. These are direct integrations, mind you. This isn't just an agent to agent API. This means that users building AI Agents with IBM will have native integrations with some of the largest enterprise software platforms globally. This enables the execution of multi agent and cross software tasks. This is rather splendid.

IBM's strategic narrative at Think 2025, articulated by Arvind Krishna, centered on the idea that AI is no longer a future aspiration but a present day "productivity engine" for businesses. Krishna stressed the need for enterprises to fully leverage their data and systems through widespread AI adoption. IBM's focus is on providing platforms and solutions that enable businesses to realize tangible value from their AI investments, put simply, IBM is focused exclusively on enterprise AI and not flashy consumer deployments. This emphasis on practical application and return on investment appears to be a direct response to the growing demand for demonstrable business outcomes from AI initiatives.

A key element of IBM's strategy acknowledges the challenges many organizations face in achieving the anticipated returns from their AI investments. Success rates, according to Krishna, hover around 25%. He suggests that by adopting the right approach, utilizing IBM's integrated platforms and open ecosystem, this success rate could potentially double or even triple. This statement reflects IBM's confidence in its ability to guide enterprises toward more successful AI deployments through a comprehensive and integrated technology stack. Furthermore, Krishna highlighted the fundamental role of hybrid cloud in enabling successful enterprise AI, emphasizing the necessity for seamless integration and orchestration across increasingly fragmented IT environments. This reinforces IBM's long standing commitment to hybrid cloud as a central pillar of its overall strategy.

IBM's approach at Think 2025 was distinctly oriented towards addressing the specific needs and challenges of enterprise clients. The company framed its announcements around key pain points, such as the existence of data silos that impede effective AI adoption. The unveiling of watsonx.data with its data fabric capabilities directly addresses this issue by aiming to unify and govern data across diverse sources, making it more readily accessible and usable for AI applications. Recognizing the complexities of integrating disparate systems in hybrid cloud environments, IBM also launched webMethods Hybrid Integration to streamline connectivity across various applications, APIs, and data sources. This focus on simplifying integration underscores IBM's understanding of the practical challenges enterprises encounter when implementing modern technologies. Moreover, with the escalating threat landscape, IBM emphasized security and resilience in its offerings, exemplified by the the introduction of IBM Concert Resilience Posture. This announcement illustrates IBM's commitment to providing solutions that address the multifaceted needs of enterprise clients, spanning data management and integration to security and business continuity.

A notable aspect of IBM's strategic narrative was the articulation of the "small is the new smart" doctrine, championed by Arvind Krishna and others across keynotes and the Analyst only AMA’s. This strategy emphasizes the effectiveness of smaller, domain tuned AI models, such as the Granite family, in delivering comparable or even superior accuracy to massive general purpose models, while offering significant advantages in terms of inference costs and hardware requirements. The release of Granite 4.0 Tiny Preview on Hugging Face under an open source license exemplifies this approach, allowing developers to experiment with a compact yet powerful language model. Krishna argued that these smaller, specialized models are better suited for enterprise use cases, where specific domain knowledge and cost efficiency are critical. This strategic bet differentiates IBM from competitors heavily invested in large foundation models and suggests a focus on practical, use case specific AI deployments that can be more readily integrated into existing infrastructure. By making the Granite 4.0 Tiny model accessible on consumer grade GPUs, IBM aims to broaden access to advanced AI capabilities, potentially fostering wider adoption among a diverse range of developers and organizations.

Furthermore, IBM consistently emphasized the importance of moving beyond initial AI experimentation towards the practical integration of AI into existing enterprise workflows and systems. The introduction of watsonx Orchestrate, with its capability to rapidly build custom AI agents and its extensive integration with over 80 leading enterprise applications, directly addresses this need for seamless integration. The availability of pre-built domain agents within watsonx Orchestrate further facilitates the practical application of AI by providing ready to use solutions for common business functions. Similarly, the launch of webMethods Hybrid Integration underscores IBM's commitment to providing the necessary tools for managing and automating integrations across the diverse IT landscapes prevalent in most enterprises. This strategic focus on integration reflects a maturing AI market where businesses are now seeking tangible returns from their AI investments, requiring AI to be seamlessly woven into their existing operational fabric.