IBM Think 2025: Enterprise Gen AI Moves Beyond the Hype

IBM used its annual Think conference on Monday to introduce new generative AI features designed to help large enterprises overcome the practical challenges of scaling AI in complex environments. The updates include an expansion of its watsonx platform, a tighter integration with Red Hat OpenShift, and new AI assistants aimed at making the technology more accessible to developers and business users alike.

The announcements reflect IBM’s strategic focus on hybrid cloud infrastructure and open-source compatibility as the foundation for trustworthy, enterprise-grade AI. According to IBM Chair and CEO Arvind Krishna, success in generative AI will depend less on experimental models and more on the ability to integrate AI with existing IT systems and domain-specific data.

“You want a model that works with your business,” Krishna said during a keynote presentation. “You want to run it in your environment, and you want to make sure your data stays yours.”

Watsonx Expands to Meet Enterprise Demand

IBM launched watsonx a year ago as a platform to help organisations build, train, deploy, and govern AI models. At Think 2025, the company introduced several updates, including InstructLab, a new open-source technology developed with Red Hat. InstructLab enables organisations to customise open-source foundation models using their own data and subject-matter expertise—all while preserving model integrity and accelerating time to value.

Also new is IBM Granite 13B, a family of open-source language and code models released under an Apache 2.0 licence. These models were trained on enterprise-focused datasets and will be integrated into watsonx, supporting more secure and tailored deployments across industries.

“We’re at an inflection point where enterprises need models that are not only powerful but also efficient and responsible,” said Dinesh Nirmal, Senior Vice President of Products at IBM Software. “Granite 13B is designed to help businesses get started faster without sacrificing control or compliance.”

AI Assistants Embedded Across the Stack

IBM is also betting on domain-specific AI assistants to drive adoption. The company introduced new and updated watsonx assistants for coding (Code Assistant), customer service (Virtual Agent), and IT operations (AIOps). These tools are built to run across hybrid cloud environments and integrate with enterprise systems like SAP, Salesforce, and ServiceNow.

The latest update also includes significant agent capabilities within watsonx Orchestrate, enabling businesses to deploy and manage intelligent agents that interact with a wide range of applications and services, further enhancing automation and efficiency across the enterprise. The agent-driven approach is already showing strong results in automating workflows in industries like finance, healthcare, and retail.

IBM has further solidified its position in the hybrid cloud space with integrations extending beyond traditional enterprise systems. The company now offers deeper integrations with AWS and Salesforce, streamlining the deployment of AI-powered services across multi-cloud and hybrid environments.

Early client engagements suggest significant productivity gains, particularly in automating repetitive tasks and generating boilerplate code. IBM Consulting, which is working with more than 1,000 clients on Gen AI projects, will use the updated watsonx capabilities to help businesses build out proprietary AI solutions. The consulting division has grown its AI expertise significantly over the past year, with 40,000 practitioners now trained in AI and more than 1,700 client engagements underway.

The Hybrid Cloud Advantage

The company’s hybrid cloud strategy underpins its Gen AI efforts. IBM has tightly integrated watsonx with Red Hat OpenShift, enabling AI workloads to run flexibly across on-premises, private, and public cloud environments. This approach is designed to give enterprises more control over where their models are trained and how their data is used—a key requirement in regulated industries like banking, healthcare, and government.

“Clients want the ability to run AI where their data resides,” said Rob Thomas, Senior Vice President of Software and Chief Commercial Officer at IBM. “Hybrid cloud is what makes this possible—and scalable.”

Further strengthening its hybrid integration strategy, IBM has collaborated with HashiCorp, CoreWeave, and Intel and NVIDIA to provide more robust infrastructure support for enterprises running AI models in multi-cloud environments. These collaborations are aimed at improving performance, scalability, and security for enterprise AI workloads, ensuring that IBM’s solutions meet the growing demand for high-performance compute resources in the enterprise AI space.

IBM also reaffirmed its commitment to AI governance, citing its AI & Data Platform governance tools and its contributions to responsible AI frameworks. Features like model monitoring, bias detection, and data lineage are increasingly seen as prerequisites for adoption, especially in industries facing stringent compliance obligations.

In line with these efforts, IBM announced IBM LinuxONE 5, a powerful new platform designed to support the most demanding AI workloads with unmatched security and performance. LinuxONE 5 can process up to 450 billion AI inference operations per day, making it an ideal solution for businesses looking to scale their AI applications while maintaining stringent security standards.

Finally, IBM announced its intent to acquire DataStax, a leader in enterprise data management, in a move designed to enhance its capabilities in managing large-scale, unstructured data. DataStax’s expertise in real-time data streaming will integrate into IBM’s broader AI strategy, particularly with watsonx, enabling more advanced generative AI applications.

A Shift Toward Practical AI

IBM’s emphasis on integration, openness, and governance marks a contrast with some of the more experimental or consumer-oriented uses of generative AI. Rather than chase headline-grabbing chatbots, IBM is positioning itself as the vendor of choice for CIOs and CTOs who need scalable, secure, and enterprise-specific solutions.

At Think 2025, the message was clear: AI is no longer a futuristic add-on — it is becoming embedded in how business gets done. But success will depend on solving unglamorous but essential challenges: connecting data, ensuring privacy, managing costs, and proving ROI.

As Krishna put it: “It’s not about the model. It’s about the end-to-end capability. That’s where the value is.”

Read more: AI unlocks $15.7 Trillion opportunity, but hurdles await Africa