IBM Think 2025 felt like a company at the crossroads of clarity and complexity. With Jimit Arora (CEO, Everest Group) and Shirley Hung (Partner) in attendance, the Everest Group team studied IBM’s evolving narrative around enterprise AI, hybrid cloud, and modernization. From the keynote stage to analyst briefings, Think 2025 revealed both IBM’s conviction and its challenges in orchestrating relevance in an increasingly platform-driven enterprise technology landscape.
Shoutout: A special thanks to Beth Bamonte and Warren Lewis from IBM Analyst Relations for curating a seamless, insightful experience. Their thoughtful hosting made the difference.
The big three: Keynotes that framed the IBM story
| Speaker | Key Message | Takeaway |
| Arvind Krishna (CEO) | Positioned IBM as the enterprise AI leader built on trust, governance, and hybrid flexibility, emphasizing the rise of small language models (SLMs), open-source contributions, and digital talent readiness. | Strong alignment with enterprise needs; however, IBM needs to demonstrate developer adoption and AI-led productivity in real-world use cases. |
| Mohammad Ali (SVP, IBM Consulting) | Introduced the IBM Consulting Advantage Platform, a foundational layer for embedding AI across industries, integrating IP, accelerators, partner services, and reusable components. | A cohesive vision for scaled consulting-led AI delivery, but sustained execution and industry adaptation will be critical. |
| Dinesh Nirmal (SVP, Products, IBM Software) | Detailed enhancements across watsonx components – Orchestrate, Assistant, data, governance – and unveiled Granite Tiny for efficient model deployment, with a continued focus on hybrid deployments and open-source model access. | Rich technical evolution across the software stack; integration depth, developer tooling, and cross-portfolio orchestration still have room to mature. |
Software first: My lens on IBM’s announcements
With my focus now squarely on IBM Software, Think 2025 gave me a front-row view into the platform’s depth and direction. IBM is staking its ground on a modular, open AI stack with watsonx, powered by foundation models (Granite), agent-based orchestration, and strong data governance. But there are two sides to this story.
What Worked:
- watsonx Orchestrate brought multi-agent orchestration with 150+ enterprise tool connectors – a practical step toward scalable enterprise automation
- Granite 4.0 Tiny Preview demonstrated a lean, open-source model with an extended context window– a nod to cost-aware AI design
- watsonx.data enhancements made meaningful strides in structuring unstructured data and operationalizing governance
- Focus on SLMs and open-source AI resonated well with enterprise needs for flexibility, efficiency, and local deployment
What could be sharper:
- The deployment maturity of watsonx agents was underexplored, and enterprises need evidence of production-ready benchmarks
- Granite’s evolution roadmap lacks specificity that will provide enterprises needed visibility into support, tuning, and lifecycle
- Interoperability across LinuxONE, webMethods, and watsonx needs a stronger integration story
- Talent readiness and upskilling around new tooling and AI development environments need more emphasis and support mechanisms.
AI at Think: Big themes and IBM’s positioning
| Enterprise Demand Theme | IBM’s Positioning | Everest Group Take |
| Custom AI | watsonx fine-tuning and domain adaptation capabilities | Solid technical base; needs easier tooling for adoption. |
| Open Models | Granite models, open-source availability | Competitive differentiator; brand perception still catching up. |
| Small Language Models (SLMs) | Granite Tiny and domain-specific SLMs | Smart pivot; execution and performance validation are key. |
| AI Governance | watsonx.governance for trust, explainability | Strong architecture; adoption across client base still evolving. |
| Multimodal/Agentic AI | watsonx Assistant, Orchestrate | Agent orchestration is credible; multimodal depth trails top-tier peers. |
| Digital Talent Readiness | Emphasis on tools and platforms to enable AI-native work | Promising narrative; more structured enablement efforts needed. |
The verdict: Progress with caveats
IBM Think 2025 reinforced IBM’s commitment to enabling enterprise-grade AI with modularity and governance at its core. But the path from architecture to adoption remains steep. While IBM Software brought real momentum, the integration gaps and messaging complexity may hinder acceleration. IBM Consulting’s promise of industry-contextualized AI is strong, and the Consulting Advantage Platform is a compelling step, but delivery agility will be key.
If IBM can converge its platforms, simplify developer experience, and lead with proof over pitch, its bets on AI and hybrid cloud could pay off.
Looking ahead: The months leading to the general availability of Granite 4.0 and the broader deployment of watsonx agents will be critical milestones. Also, watch out for the Tech Launch Perspective report I am co-authoring with my colleague, Vishal Gupta, where we will be studying the implications of IBM’s AI announcements.
If you would like to discuss this topic in more depth or have any further questions, please contact Abhishek Singh ([email protected]).
