The AI war IBM isn't fighting -- and the one it thinks it can win

The strategy laid out by IBM during its Think conference isn't about competing head-to-head with hyperscalers on every front. It's about owning the enterprise segments where sovereignty, governance and hybrid infrastructure are non-negotiable -- and proving it can deliver at scale in the AI era. 

"The through-line of Think 2026 was IBM's explicit framing of itself as the orchestration and governance layer for the agentic enterprise," said Stephanie Walter, AI stack practice leader at HyperFRAME Research. IMB is making a meaningful strategic choice to be the system that makes heterogeneous AI work inside complex organizations, she noted.  

The other hyperscalers all have their own niches. Microsoft has productivity distribution, AWS has infrastructure and developer depth, and Google has models and AI-native cloud momentum, Walter said. But IBM is standing strong on hybrid enterprise execution. 

IBM isn't trying to out-scale AWS, Microsoft or Google -- it's positioning to manage the complex environments and provide an integrated platform experience tailored to enterprise needs. 

It's not about building agents 

While IBM Think in May featured a lot of focus on AI agents and related technologies, including IBM Bob for more efficient code and watsonx Orchestrate as a centralized control plane for agentic AI systems, that is not where IBM differentiates itself in the market, said Walters. All the major hyperscalers have AI agents, but not all providers are built to handle the complex hybrid environments in which some agents operate. 

IBM's message was that enterprise AI needs orchestration, real-time context, automation, sovereignty and governance built into the operating layer, according to Walter. However, it remains to be seen if IBM can make that architecture easier for customers to deploy, integrate and measure at production scale. 

So, the question isn't who has the best AI agent -- there are already enough competing for attention -- it's who can deploy AI where compliance and legacy infrastructure create constraints, and who has the integrations that can get the job done. 

The hard problem has moved past 'can we build one' to 'who owns the policy, observability and identity layer across all of them.'

"IT leaders [should] watch the move from building individual agents to operating an agentic estate. A lot of enterprises I talk to have deployed agents here or there scattered across multiple business units and using multiple frameworks. The hard problem has moved past 'can we build one' to 'who owns the policy, observability and identity layer across all of them,'" said Michael Leone, vice president and principal analyst at Moor Insights & Strategy. 

IDC predicts that there will be 1.2 billion AI agents in operation by 2029, performing 217 billion daily actions. And the creation of each agent prompts numerous questions to ensure there is proper governance, monitoring and oversight -- which highlights the need for an overall AI operating model and management strategy, according to IBM officials. 

"How are you going to have a control plane to manage those agents? What is the communication that's going to happen between those agents? Who has access to those agents? What kind of data is being accessed by those agents?" asked Dinesh Nirmal, senior vice president of software at IBM, at a Think keynote. 

It's all about connecting the pieces 

IBM has been busy over the past five years acquiring several companies, such as HashiCorp, Apptio and Confluent, to support its hybrid and AI push. However, the goal is to get all these tools to work together as an integrated platform, according to Rob Thomas, senior vice president of software and chief commercial officer at IBM, rather than remain standalone point products that create a fragmented environment. 

One of the biggest focuses that we have in R&D right now … is how do we make this a single experience across … containers to infrastructure automation to real time data and, obviously, agents. I would say, stay tuned on this.

"One of the biggest focuses that we have in R&D right now … is how do we make this a single experience across … containers to infrastructure automation to real time data and, obviously, agents. I would say, stay tuned on this," said Thomas. 

One step toward creating that single experience is IBM Concert, an agentic operations platform that the vendor built from  software it acquired. The tool's main goal is to bring everything together with a shared operational layer. Additionally, its integration of Confluent's streaming capabilities into watsonx.data provides AI systems with real-time context across a hybrid environment. At IBM Think, the vendor started to weave these pieces together. 

"This year, IBM had one coherent platform message. Their AI operating model framing pulled agents through watsonx Orchestrate, data through Confluent, automation through Concert, hybrid through Sovereign Core and the software development lifecycle through Bob into a single picture. It was a clean and integrated version of the IBM story I haven't heard them tell before," Leone said. 

IBM remains consulting-focused 

IBM has a strong consulting arm that accounts for roughly 70% of the business, versus 30% for software, said Thomas. He sees consulting as an advantage that provides insights into what clients are doing, which helps shape IBM's roadmap. Some clients of IBM Consulting highlighted at the conference include SEI and New York Life. 

IBM has demonstrated a structural advantage in regulated industries, mainframe-adjacent workloads and organizations with significant on-premises data estates, according to Walter said. With products such as Sovereign Core and z/OS integrations, paired with the consulting-led deployment model, IBM creates trust relationships that hyperscalers struggle to replicate in healthcare, banking, government and defense. 

"That's not a small market, and it's the market most urgently needing governed AI that works inside compliance boundaries," Walter said. 

Since part of IBM's customer base relies on legacy infrastructure, there needs to be strong integration capabilities to connect all the dots. IBM already has a legacy consultative approach and a track record of wading in to help customers solve complex problems, particularly the integration of technology tools with legacy systems, according to Mark Beccue, principal analyst covering AI at Omdia. 

"I believe IBM's position is strongest where governance, hybrid and consulting overlap. That's exactly where their cloud competitors have the weakest stories," said Leone.  

Informa TechTarget senior news writer Beth Pariseau contributed to this report. 

Kathleen Casey is an award-winning writer and editor covering various IT infrastructure topics, including cloud computing, networking and emerging technologies, for Informa TechTarget.