AI adoption has outrun AI payoff. IBM used its Think 2026 conference in Boston to announce new agent orchestration, real-time data, intelligent operations and sovereignty products aimed at enterprises trying to move AI from pilots into governed production systems.
The product push focuses on four systems IBM says enterprises need to operate AI at scale: agents, data, automation and hybrid cloud.
IBM Chairman and CEO Arvind Krishna tied the portfolio to a wider operating-model shift. “The enterprises pulling ahead are not deploying more AI — they’re redesigning how their business operates,” he said.
High spending meets stagnant returns
According to IBM’s CEO study, only around 25% of AI initiatives deliver expected ROI and just 16% have scaled enterprise-wide.
Morgan Stanley’s research points to the same pressure from another angle: only 21% of S&P 500 companies now cite at least one AI benefit, while companies showing measurable results are seeing cash-flow margin expansion at roughly twice the global average.
The spending side is moving faster. Goldman Sachs’ baseline model implies $765 billion in annual AI CapEx in 2026, rising to $1.6 trillion in 2031, across compute, data centers and power.
At the enterprise level, a16z’s survey of 100 VP- and C-level respondents at Global 2000 companies found average LLM spend rose from about $4.5 million to about $7 million over two years, with respondents expecting another 65% rise to about $11.6 million this year.
That split between spending and proof is the gap IBM is trying to occupy. Deloitte’s 2026 State of AI in the Enterprise report found that 25% of respondents had moved 40% or more of their AI experiments into production, while 54% expected to reach that level within three to six months.
Orchestrating the agentic control plane
IBM’s next-generation watsonx Orchestrate, now in private preview, is built for that scaling problem. IBM described it as an agentic control plane that lets organizations deploy agents from different sources with consistent policy enforcement and accountability.
The company also pointed to IBM Bob, now generally available, as an agentic development partner for enterprise teams building agents with security and cost controls.
Solving the data and infrastructure bottleneck
The data layer is the second part of the rollout. IBM is using its completed Confluent acquisition to connect real-time event streaming with watsonx.data, Kafka and Flink-based data flows, and a new context layer that applies semantic meaning, enforces governance at runtime and supports explainable AI decisions.
IBM also said a proof of concept with Nestlé delivered 83% cost savings and a 30x price-performance improvement on a global data mart spanning 186 countries.
The firm’s recently unveiled IBM Concert product claims to extend the same operating model into infrastructure and security operations. Available in public preview, the platform aims to correlate signals across applications, infrastructure and networks without requiring enterprises to replace existing tools.
IBM Concert Secure Coder, also in public preview, is marketed to embed security management into developer workflows and can generate remediations for vulnerable code, operating systems, middleware, packages and images.
Navigating sovereignty and agent sprawl
It also announced IBM Sovereign Core, which it says will bring the governance argument into regulated and cross-border environments. IBM described the platform as a way to embed policy at the infrastructure runtime level, with workload portability across hybrid and partner environments.
A separate IBM release lists customer-operated control, in-boundary identity, encryption and data services, continuous compliance monitoring, audit evidence generation and governed AI execution among its capabilities.
According to Gartner, more than 40% of agentic AI projects will be canceled by the end of 2027 because of rising costs, unclear business value or inadequate risk controls.
Gartner also warned last week that an average global Fortune 500 enterprise could have more than 150,000 agents in use by 2028, while only 13% of organizations believe they have the right AI agent governance in place.
That structure also reflects the gap identified in Writer’s 2026 enterprise AI survey, which found that 97% of executives reported benefits from AI but only 29% saw significant ROI from generative AI and 23% from AI agents.