A three-part series from Momenta on how industrial companies build operating capability in an age of Agentic AI and geopolitical pressure.
Many industrial AI pilots work, yet few scale. Post 3 of Momenta’s series explains why governance and decision rights block AI from becoming operational.
Why most industrial AI pilots fail, even when the technology works
Most industrial companies can point to successful AI pilots.
Predictive maintenance cut downtime. Scheduling tools boosted throughput. Optimization models performed well in controlled environments.
Yet very few pilots become standards. As of August 2025, Forbes reported a staggering 95% failure rate of corporate AI initiatives. The majority failed at the integration stage.
The usual explanations blame scale, data quality, or integration headaches. In practice, the real bottleneck sits elsewhere:
Decision Rights.
Agentic AI introduces systems that observe, decide, and act. Most organizations, however, never redesign who can decide, on what data, and under which conditions those systems can act autonomously. Approval chains stay intact. Accountability blurs when automated actions meet safety, quality, or commercial commitments.
As a result, AI can recommend but not execute.
This shift from insight to action reflects a deeper transition we described in our recent post, From AI that Talks to AI that Does. Asset-intensive industries are moving beyond the Inference Economy of dashboards and alerts into an Orchestration Economy, where teams of AI agents monitor, decide, and act across production processes and ecosystems.
Many organizations stall at this transition. They prove intelligence in isolation but never redesign how decisions actually flow through operations.
Workflows fragment. Operators swivel between dashboards and spreadsheets. Escalations pile up. Automation slows instead of accelerating.
Across industrial operations, the constraint is not intelligence. The constraint is governance.
Some organizations are starting to break the cycle by redesigning their operating capabilities rather than launching more projects.
A practical example is AMESA, a Momenta portfolio company built for industrial realities. AMESA links simulation, orchestration, and live operations so teams can train digital workforces under real-world constraints before deployment.
Planners and operators can test new ways of running lines and networks without risking throughput, quality, or customer commitments in live production. Learning moves from slide decks into daily operations.
In the next cycle of industrial disruption, advantage will not go to the most intelligent systems. It will go to organizations that can let intelligence operate without disrupting plants, networks, or P&L.
That is what operating capability really means.
This concludes our series on Agentic AI, geopolitics, and operating capability.
At Momenta, we work with industrial leaders and technology ventures on exactly these operating challenges.
_____________
Momenta is the leading Industrial Impact® venture capital firm, accelerating industrial innovators across energy, manufacturing, smart spaces, and the supply chain. Our team of deep industry operators has helped scale industry leaders and innovators to improve critical industries, the environment, and people's quality of life for over a decade. PitchBook has ranked Momenta as one of the ten best-performing venture capital firms for 2023 in its prestigious Global Manager Performance Score League Tables, and the firm is the only European-headquartered VC to secure a Top 10 spot on the list. For more information please visit http://www.momenta.one.
Schedule a call to learn more about our Industrial Impact investments.