About
As enterprise AI transitions from experimental Proofs of Concept (POCs) into real implementations, digital leaders are encountering a shared challenge: consumption-based public cloud token billing models do not fit neatly into traditional IT budgeting.
Daily token consumption can rise unexpectedly, pushing expenses through the roof while arbitrary caps quickly kill momentum for innovation. For many CIOs, this raises an open strategic question: How do we keep costs under control while further stimulating innovation?
During this interactive CIONET roundtable, we will explore among peers how to navigate AI tokenomics. Rather than treating public cloud LLMs as the default path, we will investigate how different execution environments, ranging from hybrid cloud setups and targeted on-premise models to specialised edge hardware, can help balance cost with performance.
This session is intentionally designed as an open peer-to-peer exchange; a space to compare notes, challenge current ways of working and architectural setups, and discuss practical approaches to building predictable, scalable AI models.
Who This is For:
CIOs, CTOs, Heads of Infrastructure & Platforms, Heads of Data & AI, Enterprise Architects, and technology leaders looking to explore strategies for building and scaling AI capabilities while maintaining financial and operational control.
Key Questions & Discussion Themes:
- Sharing experiences on where public cloud LLM billing creates friction as POCs move to production and scale.
- How are peers monitoring consumption and establishing guardrails without capping innovation?
- How to compare costs across public platforms, or build business cases for edge devices or on-premise infrastructure setups.
- How leaders use prompt optimisation, caching, batch processing and dynamic multi-model routing to improve efficiency and reduce costs.
- What architectural implications arise from AI cost optimisation? How do factors such as the price difference between input and output tokens affect application design?
- Taking into account data residency requirements, intellectual property protection and other data management aspects to help ensure compliance and manage risks.
What We Will Explore Together:
- Real-world insights from peers on managing token consumption volatility.
- Practical approaches for categorising AI uses and workloads to achieve optimal price-performance ratios.
- Perspectives on hybrid deployment options, including on-premise AI factories and edge hardware.
- Controls and guardrails to help ensure proper data protection and governance while keeping costs predictable.
.png)
.png)