Artificial intelligence is transforming cloud cost management. Adrian Anderegg, Partner Financial Services at Eraneos, explains why FinOps must evolve to address AI-specific cost drivers, governance requirements, and emerging security risks.
This article was originally published in Netzwoche on July 1, 2026, and is republished here with permission.
AI is changing cloud cost management
The rapid adoption of AI services is fundamentally changing how organizations manage cloud costs. According to the State of FinOps Report 2026, 98% of FinOps teams now manage AI-related cloud spending, compared with just 31% two years ago. As AI adoption accelerates, organizations must expand their FinOps capabilities, strengthen governance, and rethink security.
New cost drivers, new complexity
AI workloads behave differently from traditional cloud services. GPU instances, token-based billing, and dynamic workloads create cost patterns that exceed the assumptions of conventional cloud budgeting.
The cost of a single large language model (LLM) request can vary significantly depending on prompt length, context window, and model selection. Additional expenses such as embedding generation, vector databases, and model fine-tuning further increase complexity. As a result, achieving cost transparency becomes substantially more challenging.
FinOps must evolve
The core principles of FinOps remain valid for AI environments, but they must be extended.
Organizations should focus on:
- Monitoring AI resource consumption, including token usage, alongside traditional cloud resources.
- Establishing clear cost visibility and ownership across teams.
- Defining AI-specific KPIs and incentives to continuously optimize spending.
- Optimizing model selection and implementing caching strategies for recurring requests.
- Introducing AI-specific showback and chargeback models to improve accountability.
- Working closely with business stakeholders to ensure every AI initiative is supported by a measurable business case. If expected value is not achieved, AI usage should be reduced or discontinued.
Denial of Wallet: The emerging security risk
Beyond cost optimization, AI introduces a new attack vector known as Denial of Wallet (DoW).
In these attacks, malicious actors intentionally generate excessive AI requests to increase cloud costs. Public AI endpoints are particularly vulnerable.
Organizations can reduce this risk by implementing:
- Automated anomaly detection for unusual consumption patterns
- API gateways with rate limiting and throttling
- Cloud provider spending limits and budget controls
Security and financial governance must increasingly be considered together.
Governance becomes critical
Without clear governance, organizations risk uncontrolled AI experimentation and rapidly increasing cloud costs.
Companies should establish AI-specific spending policies, particularly for internally developed AI solutions and emerging practices such as vibe coding. Developers need clear guidelines for integrating AI into applications while balancing innovation with financial responsibility.
Cloud Centers of Excellence (CCoEs) should explicitly include AI cost governance and continuously monitor changes to cloud providers’ pricing models. Regular cost reviews help maintain transparency and support informed investment decisions.
Conclusion
FinOps in the age of AI requires more than traditional cloud cost management. Organizations need new capabilities, updated governance frameworks, specialized tools, and close collaboration between technology, finance, and business teams.
The ultimate objective remains unchanged: every AI investment should create measurable business value. Organizations that combine cost optimization, governance, and security will be better positioned to control cloud spending while scaling AI responsibly.
Source
Netzwoche: FinOps 2.0: Wie KI die Cloud-Kostenkontrolle herausfordert
This article was originally published in Netzwoche on July 1, 2026, and is republished here with permission.
