Artificial Intelligence (AI) offers great promise for the finance sector. It can increase efficiency, improve decision-making, and enhance customer experiences. At the recent Gartner CFO Conference, the opening keynote speakers Nisha Bhandare & Clement Christensen highlighted setbacks and stalling in four key areas: regulatory compliance, technological integration, workforce adaptation, and, notably, the escalating costs of AI.
To illustrate the costs of AI, consider the journey of a mid-size bank implementing AI to personalize customer service. This hypothetical scenario, informed by insights from the Gartner conference, leverages advanced cost calculation techniques to overcome these challenges effectively.
“Cost is one of the greatest (near term) threats to the success of AI and generative AI. More than half of the organizations are abandoning their efforts due to missteps in estimating and calculating costs.”
Gartner, 2024
The Initial Excitement and Rollout
Our bank, eager to enhance customer service, decided to implement an AI-driven system for personalized interactions. The initial excitement was palpable; the potential benefits were clear. However, the bank decided not to construct its own large language models but instead acquire an existing service. Yet, the initial rollout quickly revealed substantial costs. Software acquisition, infrastructure setup, and hiring skilled personnel all required significant investment.
Centralizing implementation and vendor management helped handle these initial costs effectively, streamlining the rollout process and reducing redundancy. Cost allocation models distributed the setup costs accurately across various departments, ensuring transparency and accountability. (fig 1: AI Cost Categories, “AI Services (Labor/People)”)
The Rise of Ongoing Usage Costs of AI
As the AI system integrated into daily operations, a new challenge emerged: ongoing usage costs of AI. Unlike traditional technologies, AI’s operational expenses increase with usage. More employees utilizing the system and the rising volume of data processing led to escalating computational costs. As the AI system processed more customer interactions, the bank saw its data processing costs increase by 20%, resulting in an additional $600,000 in annual expenses.
Advanced cause and effect cost modeling allowed the bank to simulate various usage scenarios and forecast future costs. This proactive approach helped the bank prepare for and manage these expenses, preventing budget overruns and protecting the expected return on investment. (fig 1: AI Cost Categories, “I&O/Compute/Cloud”)
For a deeper understanding of how cost and profitability modeling can enhance financial communication and performance, refer to our article on The Role of Cost and Profitability Modelling in Communicating Financial Performance.
AI Experimentation and Its Financial Implications
AI’s journey is rarely straightforward. Continuous experimentation is necessary, and our bank faced several costly iterations while refining the personalized customer service system.
Initial attempts had mixed results, leading to many expensive adjustments. For instance, the bank spent $500,000 on decentralized pilot projects that did not meet performance standards. These failures were part of the learning curve, but they also represented significant sunk costs. Managing all the decentralized, isolated pilots proved crucial.
By employing detailed cost allocation methods, the bank tracked and managed these experimental expenses. This granular approach ensured that each experiment’s financial impact was clear, aiding in better decision-making and resource allocation.
Visualizing these costs of AI helped identify the areas where AI expenses were most significant, allowing for targeted improvements. For example, the bank identified that 30% of the overrun costs were due to inefficient data processing algorithms, which required optimization. (fig 1: AI Cost Categories, “AI Software/Tools/Platform” and “Human Talent & Process”)
Strategic Refocus on AI Investments
Realizing the need for a strategic approach, the bank shifted from running isolated AI pilots to aligning AI investments with broader strategic priorities. Treating AI initiatives as a portfolio of projects allows the bank to focus on efforts that directly contribute to profitable growth and customer satisfaction.
This alignment ensures that resources are directed towards high-impact areas, maximizing the overall benefit of AI investments. By integrating cost calculation models that visualize the cause-and-effect relationships between different investments, the bank could better understand the long-term financial implications of each AI project. As Gartner points out, “Achilles Heels: Costs Can Go Awry by 500%-1,000%,” highlighting the importance of accurate cost estimation and strategic planning.
“Costs Can Go Awry by 500%-1,000%”
Gartner, 2024
Practical Example: Managing the Costs of AI for Personalized Customer Service
Consider the AI-driven personalized customer service project. The initial costs included acquiring sophisticated AI platforms at $2 million and extensive staff training at $500,000. Managing these costs required a comprehensive understanding of each expense component. As the system processed increasing volumes of customer interactions, ongoing usage costs surged. Within the first six months, data processing costs increased by 20%, adding $600,000 to operational expenses.
Advanced cost simulations helped the bank forecast these costs of AI and plan accordingly, ensuring they stayed within budget. During the experimental phase, several iterations failed to meet performance standards, leading to additional expenses of $300,000. However, tracking these costs meticulously allowed the bank to learn from each failure and improve subsequent models.
Accurate cost estimation proved critical. Continuous model updates and compliance with evolving regulations added layers of complexity, costing an additional $200,000 annually. By visualizing cost flows, the bank could identify unexpected cost drivers, such as higher-than-anticipated computational requirements or data storage needs, which added another $150,000 to the budget. This clarity enabled more precise budgeting and ensured that the AI project remained financially viable. (fig 1: AI Cost Categories, “AI-Ready Data”)
For example, the bank discovered that 25% of the cost overruns were due to inefficient use of cloud storage services. By identifying this root cause, the bank was able to optimize their data storage strategy, ultimately reducing these costs by $100,000 annually.
Costs of AI: Categories and Mitigation Strategies
The following table outlines various cost categories associated with AI, based on Gartner’s categories. We have elaborated on these categories, providing concrete examples of typical costs involved, their impact on the budget, and potential mitigation strategies:
Cost Category | Description | Typical Costs Involved | Impact on Budget | Mitigation Strategies |
---|---|---|---|---|
Risk Management | Costs related to identifying, assessing, and mitigating risks associated with AI implementation. | Risk assessments, compliance audits, insurance premiums | High if risks are not managed | Regular risk assessments, compliance training |
Environmental | Costs associated with the environmental impact of AI infrastructure and operations. | Energy consumption, cooling systems, carbon offset programs | Moderate to high | Energy-efficient hardware, renewable energy sources |
AI Services (Labor/People) | Costs for hiring and retaining skilled personnel to develop, implement, and maintain AI systems. | Salaries, training programs, recruitment costs | High due to demand for AI talent | Continuous learning programs, competitive compensation |
I&O/Compute/Cloud | Infrastructure and operational costs, including cloud services, hardware, and data centers. | Cloud storage fees, server costs, networking infrastructure | High, especially with scale | Optimize usage, negotiate with cloud providers |
AI Software/Tools/Platform | Costs for acquiring AI software, tools, and platforms required for development and deployment. | Licensing fees, subscription costs, integration expenses | Moderate to high | Evaluate ROI of software, leverage open-source tools |
Human Talent & Process | Costs related to the processes and workflows needed to support AI initiatives, beyond direct labor costs. | Process optimization, change management, collaboration tools | Moderate | Streamline processes, invest in process automation |
AI Governance & Security | Costs for establishing and maintaining governance frameworks and security measures for AI systems. | Security audits, governance frameworks, regulatory compliance | High due to stringent requirements | Implement robust governance and security frameworks |
AI-Ready Data | Costs for preparing and managing data to be used effectively in AI models, ensuring it is clean and reliable. | Data cleaning, data integration, data storage, data annotation | High if data quality is poor | Invest in data management tools, ensure data quality |
Ensuring AI Cost Transparency and Profitability
AI presents both opportunities and challenges for the finance sector. Our bank’s journey, informed by insights from the Gartner CFO Conference, highlights the importance of managing hidden costs of AI. By employing advanced cost calculation techniques, the bank uncovered hidden costs, optimized investments, and ensured sustainable AI adoption. This proactive and strategic approach enabled the bank to achieve greater cost transparency, enhance profitability, and make informed decisions, driving sustained financial health and competitive advantage in the dynamic financial landscape.
CostPerform is the ideal software for this process. Its advanced cost calculation capabilities, intuitive interface, and powerful visualization tools enable organizations to manage costs at a very granular level. This includes understanding the drivers behind AI implementation costs and calculating the resulting costs per department, per product, per channel, or even per client.
By uncovering hidden costs, optimizing resource allocation, and providing clear insights into cost drivers, CostPerform helps financial institutions manage AI adoption and achieve their goals. For any organization looking to harness the power of AI while maintaining financial control, CostPerform offers the tools and expertise necessary to succeed.
Source: Highlights From Gartner CFO & Finance Executive Conference