Cloud Financial Management: Getting a Little Help From AI/ML

Artificial intelligence and machine learning are opening opportunities for organizations to program how they want to scale up or down use of cloud resources depending upon current and anticipated future demand.

Nathan Eddy, Freelance Writer

December 2, 2022

5 Min Read
digital cloud abstract
SergiyTimashov via Adobe Stock

Although speed, agility and scalability are major goals of cloud migration, a topmost goal for many enterprises is to reduce costs.

Unfortunately, many organizations experience sticker shock when they receive hard-to-decipher bills with thousands of line items from cloud providers.

In addition, many organizations are migrating to multiple cloud platforms, each offering different options and feature (and cost) combinations -- it is hard to know whether you are getting the best price or value.

“Cloud service providers offer proprietary tools for tracking expenses, but organizations continue to lack granular visibility into cost origins and which operations are driving increases,” explains David Stodder, TDWI's senior director of research for business intelligence. “AI/ML-infused third-party tools can offer that granular visibility.”

He points out forecasting is a critical activity in cloud financial management, and AI/ML can bring predictive modeling and insights to forecasting, enabling users to examine bigger and more diverse data.

“They can continuously analyze whether forecasts are accurate and aligned with business requirements,” Stodder says. “AI/ML inside forecasting contributes to development of overall plans for managing cloud costs and how to adjust when workload demands increase.”

AI/ML Helps Capacity Management, Cost Optimization

Meanwhile, automated forecasting strengthened by predictive analytics and AI/ML enables organizations to move toward continuous forecasting, resource optimization, and capacity management.

Gartner VP Analyst Adam Ronthal explains cloud is essentially a massive cost optimization problem.

“We are trying to simultaneously cost optimize tens to hundreds of service offerings, each of which is interconnected,” he says. “So, making a change to one component will impact anything that touches that component, and will have second and third order effects as well.”

On the infrastructure side, there are hundreds of different machine instance types to choose from, so there is the complex problem on a complex infrastructure -- which is perfect for AI/ML to solve.

“In fact, it’s complex enough that it’s nearly impossible to get it right without AI/ML,” Ronthal says. “The application of AI/ML capabilities to cost optimization and budgeting problems is called augmented FinOps. It’s an emerging technology still in relatively early stages.”

He says most specific AI/ML tools are tightly scoped on solving a specific part of the problem.

For example, Oracle Autonomous Database uses AI/ML to optimize performance and security of Oracle Database, while vendors like Sync Computing optimize Spark and Databricks workflows.

Meanwhile, vendors like OtterTune and Enteros optimize database management systems (DBMS) performance.

“Other tools are broad in scope but fairly shallow,” Ronthal says. “They give you a view of where the financial hot spots are, but they aren’t making detailed prescriptive recommendations on how to solve them.”

From the perspective of Bret Greenstein, Partner, Data & Analytics, PwC, there are two great uses for AI/ML in managing cloud costs.

“The first is in automation, using AI/ML to speed up service requests and to automate the workflows that go into service requests and monitoring of cloud environments,” he explains. “The second is for prediction and optimization.”

Greenstein notes companies are using natural language processing (NLP) to manage service requests automatically (access control requests, provisioning requests, outages, etc) to operate in real time with less cost.

When it comes to optimization, AI/ML can be used to anticipate peak loads and to make decisions on the optimal instances and places to run work to balance cost, performance, and capacity.

Multiple Stakeholders Must Collaborate

Ronthal explains as the more the center of gravity for data and analytics shifts to the cloud, the more involved those parts of the organization that care about operational efficiency become: the CFO and COO.

“Generally speaking, the CFO and COO roles don’t have the deep technical capabilities to understand the value of the workloads that are run, so they will partner with CDAOs, CIOs, and line of business directors,” he says.

The successful organization will establish clear lines of communication between each of these leadership roles: CFO/COO, CDAO, CIO, LOB Director.

“However, given that cloud is essentially all about cost optimization, the CFO will become the dominant role,” he adds. “This problem requires multiple engaged stakeholders.”

Greenstein says responsibility for developing a strategy typically sits with the AI and emerging technology leaders and CIOs, but with any transformation effort they may further involve CDO’s who need to manage the accessibility, cost, and security of data across the enterprise.

“In addition, the application leaders work with business stakeholders to understand the tradeoffs between cost, performance and speed for application workloads,” he adds.

Stodder agrees for cloud computing, all need to contribute to collaborative management.

“Cloud computing is often business-driven, requiring business-side contribution to leadership,” he says. “Data scientists need to be involved to guide use of AI and AI-infused tools.”

He adds business users (including the office of finance), data scientists, application development, and IT. Some organizations bring them together in center of excellence committees.

“They manage budgeting for advancing data collection and tooling for AI-based cloud finance management,” he says.

Evolution of AI/ML Tools to Aid Planning

Greenstein says the future of AI/ML tools will make it easier to drive automation and optimization decisions.

“As cloud computing becomes more powerful, and applications, analytics, and products use it more, there will be more potential for optimization of workload sizes, locations, priorities to maximize NPS scores, improved business outcomes, and energy consumption along with cost, performance and speed,” he says.

Imagine, for example, that a retail business is about to hit a spike in demand due to an event on social media.

Using AI/ML, the business could predict that demand and provision increased capacity.

Based on the cost of doing that, it could advise if it is better to grow the environment to meet increased demand (to get the higher revenue and NPS), and that the energy impact of doing this would be accounted for.

“This in turn would drive other actions in IT to offset the energy consumption,” Greenstein explains. “With AI/ML this can happen faster, allowing dynamic decisions and avoiding costly mistakes.”

What to Read Next:

Fintech, Cloud, and Finding Ways to Bridge the Skills Gap

How to Budget Effectively for Multi-Cloud

Why the Financial Services Industry is Embracing the Cloud

Special Report: How Fragile is the Cloud, Really?

About the Author(s)

Nathan Eddy

Freelance Writer

Nathan Eddy is a freelance writer for InformationWeek. He has written for Popular Mechanics, Sales & Marketing Management Magazine, FierceMarkets, and CRN, among others. In 2012 he made his first documentary film, The Absent Column. He currently lives in Berlin.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like

More Insights