CFO on the Spot: Five minutes with Eric Muhlheim, CFO of Mozilla

I studied mathematics in college but quickly realized that theoretical math was not where I wanted to build my career. I wanted to apply that training in an environment that felt more like a team sport, which led me to finance and an early career in investment banking.

After business school, I joined The Walt Disney Company. While working in its consumer products division, I helped develop Disney English, an English-language education business for children in China. I spent three years there helping grow it from an idea into approximately 18 schools operating across five cities. I was effectively the CFO of that startup, and the experience showed me that I wanted to pursue the CFO role.

When I returned to the United States, I moved into CFO positions at a series of technology companies in education technology, advertising technology, and digital media. That path eventually brought me to Mozilla, where I also found an opportunity to contribute to a mission-based organization.

I view AI as a new industrial revolution and a tool designed to make people more productive. At Mozilla, we consider both how we use AI internally and how we incorporate it into our products.

Internally, the decision ultimately comes down to return on investment. We ask what output we expect from a particular team or product and whether AI will help us produce more or accomplish the same work with fewer resources. It then becomes a matter of tracking the investment and understanding what we are trying to achieve strategically.

When incorporating AI into Firefox, we also consider whether it adds value to the browsing experience. Some users want AI capabilities, while others do not. We therefore try to provide choices, including the ability to block AI-enhanced features or select the large language model that powers certain features.

There is no single metric that applies to every AI investment. CFOs need to begin by identifying what the company is trying to accomplish and then connect the investment to business metrics that matter.

AI can help people create a great deal of output, but more output does not necessarily produce better business results. The important question is whether the technology is improving an outcome the organization actually values.

CFOs must also attribute spending to specific workflows and use cases. Different models have different capabilities and costs. A company may not need the latest frontier model to draft an email or revise marketing copy, but it may need a more advanced model for complex technical work. Business leaders need visibility into those choices so they can use their budgets effectively.

First, identify the legitimate metrics and key performance indicators the project is expected to improve. Then determine whether the AI application is producing those results consistently and without creating excessive rework.

Consistency is especially important in finance because financial results cannot change each time a process is run. AI operates as a probability machine, which allows it to interpret uncertain or unstructured information but can also lead to inconsistent results. 

CFOs should track AI usage, rework, costs, and whether the technology is delivering the intended output. Managers also need visibility and feedback so they can identify mistakes and optimize how AI is used. 

Finance should provide guardrails by helping the organization determine how much it can afford to spend and where that spending should occur. However, I also see finance as a navigator, facilitator, and partner rather than simply an enforcer.

Finance can create central visibility into AI spending, identify the largest pools of expense, and look for opportunities to negotiate contracts or route work toward less expensive models. It can then provide managers with dashboards showing how their teams are using AI and what those activities cost.

The goal is to give business leaders an appropriate financial envelope and allow them to decide how to use it most effectively. Finance brings a different perspective and line of sight, but business leaders share the same objective of helping the company succeed cost-effectively.

Mozilla Corporation is owned by a nonprofit organization devoted to the health of the internet. As a result, our financial decisions are not based solely on quarterly earnings or an investment timeline.

We operate with a double bottom line. We must remain financially sustainable and contribute to Mozilla’s broader mission. That mission includes ensuring that Firefox has an impact on the internet and that Mozilla participates in conversations about web and AI standards.

This structure gives us the ability to sacrifice some short-term profitability when an investment supports our mission. As we add AI to the browser, maintaining users’ trust and providing them with choices are central to how we define long-term success.

The responsibility is to create room for experimentation while maintaining discipline around the core business. We define our lines of business, establish investment pools, and monitor the profitability and performance of each area.

New projects move through stages. We assess whether they are becoming viable, growing quickly enough, and aligning with Mozilla’s mission. Some projects may be interesting but fail to progress as expected, which means we sometimes have to stop work people care about and redirect resources toward opportunities with greater long-term potential.

Finance helps turn those numbers into a story the leadership team can use to make difficult decisions. The goal is to keep investment visible and closely aligned with the company’s strategy.

New technologies naturally create excitement, and AI has been accompanied by considerable hype about what it can do. That has encouraged companies to experiment and, in some cases, maximize their use without fully understanding the costs.

Many CFOs have seen those expenses rise quickly. The next step is to ask what return the company is receiving, whether it is spending responsibly, and what framework will help employees make better decisions.

At Mozilla, we are not trying to build our own large language model. We are focused on investing in AI features that provide useful experiences for Firefox users and create value without spending simply for the sake of participating in the trend.

Go back to the basics and ask what you are trying to accomplish. AI is a tool for making people more productive, so do not become distracted by everything it could potentially do. Start with the business goal, then determine how AI can support it cost-effectively.

Consider multiple models and delivery options because each has different capabilities and costs. Avoid becoming trapped in one solution, and ensure that the people using AI understand the financial trade-offs.

Finally, keep humans involved in important processes and pay close attention to privacy, security, and responsible data use. AI can increase productivity, but people remain essential for ensuring that the technology produces the intended result without creating unacceptable risks.