Main Responsibilities Of A Finance Manager At Microsoft
Andrew, a Finance Manager at Microsoft, primarily manages the gross margin and investment strategy for cybersecurity products, essentially managing a P&L as if each product were a "standalone company," which involves optimizing infrastructure costs on Azure and planning GPU allocation far in advance due to current constraints, especially with the rise of AI products. This also includes pricing, customer telemetry analysis to avoid over-allocation, and investment research for potential acquisitions or organic product development, building financial projections and analyzing consumptive units.
Financial Planning, Cloud Computing, Infrastructure Optimization, Strategic Investment, Product Management
Advizer Information
Name
Job Title
Company
Undergrad
Grad Programs
Majors
Industries
Job Functions
Traits
Andrew Sullivan
Finance Manager
Microsoft
Wake Forest University
Georgia Institute of Technology
Finance
Technology
Finance
Honors Student, Greek Life Member
Video Highlights
1. Managing the Profit and Loss (P&L) for cybersecurity products, treating them almost as standalone companies to ensure they are margin positive.
2. Optimizing infrastructure costs, particularly Azure cloud spend, by collaborating with engineering teams to ensure efficient product operation (e.g., avoiding over-allocation of virtual machines).
3. Planning GPU allocation 18 months to two years in advance due to constrained resources, involving pricing analysis and customer telemetry to efficiently structure services and avoid overuse.
Transcript
What are your main responsibilities within your role?
I primarily manage the gross margin and investment strategy for cybersecurity products at Microsoft. This means I essentially manage the profit and loss for these products.
We often think of our individual cybersecurity products as standalone companies. Our focus is on making each product margin positive. Ultimately, we aim for a mature state where our software achieves high gross margins, typically 80-90%, whether it's a licensing or consumptive business.
A significant part of my role involves working closely with engineering teams to optimize infrastructure costs. All our products are built on Azure, but this principle applies to any cloud environment. We examine our Azure cloud spend and infrastructure to ensure each product is as efficient as possible.
Our goal is to provide customers with everything they need while also running an efficient product on our end. For example, we avoid having excessive virtual machines if fewer can support our customers. We also look into things like skew migrations.
Most recently, we've been heavily focused on GPUs. We have several new AI products that are very GPU-intensive, requiring a significant number of GPUs to operate.
GPUs are currently a constrained resource. Much of my daily work involves planning our GPU allocation 18 months to two years in advance. This lead time is necessary for Nvidia to supply them to Azure, and for Azure to deploy them in our desired data center locations.
For instance, if I needed a GPU in Australia tomorrow, it would take approximately six to 12 months. This is because data centers there are not yet built out, and all available GPU capacity is already allocated. Everyone wants them, so any GPU not personally requested for our teams will be sold to OpenAI through our partnership.
My day-to-day activities include focusing on GPU allocation and its financial implications. This involves pricing, determining current and future GPU costs, and analyzing customer telemetry and usage.
We study how to efficiently structure our services to avoid over-allocating or overusing GPUs. For example, if a customer runs 15 queries a day on a ChatGPT-like service, we calculate the tokens consumed and the number of GPUs needed to support that usage. This analysis is done at a much deeper level, of course.
As part of managing our gross margin strategy, we also handle investment requests and research. This can involve conducting due diligence on acquiring a new company in the cybersecurity space.
Alternatively, it could involve building our own product organically. When building a product, we project associated costs such as headcount, Azure spend, and expenses for other telecom items or vendors. We create profit and loss statements and two-to-three-year projections for these initiatives.
For consumptive products, we forecast monthly consumption units. I estimate this will lead into the next question, but we do a lot of our work in Excel and other Microsoft tools. That's essentially my day-to-day.
Advizer Personal Links