In this blog post, “What Microsoft Means by FinOps for AI,” I will explain why every business needs a strategy for AI spending.
Traditionally, and since the release of Microsoft 365 subscription-based licensing, we have become accustomed to buying a license with a predictable monthly cost.
The model worked well and formed the core foundation of many SaaS companies, which operated on the same monthly subscription principle.
AI is starting to change that model, as we discussed in our article about the new Copilot and advanced AI usage.
With the release of more advanced AI capabilities and services, the monthly licensing model no longer works for the following reasons:
- AI agents require underlying infrastructure that costs money.
- AI agents consume tokens, which cost money.
- Not all agents are the same, and some require different models with different base costs.
- AI usage is starting to resemble cloud infrastructure services such as Azure, AWS, and GCP.
Taking all of the above into account, it is clear that there is no way Microsoft or other AI vendors can continue supporting the old monthly subscription model when so many variables can affect the cost of the service.
Introducing FinOps
With AI costs becoming a metered service, every business needs to pay close attention to ongoing costs and put a strategy in place to ensure that AI activities deliver value and clear business outcomes.
To navigate the headwinds of high AI costs, SMBs and enterprises need to start planning their spending policies, limits, and consumption plans, and place a lid on unrestricted access to advanced models.
Use Case
At CPI, we engage with many SMBs and help them design and implement AI spending policies that allocate spending according to business needs and product delivery.
In a recent engagement, we implemented a spending strategy that prioritized the AI budget for teams with a clear outcome and a product that translated into real business value.