Few organizations buy their AI from a single source. Spend spreads across multiple model providers, several AI-enabled SaaS products, and a scattering of developer tools, each with its own billing portal, its own export format, and its own review cycle. No single vendor dashboard shows the total, which is exactly why the total is so often a surprise.

Why multi-vendor spend hides

Each vendor shows you only your spend with them. The model provider shows API usage. The SaaS tool shows a subscription. The developer tool shows seats. These land in different expense systems, under different cost codes, reviewed by different people. No one is looking at the sum, so the sum grows unwatched.

Build one normalized view

The fix is a single view that normalizes across vendors. Pull each vendor's spend into a common structure: vendor, spend type (subscription or consumption), monthly cost, owning team, and use case. The structure matters more than the tool holding it. A maintained spreadsheet that aggregates every vendor beats a sophisticated dashboard that only sees one.

Separate the two spend types

Multi-vendor spend comes in two shapes that behave very differently. Subscription spend is predictable and headcount-driven. Consumption spend scales with usage and can swing month to month. Tracking them together as one number hides both. Separate them, and a seat problem and a consumption problem become two distinct issues with two distinct fixes.

Attribute across the whole picture

Attribution has to span vendors, not sit inside one. Map every subscription seat and every API key to a team and cost center, regardless of which vendor it belongs to. Once every line has an owner, you can produce a per-team AI cost that sums across all vendors, which is the number a team lead can actually act on.

Find the overlap

Tracking across vendors surfaces something a single dashboard never will: duplication. Most organizations running a multi-vendor audit discover they are paying two or three vendors for tools that do roughly the same job, adopted by different teams at different times. Consolidating the overlap saves money and simplifies every future tracking cycle.

One picture, reviewed on one cadence

The goal is a single monthly picture that shows total AI spend, broken down by vendor, by spend type, and by team, reviewed on one cadence by one owner. Individual vendor portals will always show you a fragment. Tracking is the discipline of assembling the fragments into a whole before the whole surprises you.