Datadog, Inc. — AI supply-chain exposure

Datadog is an observability company: software that watches other software, tracking latency, cost and errors across the systems that run modern applications. Its move into the AI chain comes through LLM Observability, a product line that monitors how models behave in production, including how fast they respond, how much each call costs, and whether output quality is drifting. It sits in the applications layer as a producer of the evaluation and monitoring tooling that teams wrap around models they did not build themselves.

The structural hook is that every company deploying a model needs to know whether it is working and what it is costing, and few build that plumbing in-house. Datadog already lives inside enterprise stacks, so extending its dashboards to cover models is a natural adjacency rather than a fresh sale, and AI-native customers are a fast-growing slice of its base. The model keeps it toward the edge of the chain: it captures value from AI adoption without owning compute, models or data, acting more as a toll-collector than a load-bearing link.