Datadog's GPU Monitoring Launch: AI Monetization Push Amid Optimization Risks

What happened

Datadog has launched GPU Monitoring, a product designed to help businesses optimize GPU spend and performance as they scale AI projects. This aligns with the company's strategy to monetize AI-driven telemetry growth, which contributed ~7 percentage points to YoY revenue in Q4 2025. However, the DeepValue report warns that AI-native customers can rapidly optimize usage, creating revenue volatility and threatening the dollar-based net retention rate (DBNRR) of about 120%. The launch is a tactical move to capture more value from AI workloads, but it does not address the core risk of customers suppressing ingest costs to reduce bills. Thus, while reinforcing Datadog's AI narrative, this product enhancement leaves the fundamental investment thesis unchanged.

Implication

The GPU Monitoring launch strengthens Datadog's position in AI observability, potentially attracting more AI-native customers and supporting usage growth. However, it also highlights competitive pressures to provide cost-control tools, which could encourage optimization and compress billable usage per workload. Investors must monitor whether this product helps sustain DBNRR near 120%, as any decline could trigger multiple compression given the high valuation (P/E 402x). The launch does not mitigate the risk of telemetry suppression by competitors like Grafana or Dynatrace, which threaten usage-based monetization. Therefore, while positive for growth, it underscores the need for cautious evaluation of expansion metrics before any investment shift.

Thesis delta

The investment thesis remains unchanged: Datadog's success hinges on DBNRR holding near 120% despite optimization risks from AI-native customers. The GPU Monitoring launch is consistent with the company's focus on AI monetization but does not alter the fundamental drivers or risks. Investors should continue to wait for clearer evidence of sustainable expansion, as highlighted in the DeepValue report's WAIT rating.