Datadog Gives AI Cost Clarity as Anthropic Aims for $10B Scale

Anthropic, the developer of the Claude series of large language models (LLMs), is reportedly in $10 billion funding talks, signaling a significant capital-raising effort as demand for advanced AI models continues to expand. The talks, though still in early stages, reflect the growing interest from investors in the firm's technology and its potential to reshape enterprise applications of AI. The company, which has gained traction with models like Claude Opus and Claude Sonnet, is now aiming to scale its infrastructure and offerings to meet the rising usage and complexity of AI workloads across industries [1].

One key challenge facing companies integrating models like Claude into their systems is managing the associated costs. As teams increasingly rely on AI, they encounter fragmented data, inconsistent usage metrics, and a lack of visibility into their spending. To address this, Datadog recently launched a native integration with Anthropic’s Usage and Cost Admin API, allowing enterprises to monitor, troubleshoot, and secure their LLM applications from within the Datadog Cloud Cost Management (CCM) platform. This integration enables firms to track detailed cost data across models, workspaces, and teams, offering greater accountability and control [1].

The integration normalizes usage data through the FinOps Foundation’s FOCUS format, making it accessible in CCM dashboards, reports, and alerts. This allows organizations to understand the total cost of services—including cloud, SaaS, and now Claude models—by breaking down usage by input and output token counts, service tiers, and cache hit rates. With the data centralized, teams can identify cost drivers and optimize their AI usage in real time. For example, by analyzing token usage over time, organizations can detect peak demand periods and implement strategies such as request queuing to reduce unnecessary costs [1].

Datadog's integration also supports tagging and service ownership mapping, enabling organizations to allocate costs across workspaces, API keys, and teams. This feature helps firms integrate AI spending into their unit economics and ensures that usage is tied to specific services or products. Through the CCM Explorer, teams can view cost summaries by model and service, with changes in cost—both monetary and percentage-based—highlighted for further investigation. This level of visibility fosters accountability across engineering, FinOps, and finance departments [1].

Beyond monitoring, the integration enables proactive alerting on cost anomalies, such as abnormally low cache hit ratios, unexpectedly high token usage, or sudden shifts in service tier usage. Custom monitors can also be created to track usage patterns and detect deviations that may indicate inefficiencies or misconfigurations. This real-time feedback allows firms to take corrective actions before costs spiral out of control. Additionally, by correlating LLM usage with broader infrastructure and application telemetry, organizations gain deeper insights into how AI contributes to overall system performance and expenses [1].

As AI adoption accelerates, the ability to manage and optimize the costs of foundation models becomes increasingly critical. Datadog’s integration with Anthropic’s cost and usage data provides enterprises with the tools they need to align AI spending with business objectives, ensuring efficient and sustainable use of AI resources. With Anthropic in the midst of raising up to $10 billion, the broader ecosystem of partners and platforms like Datadog is likely to play a pivotal role in enabling seamless, cost-effective AI deployment at scale [1].

Source: [1] Datadog - Monitor Claude usage and cost data with Datadog Cloud Cost Management (https://www.datadoghq.com/blog/anthropic-usage-and-costs/)