Datadog (NASDAQ:DDOG) just made a notable AI move. On June 30, 2026, the company announced a deal to acquire Adaptive ML, a frontier AI startup focused on Reinforcement Learning Operations, or RLOps. Terms were not disclosed.
Adaptive ML is expected to join Datadog AI Research, where the focus will be world models and agentic LLM post-training for observability and security.
This comes right after a strong Q1 2026 for Datadog. Revenue crossed $1 billion for the first time, rising 32% year over year. ARR moved above $4 billion, while free cash flow reached $289 million.
Datadog also said over 6,500 customers now send data for one or more AI integrations. That group is only 20% of customers, but it represents about 80% of ARR.
So, the timing matters. Datadog is not buying AI buzz. It is buying deeper AI infrastructure capability.
Turning Telemetry Into First-Party Intelligence
The biggest synergy sits inside Datadog’s data advantage. Datadog already sees large volumes of real-world signals across infrastructure, applications, logs, security, user experience, and cloud environments. Adaptive ML could help convert that raw telemetry into more useful AI behavior.
That is especially relevant for agents that need feedback loops, not just static training data. Adaptive ML’s RLOps platform was built around improving specialized models and agents over time. Datadog’s platform gives it a large operating environment where those ideas can be tested.
This matters because observability is moving from dashboards toward action. Customers do not only want alerts. They want systems that explain issues, triage incidents, and suggest fixes.
Datadog already has Bits AI SRE Agent, Bits AI Security Agent, Bits Assistant, and MCP server. Adaptive ML could sharpen these products by improving post-training, reinforcement learning, and model evaluation.
In plain English, Datadog may gain better ways to make its agents learn from production reality.
Strengthening AI Training & GPU Observability
Datadog’s Q1 call made one thing clear: AI training is becoming a real market for the company. Management said training workloads used to be narrow and homegrown. Now they are becoming production workloads.
That creates demand for reliability, efficiency, and deeper monitoring. Datadog also landed large deals with AI research divisions at two major technology companies. These teams are using Datadog to monitor hyperscale AI training workflows and optimize GPU usage.
Adaptive ML could fit neatly into that shift. Its work around RLOps and specialized agents gives Datadog more depth in the AI development lifecycle. This could support GPU monitoring, LLM observability, and future tools aimed at training teams.
The synergy is not just about watching GPUs run hot. It is about linking model behavior, infrastructure performance, data pipelines, and engineering workflows. That could make Datadog more relevant to AI labs, hyperscalers, and enterprises training their own models.
Deepening Agentic Workflows Across Observability & Security
Datadog is already seeing strong usage growth across its AI tools. Bits AI SRE investigations more than doubled from December to March. LLM observability spans nearly tripled quarter over quarter. MCP server tool calls quadrupled. Bits Assistant messages rose by a factor of 12.
These are early signs that both humans and agents are using the platform more often. Adaptive ML could help Datadog make those agentic workflows more capable and more reliable.
The security angle also matters. Datadog’s Bits AI Security Agent can triage Cloud SIEM signals and investigate threats. Management said some investigations that once took hours can now take as little as 30 seconds.
Adaptive ML may help improve this kind of agent behavior through better reinforcement learning and domain-specific post-training. The result could be stronger automation across incident response, threat analysis, and root-cause workflows.
Still, agentic systems need guardrails, permissions, and auditability. That work remains hard.
Expanding Platform Stickiness Through RLOps
Datadog’s platform model is already showing strong expansion. In Q1, 56% of customers used four or more products. About 35% used six or more. Another 20% used eight or more.
That matters because Adaptive ML could add more reasons for customers to stay inside Datadog’s ecosystem. If AI teams can monitor models, improve agents, evaluate behavior, and connect that data to production systems, Datadog becomes more than an observability vendor.
There is also a go-to-market benefit. Datadog is selling into AI-native companies, banks, insurers, travel groups, fintechs, and public sector customers. It also received FedRAMP High certification and plans a new U.K. data center.
Adaptive ML’s technology could help Datadog serve regulated enterprises that want specialized agents without losing control. The phrase “build, own, and deploy” matters here. Enterprises may want AI systems tuned to their own workflows, not generic assistants bolted onto legacy tools.
Final Thoughts
Datadog’s Adaptive ML deal could add useful AI depth at the right time. The company already has scale, customer breadth, and a growing AI product stack. Adaptive ML could bring stronger post-training, RLOps, and agent-improvement capabilities.
That may help Datadog build better tools for observability, security, GPU monitoring, and AI training workflows.
Still, this could be a double-edged sword. AI research talent is expensive. Integration can distract teams. Agentic products also need trust, controls, and clear customer value. Datadog is already investing heavily in R&D, so the payoff must show up in product adoption over time.
Valuation leaves little room for casual execution. As of June 30, 2026, Datadog traded at 24.29x LTM EV/revenue, 25.24x LTM price/sales, and 30.40x LTM EV/gross profit. Its LTM P/E was also very high at 682.54x.
Those multiples suggest investors are already pricing in strong growth and AI upside. The acquisition may help support that narrative, but it also raises the bar for execution.
Disclaimer: We do not hold any positions in the above stock(s). Read our full disclaimer here.
