AI Is Hitting Operational Limits

Learn more about Datadog

As AI adoption accelerates, operational complexity — not model intelligence — is becoming the primary barrier to reliable AI at scale, according to the State of AI Engineering 2026 from Datadog.

The report highlights a compounding complexity challenge as AI systems scale. Nearly seven in ten companies (69%) now use three or more models alongside increasingly complex agent workflows. Around 5% of AI model requests fail in production, with nearly 60% of those failures caused by capacity limits — leading to slowdowns, errors, and broken experiences in AI-powered applications.

Additional key findings:

  • Multi-model is now the norm: OpenAI remains the most widely used provider at 63% share, alongside rising adoption of Google Gemini and Anthropic Claude which grew by 20 and 23 percentage points, respectively.
  • Agent framework adoption doubled year-over-year, accelerating development but also introducing more moving parts into production systems.
  • The amount of data sent to AI models per request is also rising: the average number of tokens more than doubled for median use teams (50th percentile of usage volume) and quadrupled for heavy users (90th percentile).

"AI is starting to look a lot like the early days of cloud," said Yanbing Li, Chief Product Officer at Datadog. "The cloud made systems programmable but much more complex to manage. AI is now doing the same thing to the application layer. The companies that win won't just build better models — they'll build operational control around them. In this new era, AI observability becomes as essential as cloud observability was a decade ago."

Speed Requires Control

Competitive pressure is accelerating AI deployment across startups and large enterprises alike. But as systems scale, speed without control creates risk. Failures are increasingly driven by system design, including fragmented workflows, excessive retries, and inefficient routing.

"Innovation alone isn't enough," added Li. "To scale AI with confidence, organizations need real-time visibility across the entire stack — from GPU utilization to model behavior to agent workflows. Visibility and operational control are what allow teams to move fast without sacrificing reliability or governance. At scale, how you operate AI may matter more than the models you choose."

Methodology: Datadog analyzed anonymized usage data from thousands of customers using LLMs in production environments, with global coverage across industries and geographies.

Hot Topics

The Latest

The Corporate Network Is Fading - Here's What Replaces It

For decades, enterprise networks were designed around a simple assumption: work happened inside the office. Applications lived in centralized data centers, employees connected through internal infrastructure, and security focused on protecting the perimeter that surrounded everything ... But the way organizations operate today bears little resemblance to that environment. Cloud platforms host critical applications, employees connect from homes and airports as often as they do from offices, and partners collaborate through shared systems that exist far beyond corporate walls. In short, the corporate network no longer resembles the environment it was designed to protect ...

IT Leaders Must Establish a Network Data Architecture Practice

As an analyst who researches how IT organizations design, build, and operate their networks, I find that network data is a constant source of pain. Network teams struggle with data quality, fragmentation, authority, access, and trust. And these issues undermine everything they try to do. Here are the numbers: Only 45% of network teams are completely confident in the accuracy of their network source of truth, which documents the intent of their network ...

Data Center Outages Are Decreasing but More Costly

The 2026 Global Data Center Survey from Uptime Institute reveals an industry navigating workforce constraints, escalating outage expenses, even as rising costs remain the top concern for management teams ...

Observability Has a Physical Layer Now: What AI Infrastructure Teaches IT Operations

The next observability gap may not be in the code. It may be under the rack. That sounds strange until you think about how AI incidents actually feel in the middle of an investigation ... The application dashboard may be accurate. It may also be stopping at the wrong boundary. AI systems depend on software, but they also depend on a dense physical stack: racks, power paths, thermal margin, maintenance activity and, in many environments, liquid cooling. Those physical dependencies can change slowly before they look like a software incident ...

TLS Certificate Expiration Is Becoming an Observability Problem

Certificate expiration is the rare outage you can see coming. Every TLS certificate carries the date it stops working, so the moment it will begin breaking connections is knowable in advance. That's what makes an expired certificate such a frustrating way to lose a service. What's changing now is how often that date comes around ...

Enterprise Cloud Observability: Choosing the Right Starting Point

Enterprises operate different combinations of workloads across cloud, hybrid and multicloud environments. For business-critical workloads, teams need to consider monitoring and observability early so they can detect health issues, investigate failures, and understand operational impact. Organizations place workloads on cloud platforms based on a combination of technical requirements, economics, existing dependencies, organizational standards, and business priorities. Their monitoring priorities therefore depend on what they operate and where those systems run. Those priorities will not look the same for every organization ...

AI Is Making Incident Management More Complex - Bad Data Makes It Harder

Top-performing businesses prioritize data-driven decision making, enabling leaders to move from intuition and gut feel towards evidence-based judgment. But that judgment is only sound when the data underpinning decisions is accurate. With incident management, data accuracy is particularly important. Long-term revenue, customer trust, and operational stability depend on high-quality data that enables teams to quickly identify and address the root cause of major incidents. Against this backdrop, governance becomes a critical endeavor to ensure the right data drives the right action ...

MEAN TIME TO INSIGHT Podcast - Episode 26: Network Compliance

In MEAN TIME TO INSIGHT Episode 26, Shamus McGillicuddy, VP of Research, Network Infrastructure and Operations, at EMA discusses network compliance ... 

Why VM and Container Monitoring Is Not Enough for Cloud-Hosted Autonomous Agents

Most production autonomous agents do not run in a vacuum. They run inside cloud infrastructure: virtual machines, containers, pods, managed clusters or private servers. That is where most operations teams start monitoring. Is the VM alive? Is the container running? Did the pod restart? Is memory stable? Is CPU too high? Did the health check pass? Those signals are useful. They tell you whether the shell around the agent is alive. They do not tell you whether the agent inside is actually operational ...

Beyond Monitoring: Why IT Operations Must Evolve into Decision Operations

Enterprise IT environments have never been more observable ... Yet many organizations still grapple with outages, lengthy incident resolution cycles, and increasing complexity. Most teams do not suffer from a shortage of data. They struggle to determine what deserves attention and what action to take next ... Enterprise IT operations must move beyond monitoring and visibility. The next stage of maturity is decision operations, an approach that helps teams make faster, better-informed decisions ...

AI Is Hitting Operational Limits

Learn more about Datadog

As AI adoption accelerates, operational complexity — not model intelligence — is becoming the primary barrier to reliable AI at scale, according to the State of AI Engineering 2026 from Datadog.

The report highlights a compounding complexity challenge as AI systems scale. Nearly seven in ten companies (69%) now use three or more models alongside increasingly complex agent workflows. Around 5% of AI model requests fail in production, with nearly 60% of those failures caused by capacity limits — leading to slowdowns, errors, and broken experiences in AI-powered applications.

Additional key findings:

  • Multi-model is now the norm: OpenAI remains the most widely used provider at 63% share, alongside rising adoption of Google Gemini and Anthropic Claude which grew by 20 and 23 percentage points, respectively.
  • Agent framework adoption doubled year-over-year, accelerating development but also introducing more moving parts into production systems.
  • The amount of data sent to AI models per request is also rising: the average number of tokens more than doubled for median use teams (50th percentile of usage volume) and quadrupled for heavy users (90th percentile).

"AI is starting to look a lot like the early days of cloud," said Yanbing Li, Chief Product Officer at Datadog. "The cloud made systems programmable but much more complex to manage. AI is now doing the same thing to the application layer. The companies that win won't just build better models — they'll build operational control around them. In this new era, AI observability becomes as essential as cloud observability was a decade ago."

Speed Requires Control

Competitive pressure is accelerating AI deployment across startups and large enterprises alike. But as systems scale, speed without control creates risk. Failures are increasingly driven by system design, including fragmented workflows, excessive retries, and inefficient routing.

"Innovation alone isn't enough," added Li. "To scale AI with confidence, organizations need real-time visibility across the entire stack — from GPU utilization to model behavior to agent workflows. Visibility and operational control are what allow teams to move fast without sacrificing reliability or governance. At scale, how you operate AI may matter more than the models you choose."

Methodology: Datadog analyzed anonymized usage data from thousands of customers using LLMs in production environments, with global coverage across industries and geographies.

Hot Topics

The Latest

The Corporate Network Is Fading - Here's What Replaces It

For decades, enterprise networks were designed around a simple assumption: work happened inside the office. Applications lived in centralized data centers, employees connected through internal infrastructure, and security focused on protecting the perimeter that surrounded everything ... But the way organizations operate today bears little resemblance to that environment. Cloud platforms host critical applications, employees connect from homes and airports as often as they do from offices, and partners collaborate through shared systems that exist far beyond corporate walls. In short, the corporate network no longer resembles the environment it was designed to protect ...

IT Leaders Must Establish a Network Data Architecture Practice

As an analyst who researches how IT organizations design, build, and operate their networks, I find that network data is a constant source of pain. Network teams struggle with data quality, fragmentation, authority, access, and trust. And these issues undermine everything they try to do. Here are the numbers: Only 45% of network teams are completely confident in the accuracy of their network source of truth, which documents the intent of their network ...

Data Center Outages Are Decreasing but More Costly

The 2026 Global Data Center Survey from Uptime Institute reveals an industry navigating workforce constraints, escalating outage expenses, even as rising costs remain the top concern for management teams ...

Observability Has a Physical Layer Now: What AI Infrastructure Teaches IT Operations

The next observability gap may not be in the code. It may be under the rack. That sounds strange until you think about how AI incidents actually feel in the middle of an investigation ... The application dashboard may be accurate. It may also be stopping at the wrong boundary. AI systems depend on software, but they also depend on a dense physical stack: racks, power paths, thermal margin, maintenance activity and, in many environments, liquid cooling. Those physical dependencies can change slowly before they look like a software incident ...

TLS Certificate Expiration Is Becoming an Observability Problem

Certificate expiration is the rare outage you can see coming. Every TLS certificate carries the date it stops working, so the moment it will begin breaking connections is knowable in advance. That's what makes an expired certificate such a frustrating way to lose a service. What's changing now is how often that date comes around ...

Enterprise Cloud Observability: Choosing the Right Starting Point

Enterprises operate different combinations of workloads across cloud, hybrid and multicloud environments. For business-critical workloads, teams need to consider monitoring and observability early so they can detect health issues, investigate failures, and understand operational impact. Organizations place workloads on cloud platforms based on a combination of technical requirements, economics, existing dependencies, organizational standards, and business priorities. Their monitoring priorities therefore depend on what they operate and where those systems run. Those priorities will not look the same for every organization ...

AI Is Making Incident Management More Complex - Bad Data Makes It Harder

Top-performing businesses prioritize data-driven decision making, enabling leaders to move from intuition and gut feel towards evidence-based judgment. But that judgment is only sound when the data underpinning decisions is accurate. With incident management, data accuracy is particularly important. Long-term revenue, customer trust, and operational stability depend on high-quality data that enables teams to quickly identify and address the root cause of major incidents. Against this backdrop, governance becomes a critical endeavor to ensure the right data drives the right action ...

MEAN TIME TO INSIGHT Podcast - Episode 26: Network Compliance

In MEAN TIME TO INSIGHT Episode 26, Shamus McGillicuddy, VP of Research, Network Infrastructure and Operations, at EMA discusses network compliance ... 

Why VM and Container Monitoring Is Not Enough for Cloud-Hosted Autonomous Agents

Most production autonomous agents do not run in a vacuum. They run inside cloud infrastructure: virtual machines, containers, pods, managed clusters or private servers. That is where most operations teams start monitoring. Is the VM alive? Is the container running? Did the pod restart? Is memory stable? Is CPU too high? Did the health check pass? Those signals are useful. They tell you whether the shell around the agent is alive. They do not tell you whether the agent inside is actually operational ...

Beyond Monitoring: Why IT Operations Must Evolve into Decision Operations

Enterprise IT environments have never been more observable ... Yet many organizations still grapple with outages, lengthy incident resolution cycles, and increasing complexity. Most teams do not suffer from a shortage of data. They struggle to determine what deserves attention and what action to take next ... Enterprise IT operations must move beyond monitoring and visibility. The next stage of maturity is decision operations, an approach that helps teams make faster, better-informed decisions ...