Harness Is Building the Enterprise AI Stack

The company that started as a CI/CD platform just raised $240M, hit a $5.5B valuation, and quietly became the infrastructure layer every enterprise engineer needs.

AI coding tools have transformed software development into a real-time creative process. With tools like Cursor, GitHub Copilot, & Anthropic Claude Code generating production-ready code in seconds, the bottleneck is no longer writing software.

The real challenge begins after the code is generated.

Testing, security, deployment, compliance, observability, cloud cost control, and production reliability still rely on fragmented workflows stitched across dozens of enterprise tools. While AI has accelerated the developer inner loop, the operational layer surrounding software delivery remains painfully manual.

This is the gap Harness is positioning itself to solve: building an AI-native enterprise software delivery stack designed not just to generate code, but to run, secure, optimize, and govern it at scale.

The AI Velocity Paradox

AI has dramatically sped up code generation. But everything that happens after the code is written, like testing, security checks, deployments, compliance approvals, and cloud cost management, still depends heavily on manual workflows and disconnected tools.

So while developers can now ship code faster than ever, the rest of the software pipeline struggles to keep up.

And when only one part of the system gets faster, you don’t get faster software delivery. You just create a larger backlog waiting to be reviewed, secured, deployed, and managed.

The Problem: Inner loop (writing code) is getting 10x faster. The outer loop (everything after code) is stuck where it was in 2020.

The Solution: Whoever automates the outer loop owns the enterprise AI budget. That is a much larger category than most people realize.

Jyoti Bansal, Harness io co-founder and CEO, has a name for what every engineering leader is quietly dealing with right now. He calls it the AI Velocity Paradox. He said this

Source PR Newswire

Source PR Newswire

This isn’t a small or theoretical problem. Most engineering teams already spend more time on everything around the code than on writing the code itself, deployments, security checks, fixing incidents, managing infrastructure, and controlling cloud costs.

And AI coding tools won’t reduce that workload.

In fact, as tools like Cursor and GitHub Copilot help developers generate code faster, the pressure on the rest of the software pipeline only increases. More code means more systems to test, secure, deploy, monitor, and optimize.

That’s the problem. Harness is trying to solve.

Not by building another AI chatbot for developers, but by automating the infrastructure and operational layer that software teams depend on every day.

What $240M Buys You in Enterprise AI

In December 2025, Goldman Sachs led a $240 million Series E into Harness, valuing the company at $5.5 billion. That is a 49 percent jump from the $3.7 billion valuation in their 2022 Series D. The round included IVP, Menlo Ventures, and Unusual Ventures, plus a $40 million tender offer designed to give liquidity to long-term employees.

The numbers behind the raise tell the story clearly. On track to exceed $250 million in ARR in 2025. Fifty-plus percent year-over-year growth. 128 million deployments powered. 81 million builds. 1.2 trillion API calls protected. $1.9 billion in cloud spend optimized. Over 1,000 enterprise engineering teams across North America, EMEA, and APAC.

These are not startup metrics. These are infrastructure company metrics.

The Platform Nobody Is Talking About

Most conversations around enterprise AI focus on the model itself. Is the company using OpenAI or Anthropic? GPT-4, Gemini, or Opus?

But that is slowly becoming the less important question.

The bigger question is what sits around the model. What systems, data, permissions, and workflows give the AI real context to operate inside an enterprise environment?

Harness believes the answer is its Software Delivery Knowledge Graph.

At a basic level, modern software delivery is deeply connected. Deployment pipelines, cloud infrastructure, services, dependencies, security policies, and build systems all affect each other. But that information usually lives across dozens of separate tools, making it difficult for AI agents to fully understand what is happening inside an engineering environment.

generated by @techwith_ram

generated by @techwith_ram

Harness is trying to solve that by creating a centralized knowledge layer that connects all of this data together. The goal is to give AI systems a much more complete understanding of how software is built, deployed, secured, and managed.

The best example of this approach is the Cursor integration launched in April 2026.

Developers can stay inside Cursor and use natural language to trigger deployments, run pipelines, or check governance rules directly from chat. Behind the scenes, the system connects to a Harness-hosted MCP server powered by the Knowledge Graph.

And because it uses existing RBAC permissions, companies do not need to rebuild access controls or create a separate security layer for AI workflows.

Designed by @techwith_ram

Designed by @techwith_ram

Cursor understands the inner loop, what the code is doing. Harness understands the outer loop, how that code gets tested, secured, deployed, and managed. The plugin connects the two. That is context engineering applied at enterprise scale.

The Enterprise Validation Is Arriving

One of the biggest signs that a software platform is becoming core infrastructure is when large enterprises start depending on it for real production workloads.

That is exactly what is happening with Harness.

In March 2026, Workday selected Harness to support agentic AI software delivery at enterprise scale. A month later, Infosys announced a strategic partnership with Harness to automate the entire path from code to production across hybrid and multi-cloud environments.

Around the same time, Harness also expanded its integration with Google Cloud Developer Connect and received the 2026 Google Cloud Technology Partner of the Year Award for Application Development and DevOps.

And the customer list keeps growing. United Airlines, Morningstar, and Choice Hotels are not just testing the platform. They are using Harness to run real software delivery operations in production.

The Insight Most Teams Are Missing

There is a pattern that shows up in every major AI wave.

At first, everyone focuses on the most visible part of the technology. Which model is better? Which benchmark scores are higher? Which AI generates better results?

But over time, the focus shifts.

People start realizing that the model alone is not enough. The real value comes from the infrastructure around it. The systems that provide context, security, governance, observability, compliance, and cost control.

That is where enterprise adoption actually happens.

Today, most frontier AI models are already good enough for enterprise use. For many companies, choosing between OpenAI, Anthropic, or Gemini is no longer the biggest challenge.

The bigger challenge is everything around the model.

How do you securely connect AI to internal systems? How do you manage deployments? Monitor infrastructure? Control cloud costs? Enforce permissions and compliance rules? Give AI enough context to operate safely inside production environments?

That is the layer Harness has been building for years.

Long before people started calling it the “outer loop,” Harness was already focused on automating software delivery infrastructure. Since 2017, the company has expanded far beyond CI/CD into security testing, cloud cost optimization, feature flagging, database change management, internal developer portals, and governance tooling.

All of these products connect to the same idea. The fastest engineering teams are not simply the teams using the best AI coding assistant. They are the teams with automated, observable, and well-governed delivery infrastructure behind the scenes.

AI did not change that idea. It made it even more important.

And now, as AI-generated code increases the speed of software development, more enterprise teams are realizing they also need the infrastructure layer that can manage everything that comes after the code is written.

What the Next 12 Months Look Like

The new $240 million funding round is being used to push three major priorities for Harness: improving the platform, expanding globally, and continuing to build its AI software delivery platform.

The company now has more than 1,200 employees across 14 offices worldwide. Recent launches like the MCP integrations, the Cursor plugin, and the expansion of its Knowledge Graph with Google Cloud are all part of the same strategy.

Harness wants to become the infrastructure layer that connects AI coding tools to real production systems.

CEO Jyoti Bansal describes this category as “AI for everything after code.”

And that market is much bigger than traditional DevOps.

As AI coding tools make software generation faster, companies everywhere are running into the same problem: generating code is easy, but securely testing, deploying, governing, and managing that code at scale is still difficult.

That creates a huge opportunity for Harness.

Almost every engineering team is now facing the same challenge: code velocity is increasing, but software delivery is not speeding up at the same pace. The companies that solve that bottleneck will become a critical part of the modern AI software stack.

Harness is betting that infrastructure, not just AI models, will define the next phase of enterprise software development.

Thoughts

The companies winning with enterprise AI in 2026 are not the ones with the best models. They are the ones who automated the 60-70% of engineering work that happens after the code is written. Harness is the infrastructure layer making that possible. The $5.5B valuation reflects the market starting to realize what that means.

Follow @techwith_ram for more such posts.