The migration story is mostly told. The Fortune 500 has moved — or is committed to moving — its workloads off on-premise hardware. The hyperscalers absorbed that shift. They are now running the compute, the storage, the networking. What most of the industry press calls "cloud adoption" is essentially over as a first-order opportunity for investors.
What is not over — what is arguably just beginning — is the second-order build. Everything that sits between the raw infrastructure and the applications that enterprises actually run.
The Gap Nobody Talks About
Here is the thing about running workloads on a major cloud provider: you get phenomenal raw capability and a fairly brutal DIY experience above the API surface. The compute is elastic. The storage is durable. The networking is fast. But then you try to manage security posture across three regions, enforce compliance for HIPAA or SOC 2 across 40 engineering teams, or orchestrate deployments across a hybrid environment — and you discover that the hyperscaler does not solve that for you. Not cleanly. Not at enterprise scale.
That gap is a $200 billion market being built right now. We have been investing in it for six years.
Three Layers That Are Still Incomplete
The way we think about the opportunity breaks into distinct layers, each with a different stage profile and buying motion:
Security and compliance posture management. As enterprises moved to cloud, their attack surface changed shape. Perimeter-based security no longer makes sense when developers are spinning up infrastructure from a laptop. The tooling to manage identity, access, and compliance state across dynamic cloud environments is still being figured out. The incumbents in this space are either too narrowly focused on specific compliance frameworks or too broad to be operationally useful. There is meaningful white space for products that enforce posture at the engineering workflow layer rather than as a compliance audit layer.
Multi-cloud orchestration. Most large enterprises are not running on one cloud provider. They have AWS for certain workloads, another provider for others, and a legacy on-premise data center for systems too risky to migrate. Managing workloads across that landscape — with consistent policies, unified observability, and intelligent cost optimization — requires a layer that does not exist natively anywhere. We have backed one company specifically in this space and are actively evaluating two others.
Developer experience at scale. When an engineering organization grows past 500 people, the productivity problems compound. Onboarding a new engineer takes weeks instead of days. Infrastructure provisioning is inconsistent. Deployment pipelines break in ways that are difficult to trace. The internal developer platform category — tools that abstract the cloud operations layer into a workflow engineers can actually use without deep ops expertise — is growing faster than any comparable infrastructure category we have tracked. Net promoter scores among engineering buyers are the highest we see across all the sub-markets we cover.
Who Is Buying and Why Now
Enterprise IT budgets have shifted. Five years ago, the primary cloud spend was migrating existing systems. That migration budget is now infrastructure maintenance. The new budget line — the one that has grown every quarter since 2022 without meaningful cyclical dip — is optimization and governance. Enterprises that moved fast and spent freely during the migration wave are now dealing with the consequences: ballooning cloud bills, inconsistent security policies, and engineering teams that struggle to move at the pace the business needs.
The buyers we talk to are not looking for another point solution. They have too many of those. They want fewer vendors, deeper integrations, and tools that actually talk to each other. That preference for platforms over point solutions is something we factor into how we evaluate companies at Series A. A narrow wedge is fine — in fact, it is often necessary to win an initial deployment. But the roadmap needs to show a credible path to becoming the layer, not just a feature within someone else's platform.
What Good Looks Like From a Diligence Perspective
We have evaluated close to 200 cloud infrastructure companies over the past four years. The pattern of good early signals is fairly consistent:
- Initial deployment in one engineering team, organic expansion to three or four others within the first year without any sales involvement
- Net revenue retention above 130 percent in the first cohort of enterprise customers
- A founding team where at least one person spent time actually operating infrastructure at scale, not just building tools at a startup
- A clear explanation of why the problem is not solved by the hyperscaler natively — and why it will not be solved by them in the next three years
That last point matters more than founders often realize. We have seen impressive teams building excellent products in spaces where the hyperscaler eventually ships a native solution. It does not always kill the startup — sometimes the native solution validates the category and the startup wins on depth. But you need to have a clear thesis about why you survive that moment.
The next decade of cloud infrastructure will be built by the teams solving the second-order problems. The raw compute is commoditized. The interesting work is everything else.
