Infrastructure, unfiltered
Essays on infrastructure, AI, and developer experience from the team building ContextOS.
Latest
- Perspective
Different Logo, Same Cage
Everyone is fleeing VMware, and most teams are landing in the next walled garden. A forced migration is the only moment anyone re-architects — and the industry is about to spend the biggest one in a decade on changing vendors instead of changing the layer.
- Perspective
Pilot Purgatory Is an Infrastructure Problem
95% of enterprise GenAI pilots show zero measurable return. The industry blames data readiness, change management, model choice. The cause nobody dwells on: the path from a working demo to a secure, scaled, cost-governed production system is still hand-built — and most organizations can't build it.
- Perspective
We Built the Automation. We Never Built the Memory.
Enterprise AI projects keep failing, and the blame keeps landing on the model. But most organizations can't tell you what's running in production, let alone why — and no amount of prompting fixes an environment that has lost its own reasoning. The gap between prototype and production was always an infrastructure problem.
- Thomas Hatch
Different Logo, Same Cage
Everyone is fleeing VMware, and most teams are landing in the next walled garden. A forced migration is the only moment anyone re-architects — and the industry is about to spend the biggest one in a decade on changing vendors instead of changing the layer.
- Alex Peay
Pilot Purgatory Is an Infrastructure Problem
95% of enterprise GenAI pilots show zero measurable return. The industry blames data readiness, change management, model choice. The cause nobody dwells on: the path from a working demo to a secure, scaled, cost-governed production system is still hand-built — and most organizations can't build it.
- Thomas Hatch
We Built the Automation. We Never Built the Memory.
Enterprise AI projects keep failing, and the blame keeps landing on the model. But most organizations can't tell you what's running in production, let alone why — and no amount of prompting fixes an environment that has lost its own reasoning. The gap between prototype and production was always an infrastructure problem.
- Alex Peay
Cloud Waste Just Reversed a Five-Year Trend. That Was Inevitable.
After five straight years of decline, wasted cloud spend climbed back to 29% in 2026 — in the same year FinOps teams and Cloud Centers of Excellence reached record adoption. The industry's answer is more FinOps. You can't tool your way out of a business model.
- Thomas Hatch
Containment Was Never a Prompt
A gym-booking assistant and a frontier lab's own sealed security tests both broke out of their boundaries this summer. Neither needed a sophisticated attack — just a boundary that existed as an instruction instead of as a property of the network. Twenty years of bolting security onto infrastructure is what made that possible.
- Alex Peay
95% Idle: The Most Expensive Waste in Enterprise IT
Enterprises spent two years solving GPU scarcity and created a GPU utilization crisis. Average utilization across tens of thousands of production clusters is 5% — while the price of the hardware went up. This isn't a silicon problem. It's an orchestration problem.
- Thomas Hatch
Intelligence Got Cheap. Infrastructure Didn't.
Open-weight models have closed the frontier gap to a matter of months, and a 27-billion-parameter model now runs on a single GPU. That doesn't mean AI is slowing down — it means the value is moving to whoever can actually deploy and run it. The bottleneck just went back to where it has been sitting since 1969.
- Alex Peay
The Bottleneck Moved. Nobody Told Infrastructure.
AI made writing software fast. It didn't make shipping it fast. Two six-hour outages at the world's biggest storefront, a 4.6x review backlog, and 43% of AI code failing after it passes staging all point at the same layer: infrastructure was built for human-speed change and never resized. Adding reviewers won't fix it. Removing what they're reviewing will.
- Thomas Hatch
AI Won't Fix Your Infrastructure (It'll Just Break It Faster)
When AI compresses the cost of creation to near zero, it doesn't just change how fast things get built — it changes what gets built, how much, and whether any of it is worth anything.
- Alex Peay
Your AI Stack Has No Security Boundary (And That Should Terrify You)
Why ContextOS is building a unified platform instead of chasing the latest model serving framework — and why that answer should concern everyone building AI applications today.
- Alex Peay
Three Strategic Missteps That Cost Infrastructure Companies YEARS
Three early-stage decisions infrastructure startups make that look smart in the moment but compound into years of lost opportunity by year three.
- Alex Peay
Stop Building Spikes: Why the Sharpest Product Isn't Always the Strongest Company
Too many founders hear the advice to build a wedge but end up building a spike instead — a sharp product with no platform behind it to drive expansion.
- Alex Peay
The Developer Productivity Crisis: Why Your Engineers Are Drowning in Infrastructure
Developers spend only 16% of their working time actually writing code, with the remaining 84% consumed by operational tasks like CI/CD pipelines, infrastructure management, and security compliance.
Under the Hood: How ContextOS Automates Database Provisioning
Spinning up a database seems straightforward, but provisioning one for production typically demands far more expertise and time than engineers expect.
- Thomas Hatch
Developers should be Drivers, Not Mechanics: Infrastructure That Finally Gets Out of Your Way
Infrastructure should work like driving a great car: developers should be able to just ship code while the complexity stays hidden in the platform.
- Thomas Hatch
The DSM and Software Drivers: Deployment Done Right
Current infrastructure deployment approaches suffer from significant complexity, and the ContextOS Distributed State Machine and Software Drivers fix that with consistent, idempotent, autohealing deployment.
- Thomas Hatch
The VFS: Where Distributed State Actually Works
The ContextOS Virtual File System is an active source and declaration of truth that changes how distributed systems manage state by splitting active and desired state and embracing eventual consistency.
The CTX/ICC Construct: Making Distributed Computing Actually Work
Every developer understands how to use multiple CPU cores on a single computer through threads and processes, and ContextOS extends that same model across multiple physical machines.
- Thomas Hatch
We've Been Solving the Wrong Problem: Why ContextOS Exists
After 20 years building datacenter infrastructures for everyone from startups to the Fortune 500, the realization is that 98% of infrastructures have nearly identical goals.
- Alex Peay
Built for Developers, Ready for Enterprises
At SaltStack we built one of the most widely-used open source DevOps tools in the world, but struggled to help customers understand how easy it was to get started.
- Alex Peay
Cloud Pricing: How your vendor is weaponizing pricing
Major cloud vendors like AWS, Azure, and Broadcom have turned pricing complexity into a strategic weapon that drains organizations through confusion, lock-in, and opacity.
The Infrastructure Industry's Dirty Secret
In the enterprise software world, complexity isn't a bug — it's the business model, because vendors make more money when infrastructure is complicated.
The DevOps Paradox: How Automation Created More Work
The promise was that automation would set us free, yet infrastructure teams today have more tools and more automation than ever and are somehow working harder than they ever did.
The future of Infrastructure
Infrastructure solutions must address the distinct needs of three overlooked personas — executives, IT administrators, and developers — to deliver a system that is automated, easy, secure, and cheaper.
- ContextOS Team
Between a Rock and Hard Place
Companies moving workloads back on-premise are caught between a cloud-created IT skill gap and antiquated, lock-in-focused infrastructure tools, creating an opening for disruption.
Are you trapped in a storm?
Despite two decades of cloud hype, not every workload belongs in the cloud, and many companies are now moving workloads back on-premise to escape rising costs and vendor lock-in.