From Gaming Rigs to GPU Infrastructure: How Christian Ondaatje is Democratizing AI Compute with Aranya

The GPU Origin Story

There’s something special about founders who’ve been obsessed with a single piece of technology for years before starting a company. Christian Ondaatje is one of those people.

His love affair with GPUs started early—like most good tech stories, through gaming and VR in college. But unlike most people who play games and move on, Christian decided to build an eGPU company while still in his CS program at Harvard. It didn’t work out (thanks, cease and desist from Apple), but it taught him something invaluable: the startup game is messy, unpredictable, and somehow only made him more obsessed with GPUs.

Fast forward to today, and Christian is the co-founder and CEO of Aranya—a company solving one of the most pressing infrastructure problems in AI: how to get from bare metal servers to production-ready GPU clusters without losing weeks to setup and configuration.

Outside of work, he’s channeling that same technical obsession into building a custom Linus VR racing simulator rig. (This is the kind of founder detail that tells you everything you need to know about someone’s approach to problem-solving: they don’t just think about problems, they build solutions in their spare time.)

The Problem: GPU Infrastructure is Still Broken

Here’s the thing about the AI boom: demand for compute is at an all-time high. Companies are desperate for GPU capacity. But there’s a massive gap between what they need and what’s actually available.

Most AI companies started their journey in the hyperscalers—AWS, Azure, Google Cloud. They built their infrastructure, their workflows, their entire mental models around managed services. But as they scale, costs become prohibitive. The economics push them toward private bare metal data centers, which offer significantly cheaper compute.

The problem? Those data centers don’t come with the orchestration layer that modern AI companies have come to expect. There’s no managed Kubernetes. No pre-built automation. No way to seamlessly span massive amounts of compute across multiple clusters.

So what happens? Companies spend weeks—sometimes months—trying to bridge that gap themselves. They’re hiring infrastructure engineers just to set up the plumbing. They’re writing custom scripts. They’re managing technical debt that could’ve been avoided.

This is the exact problem Aranya solves.

The Solution: 48 Hours to Production

Christian’s claim to fame is deceptively simple: Aranya can take any bare metal infrastructure and turn it into a production-ready inference cluster in 48 hours.

Think about that for a second. 48 hours. Not weeks. Not months. Two days.

“We bridge the gap from the reality of what modern private data centers offer, which is bare metal,” Christian explains. “And then orchestrate that and build custom architectures on the fly so that it matches very cleanly with what AI companies need.”

The magic is in the orchestration layer. Aranya drops in, handles the complexity, and suddenly you have a managed, scalable, production-ready system. The result? Customers can totally outscale their competition. And according to Christian, they’ve already seen this play out multiple times this year.

The Technical Foundation: A Decade of Orchestration

Here’s where the origin story gets interesting. The core technology powering Aranya didn’t start with Aranya.

Ten years ago, Christian was designing systems for orchestrating heterogeneous compute into cohesive wholes. This evolved into an open-source technology called ClusterDost. The crucible? Running a previous company (Tokaweer) as a solo founder while managing an entire data center’s worth of compute.

Necessity breeds innovation. ClusterDost emerged from that pressure—a tool that allowed Christian to coherently manage massive amounts of compute alone. Eventually, that IP was transferred to Aranya to form the core of their single cluster engine.

“Cluster analysis basically the tool that allows us to go multi-cluster super seamlessly,” Christian notes. “Because each individual cluster becomes very clean to orchestrate.”

This is classic founder wisdom: the best solutions come from having actually lived the problem.

The Open Source Decision

One of the most interesting choices Christian made was keeping ClusterDost open source—and doing the same with Aranya’s foundational technology.

This wasn’t an obvious call. Many founders wrestle with open source strategy. But for Christian, it’s non-negotiable.

“Open source is very near and dear to me,” he says. “It’s something that I owe much of my capabilities and engineering to the various open source communities that I’ve been a part of.”

More importantly, he understands something fundamental: the only way to create a category-defining technology that lasts multiple generations is through open source. Look at Linux. Look at Kubernetes. These didn’t dominate because they were proprietary—they dominated because they were genuinely useful to the community.

By keeping their technology open, Aranya creates the conditions for adoption at scale. They can add features that benefit the ecosystem (like self-hosted Postgres or Minecraft servers on ClusterDost) without worrying about immediate commercial payoff. It’s a bet on the long game.

Product Roadmap Driven by Users

So how do you decide what to build next? Christian’s philosophy is refreshingly simple: “The customer is the PM. The user is the PM.”

This isn’t theoretical. It directly shaped how Aranya built their primary interface.

Originally, they planned to launch a comprehensive web UI. Seemed logical, right? But then they came to terms with reality: 99% of compute operations happen over text channels. Slack. Telegram. WhatsApp. iMessage. These are the channels where infrastructure actually gets managed at the human level.

So Aranya made the call to embrace that chaos instead of fighting it. They built their primary interface for the multi-cluster operating system to live in Slack and email.

“I think a lot of tech companies can be divided into two categories,” Christian says. “You have companies that see chaos and break out an iron and board. Then you have companies that see chaos and break out a surf board. We try to see the chaos and just go with it.”

This is the kind of product thinking that creates delightful user experiences. You’re not forcing behavior change. You’re meeting people where they already are.

Building the Winning Team

Behind every great technical founder is a team of people who share that obsession and curiosity.

Christian has two co-founders. Sasi (CPO) grew a department at Jaguar Land Rover from 20 to 300 people with a $12M R&D budget. Arya (the other co-founder and Christian’s spouse) has operated at the highest levels of venture capital and decentralized compute.

They met while Arya was getting her MBA at MIT, and they’ve been building together ever since.

For the broader team, Christian was intentional. They didn’t make their first hire for four to five months. Why? Because they wanted to plant the seeds of an amazing team full of people with curiosity and the ability to learn on the fly.

This matters because the job is genuinely hard. You’re managing tens of small data centers. Things break constantly. A switch port goes down. A tornado takes out a backbone router. You need brilliant, creative people who can debug complex systems under pressure.

The Bottom Line

What Christian and Aranya are building is infrastructure for the AI era. They’re solving the gap between what companies need and what’s actually available. They’re doing it with a team that’s obsessed with the problem. And they’re doing it in a way that embraces the real world instead of fighting it.

The result? Companies can go from bare metal to production in 48 hours. That’s not just a feature—that’s a competitive advantage.

Want to hear the full story? Listen to the complete episode with Christian Ondaatje on Code Story to dive deeper into how Aranya is reshaping GPU infrastructure, the hard lessons from his Apple cease-and-desist, and why embracing chaos might be the best product decision you can make. Head over to Code Story now.