Q&A with Alon Yariv on the Atero acquisition and the future of Crusoe Cloud
Crusoe's acquisition of Atero enhances Crusoe Cloud's performance and efficiency. In a Q&A, Atero's co-founder, Alon Yariv, discusses the technology's benefits for AI workloads and the shared vision for an AI-first future.

In the world of artificial intelligence, a year is an eternity. It’s long enough for a new model to redefine an industry, for a bold idea to become a global platform, and for a small, deeply technical team to build something so powerful it gets noticed.
They moved fast. So fast, in fact, that they’re now a core part of Crusoe.
Our recent acquisition of Atero accelerates the performance and efficiency of Crusoe Cloud and establishes our first office in Tel Aviv. To understand what this means for our customers and the future of AI infrastructure, I sat down with Atero’s co-founder and CEO, Alon Yariv, who now leads our product innovation programs and partnerships with model labs.
In this conversation, we explore the core problems Atero solved, the immediate benefits for Crusoe Cloud users, and the shared vision for building the future of AI, faster.
A Q&A with Alon Yariv
1) From a customer's perspective, what was the core problem you set out to solve with Atero?
The main problem we tackled was the lack of flexibility in GPU clusters. We observed a few critical challenges across the AI landscape: model sizes are exploding, GPUs often sit idle, autoscaling AI workloads is painful, and the context size models can handle has grown by orders of magnitude.
All four of these issues share the same bottleneck: memory.
In AI, assets like model weights and KV caches are gigantic — we’re talking terabytes and gigabytes, not the kilobytes you see in standard user applications. It became clear that traditional methods for moving these assets around were simply not adequate. This is what prompted us to create what we call the unified memory layer, which utilizes the networking assets across an entire GPU cluster to transport memory assets in the fastest way possible.
Essentially, we created a shared virtual memory across the whole cluster, eliminating the need to localize memory on a single machine.
2) For our customers building and deploying AI, what is the core benefit of your technology that they will now experience on Crusoe Cloud?
First and foremost, flexibility. Scaling a model from one to a thousand GPUs — and back down to zero — is now possible in real time, which is very relevant in inference workloads in particular.
Second is significantly better usability and throughput. Now you can share the KV cache across the entire cluster. Typically, a user’s session is tied to a specific, physical GPU, which can lead to inefficient, jagged utilization. With our technology, the cache can be spread across machines, leading to much better GPU utilization, lower latency, and a better cost structure.
Ultimately, it provides simpler infrastructure. For companies deploying models at scale, building an inference stack is extremely difficult. It often requires patching together open-source elements that rarely work together in an optimized way. We provide a much simpler path to deploy models and work with AI, with much better performance from day one.
3) How does the technology and talent you've brought to Crusoe help our customers accelerate their own AI journeys?
In the very near-term, customers with models running on Crusoe Cloud will benefit from our KV cache superiority, which delivers better performance and lower latency.
4) Why was Crusoe the right partner for Atero? What about our vision or vertical integration model made this feel like the right fit?
There were a few reasons. First, the people. We felt a great connection with Chase, Cully, and the rest of the team.
Second, the strategy. When I looked at the AI market, I realized one of the biggest future bottlenecks would be energy. Crusoe comes from energy; it’s clear that the team understands how to break through a wall that other companies are going to hit.
Finally, we wanted a partner that would give us the agency to bring our vision to fruition. We are highly opinionated about building the best stack for inference workloads. At Crusoe, we found a place with the right backbone, the right infrastructure, and the right support, but also enough room for us to build what we envision. It was the perfect combination.
5) You’ve built an incredible team in Tel Aviv. Can you share a bit about what makes that city — and your team — such a great hub for innovation?
We look for people who, if they told us they were founding a company, we would invest in them. Every member of our team has a founder in them. They’re critical thinkers with an extremely high sense of accountability and a hunger for success and knowledge.
Tel Aviv has a high density of this kind of raw talent. It’s a culture of knowledge and excellence. I also feel there’s a certain existential awareness that motivates people to make the most of every day. It creates a community of brilliant people who are driven to solve impossible problems.
6) Looking at the big picture, what are you most excited to tackle now that you are part of the Crusoe team?
Inference. I think inference is going to be the largest opportunity the world has ever seen. When you hear people talk about "AI factories," they're talking about GPUs generating tokens. That’s inference.
Unlike training, which has matured, we are in the very early days of inference. There is so much innovation to come in the software stack, in user session management, and in distributed data centers. For Crusoe, this is the largest opportunity for growth beyond building data centers. It’s a huge opportunity to scale demand for our hardware. That’s where the company is going, and that's the opportunity we are now leading.
Ready to build on a cloud designed for the future of AI?
The integration of Atero’s team and technology is a major step forward in our mission to accelerate the abundance of energy and intelligence. It strengthens our vertically integrated platform and, most importantly, empowers you to build the future, faster.
Explore Crusoe Cloud or contact our sales team to learn how our AI-first infrastructure can power your next breakthrough. And if this post sparked your professional interest, we are hiring!