Crusoe CEO Chase Lochmiller on Scaling AI Infrastructure to Meet Demand
Chase Lochmiller anticipated the AI energy challenge. Now his company is powering some of the world's largest AI workloads.
September 17, 2026
Crusoe CEO Chase Lochmiller started his career in quantitative finance — a far cry from the world of data centers and power grids.
But that turned out to be the exact right background for what he built.
Crusoe is a vertically-integrated AI infrastructure company that designs, builds, and operates data centers and generates or sources the power to run them. Today Crusoe announced the closing of its $3.9B Series F, which values the business at nearly $31 billion dollars. Salesforce Ventures is a proud investor, and we recently had the chance to sit down with Chase to discuss how he thinks about the AI energy challenge and what he’s building to address it.
Here are three takeaways from our conversation.
1. The idea for Crusoe came from an appreciation of computing infrastructure
When Chase was working in quant finance in the mid-2010s, his team shifted its modeling workloads from CPUs to GPUs. That operational move became formative.
“That experience really made me appreciate the importance of the computing infrastructure, the networking stack, the whole storage ecosystem,” Chase told us. Training deep learning models on GPUs in 2015 gave him a ground-level view of what AI at scale actually required, long before that question became a mainstream concern.
His conviction at the time was simple and turned out to be accurate: “There’s nothing that won’t benefit from more AI embedded in that application and technology to drive greater efficiencies.” He saw the application wave coming and bet on the infrastructure layer underneath it after years of watching quant finance teams push networking and compute to its limits.
2. Go where the power is, not where the consumers are
Most data center operators start with demand and then figure out power. Chase and Crusoe inverted that logic from day one.
“Moving data is a lot easier than moving power,” Chase said. “Go to where you can access low-cost, abundant power, and moving the data in and out of there is a much easier problem to solve.”
While energy constraints dominate today’s discourse around AI infrastructure, Chase’s perspective was non-consensus when Crusoe broke ground on their Abilene, Texas site, which was chosen for its land availability and access to renewable resources.
That location + energy approach became Crusoe’s competitive wedge, and it also shapes how the team thinks about scale. Along with large-scale campuses like Stargate, Crusoe offers Crusoe Spark, a line of modular, prefabricated AI data centers designed for edge deployments. Spark units integrate power, cooling, and GPU racks into a single portable unit that can be delivered in as little as three months and run on diverse power sources — including solar and batteries — in an effort to reduce reliance on large data centers.
Long before energy was viewed as a key bottleneck in the AI buildout, Crusoe understood the importance of where you build AI infrastructure and why.
3. We’re still underbuilding AI infrastructure
While some voices in the industry believe the current wave of data center investment is outpacing demand, Chase believes we’re actually underbuilding given the rapid pace of AI advancement.
“There’s been a consistent shortage — no matter how much we build, there’s more demand than what we can support,” Chase said. “Continuously better models drive more usefulness, more usefulness drives more demand, and more demand drives more infrastructure spending.” Chase points to Anthropic’s publicly reported revenue growth — from $9B ARR at the end of 2025 to over $65B today — as evidence that demand is compounding.
To facilitate AI infrastructure growth, Chase anticipates a meaningful shift away from the data center industry’s focus on “five-nines” reliability (e.g., redundant power feeds, backup systems, failover architecture layered on top of failover architecture). The new model, in his view, will trade some of that reliability for significantly lower construction costs, with software absorbing failures that hardware used to prevent. He estimates this could drive roughly a 40% reduction in infrastructure development costs, which would be a major shift in the unit economics of the entire industry.
At the same time, Chase acknowledges the societal impact of the AI buildout, which is why Crusoe prioritizes community accountability in every project. The company covers its own power infrastructure costs rather than passing them to ratepayers, uses closed-loop cooling systems that consume less water annually than 20 average households, and assigns a dedicated community liaison to each project. Its projects have created more than 13,000 jobs to date and delivered over $20 billion in local investment across host communities. For example, in Warrenton, Missouri, a planned $78 million Crusoe data center is projected to generate over $77 million in local tax revenues.
_
We couldn’t be more excited for Chase and the Crusoe team on this incredible milestone, and look forward to our continued partnership.
For more on our investment in Crusoe, read our investment thesis >>>
