The USD 400 Million AMD Deal That’s Quietly Redefining AI Compute Leadership
Inside the USD400M Crusoe-AMD Deal
What does it mean when a rising AI startup backs a new silicon partner strategically and at scale?
In an industry long dominated by Nvidia, cloud infrastructure startup Crusoe is turning heads with a bold move. On June 12, 2025, co-founder and CEO of Crusoe, Chase Lochmiller, announced a USD400 million investment plan in AMD’s MI355X chips. The deal includes approximately 13,000 units, which will power Crusoe’s next-generation AI data center cluster, scheduled to launch in fall 2025.
The deal marks one of AMD’s largest GPU orders to date, a clear signal of a quiet realignment in AI infrastructure. With AI workloads surging and generative models maturing, AMD’s rise is no longer hypothetical. For Crusoe, the deal isn’t just procurement; it’s strategy.
Who is Crusoe, really?
Founded in 2018, Crusoe Energy began by converting flared natural gas into computing power for crypto mining. Today, it has redefined itself as a vertically integrated AI cloud provider.
In December 2024, Crusoe raised $600 million in Series D funding, backed by major investors including Founders Fund, Fidelity, G2 Venture Partners, and NVIDIA itself, bringing its valuation to $2.8 billion.
Crusoe now builds and operates AI-optimized data centers powered by stranded energy, renewable sources, and custom cooling infrastructure.
“Where startups lack in scale, we can compete by being nimble, fast, and having a high density of engineering talent,” said Chase Lochmiller, CEO, Crusoe.
Crusoe is also working closely with OpenAI, having signed a contract securing USD 11.6 billion in funding in May 2025 to support its next large-scale data center.
Why AMD? Why now?
Isn’t NVIDIA the industry standard for AI chips?
Yes, Nvidia still leads the AI accelerator market with its flagship H100 and B100 GPUs. But Crusoe's switch to AMD is a deliberate move. The MI355X GPUs, part of AMD’s MI300 series, are engineered for inference-heavy workloads and provide high-bandwidth memory and key features for powering LLM (large language model) operations, recommendation engines, and generative AI services.
Meanwhile, AMD’s ROCm software stack continues to mature with time, increasingly attracting developers looking beyond the Nvidia ecosystem.
Crusoe is betting on achieving comparable performance at scale, especially for cloud-native AI workloads.
According to AMD CEO Lisa Su:
“The demand for AI compute is strong… [MI300] is a catalyst for our data center GPU business, which we expect to grow to tens of billions over the next couple of years.”
This $400 million deal lends weight to that vision, positioning AMD as a real, scalable alternative in AI infrastructure.
How big is this infrastructure?
On March 18, 2025, Crusoe announced the start of the next phase of construction on its AMD-powered AI data centers, which will anchor a 200 MW, liquid-cooled data center in Abilene, Texas, which is part of a larger 1.2 GW campus under development. The facility will feature direct-to-chip liquid cooling for higher thermal efficiency, 100% clean energy sourcing, and an advanced modular design to maximize scalability and compute density.
In parallel, on October 15, 2024, Crusoe announced a $3.4 billion joint venture with Blue Owl, a leading alternative asset manager, and Primary Digital Infrastructure, an advisory and investor firm focused on the data center industry, to build and operate modular data centers capable of hosting up to 100,000 GPUs. It’s one of the most ambitious private compute expansions in North America to date.
“Rapidly expanding demand for purpose-built data centers proves that markets know the future will be powered by AI,” said Chase Lochmiller, CEO and co-founder of Crusoe. “We've designed this data center to enable the largest clusters of GPUs in the world that will drive new breakthroughs in AI. An investment of this magnitude from a leading, trusted asset manager like Blue Owl and Primary Digital Infrastructure is a reflection of Crusoe’s proven ability to meet growing demand for AI compute and to power these workloads sustainably.
What does this shift mean?
“Inference surpasses training … The entire AI infra TAM expands… New infrastructure and chip players emerge.”
- Daniel Newman, Futurum Group CEO
This move signals a broader transformation in AI infrastructure.
For AI developers, it opens doors to scalable, cost-effective compute with reduced vendor lock-in.
For the infrastructure market, it marks a decisive push toward decentralized, climate-conscious, startup-led innovation.
And for AMD, it represents a breakthrough, validating its MI-series GPUs not just as affordable alternatives, but as performance-grade hardware ready to compete at hyperscale.
As AI scales beyond R&D and into enterprise and real-world deployments, the demand for hardware diversity, energy efficiency, and tailored data center infrastructure continues to accelerate. Crusoe is rising to meet that demand, and AMD is powering the shift.
Final Thoughts
Crusoe’s $400 million investment in AMD chips isn’t just about hardware: it’s a signal flare for the industry. It shows that the future of AI computing won’t be written by legacy giants alone, but by fast-moving, climate-conscious startups willing to rethink the entire stack, from silicon to sustainability.
In choosing AMD, Crusoe isn’t just picking a chipmaker; it’s endorsing a philosophy: that innovation thrives when infrastructure is open, efficient, and responsibly scaled. And perhaps more importantly, that there’s room for new contenders on both sides of the chip.
As AI workloads stretch further into real-world applications, from enterprise inference to foundational model training, partnerships like this remind us:
Leadership isn’t just about who got there first; it’s about who’s building what’s next.
