See “A Primer on Carbon-Negative AI Data Centers” for more on this topic
September 2025 – California & Wisconsin.
Two very different visions for the future of artificial intelligence infrastructure are taking shape in America… both bold, both necessary.
Earlier this month, Microsoft unveiled plans for what it calls “the world’s most powerful AI data center” in Wisconsin, a multi-billion-dollar complex capable of training models that “could map the planet four times over.” Meanwhile, in California, 2CRSi, a French-founded leader in high-performance, energy-efficient computing, announced the first $290 million order under its $610 million U.S. contract. The project, located near Sacramento, California, will power a new generation of AI Factories operated by NewYork GreenCloud (NYGC), combining state-of-the-art GPU clusters with carbon-negative power.
Two Models, One Mission
At first glance, Microsoft’s and 2CRSi’s approaches could not be more different.
- Microsoft’s Wisconsin Gigacenter embodies centralized scale… a hyperscale hub measured in gigawatts, built to anchor the next era of global cloud computing.
- 2CRSi and NYGC’s AI Factory network takes the opposite path… smaller, modular facilities located near renewable energy sources and closer to data demand centers.
But both strategies are advancing the same mission: enabling the explosive growth of AI with the compute, energy and reliability it requires.
In reality, these two models are complementary, not competitive.
The U.S. will need both the vast, cloud-scale backbone of projects like Microsoft’s and the agile, distributed edge that 2CRSi and NYGC are building.
Inside the California AI Factory
The first phase of 2CRSi’s California deployment, powered by biomass-based, carbon-negative electricity, will integrate one of the most advanced hardware lineups ever assembled:
- 2,304 NVIDIA Blackwell Ultra GPUs for high-efficiency AI training and inference.
- NVIDIA Quantum-2 InfiniBand X800 networking, delivering 800 Gb/s per GPU and 6.4 Tb/s per server to eliminate data bottlenecks.
- NVIDIA BlueField-3 DPUs, enabling zero-trust security, composable infrastructure, and elastic GPU scaling.
- WEKA distributed storage for ultra-low-latency access to massive datasets.
- Advanced liquid-cooling systems engineered by 2CRSi to achieve Power Usage Effectiveness (PUE) below 1.1.
As an Elite Partner of NVIDIA, 2CRSi’s collaboration with NYGC ensures that these systems are purpose-built for AI and high-performance computing, designed to deliver the world’s fastest compute with the smallest environmental footprint.
Decentralization: Efficiency Meets Security
Beyond sustainability and performance, the modular nature of the 2CRSi–NYGC model offers another critical advantage: national security resilience.
By distributing smaller, self-contained AI Factories across multiple regions, each with on-site renewable power, energy storage and fiber connectivity, this approach reduces single-point vulnerabilities in the national compute grid.
Decentralized deployment strengthens both data sovereignty and operational continuity, aligning with emerging federal priorities around secure, sovereign AI infrastructure.
Scale and Sustainability Can Co-Exist
While Microsoft pushes the outer limits of scale, 2CRSi and NYGC are redefining what’s possible at the edge… hyper-efficient, rapidly deployable and locally powered AI capacity.
The future of AI infrastructure won’t be a winner-takes-all equation. Instead, it will be a hybrid ecosystem, where hyperscale and distributed systems operate in tandem: one providing global reach, the other agility, security and sustainability.
Both models point toward the same horizon… a smarter, cleaner and more resilient foundation for the AI-driven economy.
Explore Our Investment Strategies
See how we’re aligning capital with today’s most impactful opportunities!