Edgevana Locks $700M in Take-or-Pay Deals for NVIDIA Blackwell

Edgevana has signed fully committed, multi-year take-or-pay agreements worth more than $700 million to deploy NVIDIA Blackwell systems in North America and Europe, the company announced on November 9, 2026.

The contracts cover NVIDIA B300 GPUs and GB300 systems. Behind them sits a bigger commercial picture: Edgevana reports $7 billion in contracted customer demand and an active pipeline exceeding $20 billion. Those are separate measures from the signed commitments — one counts demand already under contract, the other covers opportunities still moving through the pipeline.

The headline number is the floor, not the ceiling, of what Edgevana says it can monetize.

How does the deployment model work?

Edgevana's approach rests on pre-energized infrastructure. The company combines pre-energized land, data center facilities and high-speed networks into a single virtual campus, giving customers one control interface across a distributed compute footprint.

Edgevana attributes demand for the platform to power constraints that are delaying traditional data center construction. Its model ties together three elements:

• Long-term energy supply
• GPU allocation
• Committed customer demand secured in advance

The company says this structure can bring tens of megawatts of high-density AI compute online in 90–120 days. That compares with years of delays for greenfield construction, which Edgevana positions as the alternative enterprises currently face.

By locking in long-term contracts ahead of deployment, Edgevana says it reduces financial risk, accelerates access to compute and delivers dedicated AI capacity at a lower cost basis than traditional providers. The model limits speculative capacity builds before demand is contracted, and it folds power availability into the compute offer itself.

Power, in short, is no longer a site-selection problem — it's part of the product.

What is Edgevana's operating footprint?

The company reports 17 operational data center locations, with another 10 in active development. Platforms running on its infrastructure collectively serve more than 10,000 downstream enterprise customers.

That customer count deserves a caveat: it measures the reach of platforms built on Edgevana's infrastructure, not direct Edgevana contracts. It's a gauge of ecosystem scale rather than signed agreements.

Why inference, and why now?

Co-CEO Ryan Fay tied the strategy to what he called AI's pivot toward inference: "Training rewards centralized mega–campuses, but inference is live production traffic."

Fay cited Edgevana's track record delivering ultra-low latency for high-frequency trading and managing edge production for two of the world's largest hyperscalers. He said the company deliberately stayed in stealth through the initial training boom, waiting for the market to shift toward inference workloads.

Co-CEO Subhan Jahromi framed the constraint in physical terms. He identified heavy-duty infrastructure as AI's defining bottleneck, saying Edgevana spent years securing power, land and compute closer to end users. Deploying reliable, large-scale GPU capacity at the edge is, in his words, exceptionally difficult.

What happens next?

The agreements tether the Blackwell rollout to a model built on existing power, distributed facilities and pre-committed customers. The open question is conversion: whether the reported $7 billion in contracted demand and the $20 billion-plus pipeline translate into additional committed take-or-pay deployments will test how far Edgevana's pre-energized model can scale.

By Grace Kim