Issue #31: Voltage Park: The $1B GPU Cloud You’ve Never Heard Of

Most players in the AI infrastructure race shout from the rooftops.

Startups race to announce funding rounds. They tweet out model results. They drop press releases announcing the announcement of every rack they deploy.

But one player did the opposite.

While others were chasing headlines, they were deploying compute and optimising clusters.

$1 billion worth of H100s. 24,000 GPUS. Six Tier 3 data centres. No leasing. No debt. All paid for.

And now they’re done operating behind the scenes.

Welcome to Voltage Park.

The GPU Audio Companion Issue #31

Want the GPU breakdown without the reading? The Audio Companion does it for you—but only if you’re subscribed. If you can’t see it below, fix that here.

Company Background

Voltage Park didn’t launch with a flashy campaign or a parade of press interviews.

It stayed quiet. Built fast. Deployed faster.

Backed by crypto billionaire Jed McCaleb, founder of Mt. Gox, Ripple, and Stellar, Voltage Park took a different route.

No leasing and no debt.

Just $500 million in personal capital dropped straight into hardware. 24,000 GPUs. A total of $1 billion of infrastructure across six Tier 3 data centres. 

This means Voltage Park owns the hardware, controls the stack, and has avoided the financing bottlenecks choking much of the market.

This puts them in an interesting position of having no finance-driven GPU/h floor price.

That means there’s room for other players to make margin on Voltage Park’s hardware. And if you dig through enough reseller platforms, marketplaces, or “white-label” GPUaaS offerings in the US, you’ll probably find Voltage Park’s hardware underneath it all.

But that was 2024.

2025 is different.

TensorDock is integrated. Blackwell is on deck. And the goal is simple:

Offer world-class AI infrastructure at hyperscaler scale, without the hyperscaler lock-in.

Executive Team

Voltage Park is run by a lean crew of operators and cloud veterans:

• Ozan Kaya (CEO) – The former investment banker turned e-commerce CFO turned AI cloud pioneer leading the charge

• Saurabh Giri (Chief Product & Technology Officer) – Former director of engineering for Amazon Bedrock at AWS, now driving Voltage Park product development.

• Brandon Peccoralo (VP Sales & Partnerships) – Spearheading go-to-market and ecosystem deals.

• Jesica Church (VP Marketing) – Driving integrated marketing and brand experiences

• Elliot Darvick (People) – Scaling out team and people operations across the organisation.

• Tom Llewellyn-Smith (VP InfoSec & Compliance) – Security lead with a focus on enterprise readiness.

• Erim Tuc (Senior Counsel) – Keeping the ship buttoned up legally.

• Frank Basso (VP, Infrastructure & Support Ops) – Ensuring maximum uptime across all 24,000 GPUs.

The Edge

Voltage Park’s advantage comes down to control of capital, infra, and delivery:

• Zero Debt, Full Ownership: $1B in NVIDIA HGX H100s, fully paid for. No leasing, no liabilities. Just raw GPU firepower that’s already online.
  

• Optimised for Cost and Performance: All six data centres are in the lowest-cost US states, colocated next to major carrier hotels and peering exchanges for low-latency access and high throughput.
  

• Enterprise-Grade Everything: Each site runs 100Gbps multihomed internet, protected by Palo Alto firewalls in HA, backed by a fully tuned VAST eBox NAS, and supported 24/7 by in-house Voltage Park staff worldwide.
  

• Massive Sustainable Clusters: 12,000 GPUs and 10MW+ of expansion capacity currently being built out in their Seattle data centres, powered by 99% renewables, delivers a compelling offer to ESG-minded enterprise customers.

Recent Moves

• TensorDock Acquisition (March 2025): Voltage Park acquired TensorDock to bring their infrastructure directly to end users via an elastic GPU cloud interface. This move gives them a first-party front end with developer and enterprise reach.
  

• Expansion of On-Demand Access: Early benchmarking shows their API latency beats many incumbents. Combined with TensorDock’s provisioning engine, expect one of the fastest clouds in the game.
  

• Pre-Leasing Next-Gen Hardware: Liquid-cooled B200s, GB200s, and GB300s are already in the pipeline, with capacity being reserved ahead of public releases.

Read more on their blog here.

What’s Next?

Debt-free. Fully deployed. Built for scale.

Voltage Park enters 2025 in an enviable position. No financing gaps, no waiting on lead times, no existential scramble to fund the next rack. With one of the largest GPU clusters on the planet already live and optimised through a deep integration with Penguin Solutions, they’re not just talking scale. They’re operating it.

So while the rest of the market juggles Blackwell MOUs and term sheets, Voltage Park is already pre-leasing B200s, B300s, and GB200s.

The orders are piling in. More GPUs are rolling out. More data centre capacity is coming online. 

But the real edge? 

Recent hires, specifically in product.

Saurabh Giri was director of engineering for Amazon Bedrock. Bedrock streamlines generative AI application development on the AWS platform.

Put simply, it gives AWS users everything they need to make building GenAI apps as fast and simple as possible.

Given that many neoclouds are moving away from just GPUaaS towards fully integrated AIaaS, it’s not unreasonable to assume Saurabh will be taking Voltage Park in that direction too. Especially as the TensorDock acquisition already unlocked developer reach, liquidity, and faster utilisation. Because it would be a shame to let all that talent go to waste, right?

Couple this with their massive data centre footprint and total compute capacity, and it increasingly looks like while the rest of the industry was making noise, Voltage Park has quietly been building a machine to move faster than anyone else.

Expect aggressive growth this year.

More hardware. An integrated software stack. Faster access. Smarter deployment.

The capital stack is clean. The infra is live. And the expansion playbook is just getting started.

Only one question remains:

When the starting gun is fired, who will be able to keep up?

Keep The GPU Sharp and Independent

Good analysis isn’t free. And bad analysis? That’s usually paid for.

I want The GPU to stay sharp, independent, and free from corporate fluff. That means digging deeper, asking harder questions, and breaking down the world of GPU compute without a filter.

If you’ve found value in The GPU, consider upgrading to a paid subscription or supporting it below:

☕ Buy Me A Coffee → https://buymeacoffee.com/bbaldieri

It helps keep this newsletter unfiltered and worth reading.