Former Energy Department officials' startup Mavria raises $130 million to build data centers in a factory

Startups & Funding

Mavria Raises $130 Million at $1 Billion Valuation to Build Data Centers in Factories

Mavria, founded by former Energy Department DOGE officials, raised $130 million at a $1 billion valuation led by Paradigm to build standardized five-megawatt data center modules in a factory.

By
Grace Kim
Filed
Channel
Startups & Funding
Read
3 min read

Mavria has raised $130 million at a $1 billion valuation to build data centers in a factory, the company says. Crypto investor Paradigm led the round, the company confirmed, matching a Wall Street Journal exclusive published Monday. Liquid 2, Critical Frontier and Dream Ventures also invested.

The Texas startup was founded by three people who helped overhaul the Energy Department under President Trump: Adam Blake, Seth Cohen and Nick Poradek. The valuation and the investor list appear on Mavria's own website.

Who are the founders?

Blake and Cohen co-led the DOGE team at the Energy Department. Cohen served as chief counsel for nuclear policy there. Together the pair carried out the president's nuclear executive orders across the government, according to Mavria.

Blake previously advised the secretary of energy, the company says. He also sold a software company to Vista Equity Partners. Poradek sold a manufacturing company to Liberty Energy. The three founders combine policy, software and industrial backgrounds.

What is Terrain?

Terrain is Mavria's first product. It combines power, cooling, controls and IT pods into a standardized, factory-built data center system. Terrain 01 is being built in Houston now, the company says.

"Terrain replaces one-off data center construction with repeatable industrial manufacturing," Mavria says.

According to the Journal's report as summarized by The Verardo Group, each unit carries five megawatts of capacity. The units are designed for fast assembly and for placement near available electricity.

The company has not named customers. It has not published prices or delivery dates. Terrain 01 remains the only confirmed build.

Why factory-built modules?

The pitch targets the same bottleneck that has developers hunting for used gas turbines: getting compute capacity where power already exists. Grid interconnection delays and local opposition have slowed traditional ground-up data center construction across the United States.

Cohen framed the problem directly in July, while Mavria was still in stealth. He said data center siting had become the "second bottleneck to AI deployment" in the U.S., after nuclear waste siting. He added that projects face strong local opposition.

A five-megawatt unit size implies scale demands of a different order. A single one-gigawatt AI campus would need roughly 200 Mavria modules, based on the reported unit size. That calculation shows the gap between modular products and hyperscale demand. It also shows the potential repeat-manufacturing volume if the model works.

What changes for the supply side?

The factory-built approach shifts data center construction from bespoke projects to industrial production. Standardized modules could compress build timelines and move capacity closer to stranded or cheap power. The five-megawatt format suits distributed siting near existing substations and generation.

Mavria's founder credentials matter for permitting. Blake and Cohen executed nuclear executive orders across the federal government. That experience maps onto the siting and regulatory fights Cohen identified as the industry's core constraint.

The company has disclosed no revenue. It has announced no customer commitments. The $130 million round and the $1 billion valuation represent investor confidence in the founding team and the modular thesis rather than proven shipments.

What comes next?

Terrain 01 in Houston is the first proof point. Customer announcements, pricing and delivery schedules remain unpublished. If the modular model delivers, Mavria will compete on manufacturing speed and siting flexibility rather than raw scale — and the roughly 200 modules needed for a one-gigawatt campus define the volume race ahead.

Source: mlq.ai

Share this article:

More from Grace Kim

Grace Kim

Show full bio

Market editor covering people, launches and day-to-day practice in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.

803 articles

Related articles

« Previous articleNext article »