Mariana Minerals Raises $310M for AI-Run US Mines

Behind every chip, data center, and power line the AI industry is racing to build sits a far less glamorous dependency: raw metal. A San Francisco startup just raised serious capital to fix that supply chain domestically, using AI to run the mines itself rather than simply selling software to existing mining companies. Mariana Minerals raised $310 million in Series B funding led by Khosla Ventures, with continued backing from Andreessen Horowitz and Breakthrough Energy Ventures, bringing the company's total capital raised to roughly $400 million at a $1.5 billion valuation, according to Fortune's exclusive reporting on the round.

Mariana isn't a software vendor selling tools to miners. It directly acquires, builds, and operates mines and refineries itself, using its proprietary MarianaOS platform to manage the entire process, from construction and mine planning through refining, according to Citybiz's reporting on the raise.

This isn't purely theoretical. Mariana's Copper One project in southeastern Utah, acquired in late 2025, resumed mining operations within four months of acquisition under what the company describes as an autonomous operating model, targeting annual production of 50,000 metric tons of refined copper. The company is separately advancing Lithium One in East Texas, a facility designed to extract lithium from produced water generated by existing oil and gas operations, with commercial production scheduled for the first half of 2027, according to PR Newswire's coverage of the raise.

MarianaOS itself splits into three integrated systems: CapitalProjectOS for construction and procurement, MineOS for coordinating day-to-day mine operations, and PlantOS, which uses sensors and reinforcement learning to optimize refining in real time, according to The AI Insider's reporting on the platform's architecture.

Khosla Ventures founder Vinod Khosla framed the strategic stakes in explicitly AI-driven terms: "Critical minerals are the materials that decide whether America builds its own future or keeps depending on China to build it instead. We backed Mariana because Turner is the kind of operator who can actually build and run AI-driven mines and refineries, faster and cleaner than anyone thought possible." That framing connects directly to a dependency most AI infrastructure coverage overlooks entirely, one Uber founder and Atoms CEO Travis Kalanick made explicit to Fortune: "The AI revolution, physically speaking, depends on the mining of a huge amount of minerals and metals. Data centers, chips, the grid, robots, EVs, defense systems: it all starts with copper and other critical minerals."

Not everyone views the valuation as fully justified yet. Independent analysis from RecodeX noted Mariana has no disclosed revenue or external customer information, and its actual production hasn't been independently audited, meaning investors priced the company at $1.5 billion largely on the strength of its team and operating model rather than proven output, a genuine risk factor worth understanding alongside the broader pattern of aggressive AI-adjacent valuations we've tracked in our coverage of DeepX's own $2.2 billion chip valuation surge earlier this week.

Mariana Minerals' raise is a useful reminder that AI infrastructure dependencies extend well beyond chips and data centers into physical, resource-constrained supply chains that most business leaders never think to evaluate directly. For businesses in manufacturing, defense, or any sector dependent on copper, lithium, or other critical minerals, Mariana's vertically integrated, AI-driven approach is worth watching as a potential new domestic supply source, particularly given the geopolitical urgency around reducing dependence on Chinese-controlled mineral supply chains.

Mariana Minerals raised $310 million in Series B funding led by Khosla Ventures at a $1.5 billion valuation, aiming to build and operate AI-run copper and lithium mines domestically. The company's Copper One project in Utah is already producing under an autonomous operating model, while its Lithium One facility in Texas targets commercial production in 2027. The raise reflects growing investor recognition that AI infrastructure depends on physical mineral supply chains, not just chips and data centers, even as some analysts question whether the valuation is justified given the company's currently undisclosed revenue and unaudited production figures.