a16z · the Podbrain notes ·
4 min read

Energy, Minerals, and the Physical Stack Behind AI

Turner Caldwell, co-founder and CEO of Mariana Minerals, and Drew Baglino, founder and CEO of Heron Power, both spent extensive time at Tesla building critical infrastructure. Caldwell led Tesla's minerals and metals team, while Baglino spent 18 years running powertrain and energy, including responsibility for the...

a16z a16z
Subscribe to Notes Upgrade
a16z episode thumbnail: Energy, Minerals, and the Physical Stack Behind AI
a16z
Key Takeaways
  1. 01

    The U.S. is 50 years behind China on critical minerals supply, requiring faster design, build, and ramp-up capabilities

  2. 02

    Drew Baglino built Tesla's MegaPack factory in Lathrop, California in just 11 months from a JCPenney warehouse

  3. 03

    Labor costs represent less than 10% of factory cost of goods sold today due to automation advances

  4. 04

    Mariana uses reinforcement learning to remove humans from refinery control loops for autonomous operations

  5. 05

    Tesla's model provides techno-optimism, risk appetite, and commitment to not abandoning worthwhile projects

  6. 06

    Silicon carbide power semiconductors enable solid-state transformers to replace steel, oil, and copper systems

  7. 07

    Both companies will create over 500 construction jobs each in the next 18 months across their facilities

Get the latest ideas from a16z.

Plus the best new takeaways from other top podcasts — read in minutes, not hours.

or

By continuing, you agree to Podbrain's Terms and Privacy Policy.

These notes may contain occasional inaccuracies. Learn how Podbrain notes are made

Turner Caldwell, co-founder and CEO of Mariana Minerals, and Drew Baglino, founder and CEO of Heron Power, both spent extensive time at Tesla building critical infrastructure. Caldwell led Tesla's minerals and metals team, while Baglino spent 18 years running powertrain and energy, including responsibility for the MegaPack and 4680 battery programs.

The conversation explores how American reindustrialization depends on physical infrastructure rather than just algorithms. Mariana Minerals operates as a software-first mining and refining company, using reinforcement learning for autonomous control while building actual copper mines and lithium refineries. Heron Power develops solid-state transformers using silicon carbide semiconductors to modernize grid infrastructure that still relies on mechanical systems designed before World War II.

Both founders argue that the Tesla playbook of rapid iteration, risk tolerance, and vertical integration can be applied to transform America's critical minerals supply chain and power grid infrastructure, addressing the 50-year gap behind China in these strategic sectors.

Software-First Approach to Critical Minerals

Mariana Minerals employs a quarter of its workforce as software and machine learning engineers developing three core operating systems: Capital Project OS for project lifecycle management, Plant OS for refinery control using reinforcement learning, and MineOS for autonomous mining operations.

"We're making a big bet on autonomy in refineries, where we use reinforcement learning to actually remove humans from the loop in determining how refineries operate" - Turner, addressing the challenge of managing highly variable feedstock from heterogeneous Earth materials.

The company operates a copper mine in southeast Utah producing high-purity materials and is building a lithium refinery in Texas, with plans for 10 projects over the next decade.

Grid Modernization Through Power Electronics

Heron Power builds solid-state transformers using silicon carbide semiconductors to replace traditional steel, oil, and copper systems in power conversion at data centers and large-scale energy installations.

"The systems underpinning the grid today are the same largely mechanical systems that were developed over 100 years ago" - Drew, explaining why grid infrastructure hasn't kept pace with edge innovations like EVs and batteries.

The U.S. leads in silicon carbide production, the key power semiconductor enabling these transformers, making domestic manufacturing strategically advantageous for capturing technology benefits.

Tesla's Execution Model Applied to Industrial Scale

Tesla's approach provides three key advantages: techno-optimism about innovating on archaic systems, appetite for risk enabling fast decision-making, and commitment to fighting through challenges when outcomes justify the effort.

"If the outcome is worth it, Tesla will fight through the challenges of getting to that outcome" - Turner, contrasting with traditional companies that abandon autonomy projects after initial failures.

Drew built Tesla's MegaPack factory in Lathrop, California in 11 months, demonstrating that rapid industrial deployment is possible with proper alignment and local jurisdiction cooperation.

The do-or-die startup environment where "whether or not the paycheck will clear" depends on execution creates focus and attracts top talent to high-growth opportunities.

Manufacturing Economics and Supply Chain Strategy

Labor costs represent less than 10% of factory cost of goods sold in modern automated facilities, making the U.S. competitive with China on direct manufacturing costs.

"Everything that you could possibly need to build a car, which has 7,000 parts in it, is within less than a three hours drive" - Drew, describing China's co-located industrial areas as a model for U.S. competitiveness.

Co-located supply chains in the United States would dramatically reduce logistics costs and time while creating high-paying technical jobs requiring skill and training.

Building Industrial Workforce in America

Both companies will create over 500 construction jobs each in the next 18 months, with Turner's lithium and copper projects adding additional full-time positions as operations scale.

Drew hired talent from analog industries like high-speed bottling plants and syringe manufacturing for Tesla's 50 gigawatt hour battery facility in Texas, finding "immense depth of talent" when thinking creatively.

The mining industry faces a 35-year talent attrition challenge, but oil and gas workers provide transferable skills, while software optimization algorithms from ride-sharing and ad tech apply to plant operations.

Policy Recommendations for Industrial Acceleration

Turner advocates applying the last 50 years of oil and gas policy tools to create a minerals mandate, providing incentive structures that mobilize private capital with long-term market confidence.

Drew calls for durable industrial policy, coordinated federal-state identification of energy and manufacturing build-out zones, and a federal highway trust fund for grid infrastructure.

"How do we find a master plan of build out of linear infrastructure that maybe connects those manufacturing energy build out zones to improve resilience, reduce costs, and really move us forward as a nation" - Drew.

Resources Mentioned

and maybe 150 spreadsheets that are kind scattered around an operation

ke, what is the tech stack that they're comfortable with? And then for the most part, it is pen and paper and maybe 150 spreadsheets that are kind scattered around an operation. And what you need to

a16z
From a16z. Get a note like this from every new episode.
Subscribe to Notes Upgrade

These notes may contain occasional inaccuracies. Learn how Podbrain notes are made

0 / 0
Link copied