# Energy, Minerals, and the Physical Stack Behind AI Podcast: The a16z Show Published: May 13, 2026 Reading time: 13 min Canonical: https://podbrew.app/briefs/the-a16z-show-energy-minerals-and-the-physical-stack-behind-ai Erin Price-Wright speaks with Turner Caldwell and Drew Baglino about the foundational challenges facing America's critical mineral supply and power infrastructure. They examine the urgent need to address the physical stack underpinning the burgeoning AI economy. The discussion highlights how the US lags significantly in critical mineral supply and relies on a century-old power grid. Caldwell and Baglino detail how new approaches like automation, reinforcement learning, and solid-state transformers can compress timelines for mining and refining, and modernize grid technology to meet current and future demands. This conversation is crucial for understanding how the US can close strategic gaps, revitalize domestic industrial capacity, and ensure its competitiveness on the global stage. It underscores the importance of resilient infrastructure and advanced materials for national security and the future of technology. ## Key takeaways - The US lags 50 years behind in critical mineral supply and operates a power grid based on pre-World War II mechanical systems, both of which are critical for AI and re-industrialization. - Despite advancements in edge technologies like chips and batteries, the foundational physical infrastructure for energy and materials has not evolved sufficiently to meet current or future demands. - The increasing energy and material requirements of AI should be treated as a national call to action and an opportunity to rebuild America's industrial backbone. - Mariana Minerals integrates substantial software engineering to build operating systems for automated and accelerated critical mineral mining and refining. - Haren Power is advancing grid technology with solid-state transformers, utilizing modern power semiconductors and software to replace conventional components and enhance power conversion efficiency. - Building and operating critical mineral projects typically takes five years for construction and an additional three to five years to reach full operational capacity. - To bridge the critical mineral gap, the US must not only streamline permitting but also accelerate the entire project build-out process faster than other nations. - The core grid relies on century-old, largely mechanical systems that lack modern control and monitoring, despite innovation at the grid's edge (EVs, supercharging, energy storage). - Rapid, impactful infrastructure development is achievable, as demonstrated by the construction of a Megafactory in 11 months. - Local jurisdictional alignment is crucial for accelerating US manufacturing projects; a supportive local government can expedite processes. - In modern automated factories, labor costs are a minor factor (under 10% of COGS) in the competitiveness of US manufacturing. - Co-locating critical supply chains within the United States, reducing logistics costs and time, is a significant unlock for competitive US manufacturing. - Mariana Minerals leverages large language models for faster engineering and procurement, alongside reinforcement learning for autonomous refinery operations, eliminating manual tuning for complex processes. - Autonomy is vital due to highly variable raw materials and the scarcity of skilled US labor capable of consistently managing the intricate, constantly changing parameters of modern critical mineral refining. - Accelerating software uptake in industrial settings requires designing tools for operators' specific problems and integrating software engineers directly with operating teams. - Tesla's core philosophy includes a high degree of techno-optimism and a fundamental belief in innovating on old and archaic industrial systems. - Tesla's high risk appetite fosters rapid decision-making and empowers teams to innovate without fear of failure. - To staff new industrial operations, companies must creatively recruit from 'analog industries' where transferable skills (e.g., high-speed manufacturing, complex optimization) can be adapted. - Implement a 'minerals mandate' based on past energy mandates to create stable incentives that mobilize private capital for long-term US industrial development. - Create a federal 'highway trust fund' for the electricity grid to establish a master plan for linear infrastructure, connecting key zones to improve resilience and reduce costs. ## 00:00 - 04:01 US Faces Critical Mineral and Grid Deficiencies Essential for AI The United States lags significantly in foundational infrastructure, with its critical mineral supply trailing China by 50 years and its power grid relying on mechanical systems designed before World War II. Despite rapid innovation in areas like batteries, software, and chips, the underlying physical infrastructure, such as grid transformers and mineral refineries, has not kept pace with modern demands. This outdated infrastructure poses a serious challenge to America's ambitions for AI dominance and broader re-industrialization. These are inherently physical projects, requiring vast amounts of materials, energy, and the efficient movement of electricity. The current system is ill-equipped to support the escalating demands of new factories, autonomous systems, and advanced AI models. Concerns that AI will overstrain the faltering grid and demand more energy than can be supplied are valid. However, instead of deterring progress, these challenges should be viewed as a national call to action. The US has historically excelled at large-scale national projects and possesses the capacity to overcome these hurdles. To rebuild the industrial backbone of the United States and secure its AI future, a comprehensive rethinking of the entire stack is necessary. This involves overhauling critical mineral extraction and refining, modernizing energy generation and transmission, and developing new methods to build and interconnect infrastructure with the speed required by contemporary technological advancements. > the constraint on America's AI future, and as I mentioned, I think reindustrialization more broadly is in many ways atoms and not algorithms. ## 04:01 - 08:02 Mariana Minerals and Haren Power introduce their software-first approaches to critical mineral mining and power electronics for grid modernization. Mariana Minerals is a software-first company focused on mining and refining critical minerals. A quarter of its staff are software engineers developing three core operating systems: Project OS for project lifecycle management, Plant OS for refinery control, and Mine OS for autonomous mining operations using reinforcement learning. The company builds and operates mineral projects, currently running a copper mine in Utah and constructing a lithium refinery in Texas, with a goal of ten new projects in ten years. Haren Power builds power electronics to accelerate the electricity sector. Leveraging decades of improvements in power transistors, the company is developing solid-state transformers that use silicon and software to replace traditional materials like steel, oil, and copper in power conversion. These technologies are crucial for modernizing the grid and supporting growing electricity demand from data centers and large-scale energy installations. Both companies address the urgent need for the US to re-shore critical supply chains and innovate in physical industries. Drew Baglino emphasizes that the US developed key power semiconductor technology like silicon carbide and should commercialize its benefits domestically. Turner Caldwell highlights that the US is significantly behind in critical mineral supply, particularly compared to China, and innovation is essential to accelerate new mineral capacity development. > the US is fifty years behind on critical mineral supply. ## 08:02 - 08:51 Closing the Critical Mineral Gap Requires Accelerated Project Development Developing critical mineral projects involves extensive timelines, extending well beyond initial permitting. After securing permits, construction can take five years, followed by another three to five years for the facility to reach full operational capacity. This means a single project can demand nearly a decade from the start of building to achieving its intended output. Given that the US is significantly behind in critical mineral supply, simply reducing regulatory hurdles is not enough. To genuinely catch up with leading nations like China, the US must develop new projects at an even faster pace than the current global standard. This acceleration is crucial to overcome the existing deficit and secure a reliable domestic supply. > Even if we start to lower the burdens to play catch up with China, we actually have to go faster than China does. ## 08:51 - 10:02 Drew Baglino explains his pivot from grid-edge innovation to tackling the outdated core grid infrastructure. Drew Baglino, previously at Tesla, had a front-row seat to innovations at the grid's edge, including the expansion of electric vehicles, the supercharging network, and large-scale grid storage like Megapacks. These efforts significantly advanced the accessibility and omnipresence of clean energy solutions for consumers. However, Baglino observed a stark contrast between this rapid innovation at the edge and the stagnant core grid. He notes that the fundamental systems supporting the grid today are largely mechanical, having been developed over a hundred years ago. This outdated infrastructure lacks essential control and monitoring capabilities, making the system overbuilt and fragile. The reliance on archaic technology also limits the number of equipment suppliers, many of whom are based overseas, which Baglino views as a vulnerability for critical infrastructure in the United States. He is confident that rapid development is feasible, drawing on his experience building the Megafactory in Lathrop, California, from a JCPenney warehouse in just 11 months. > The systems underpinning the grid today are the same largely mechanical systems that were developed over a hundred years ago. ## 10:02 - 12:02 Accelerating US Manufacturing Through Local Alignment and Co-located Supply Chains Reindustrializing the US and building critical infrastructure requires gaining alignment with local jurisdictions. These local bodies have the power to either impede projects by saying no at every step or accelerate them by saying yes. When jurisdictions actively support manufacturing initiatives, it can be transformational for project timelines and outcomes. Contrary to common belief, labor costs are not the primary barrier to US manufacturing competitiveness. In today's highly automated factories, the labor differential between the US and countries like China accounts for less than 10%, potentially even less than 5%, of the cost of goods sold. The true driver of competitiveness lies in the supply chain. The key is developing co-located critical supply chains within the United States. This reduces logistics costs and time significantly, mirroring successful models seen in places like China, where all necessary parts for complex products like a car (which has 7,000 parts) are within a three-hour drive. Achieving this kind of co-location for the supply base in the US, coupled with advanced automation, represents a major opportunity. It would not only enhance manufacturing competitiveness but also create a substantial number of high-paying and important jobs. > Getting to that kind of co-location of the supply base in the United States Would be a major unlock, along with automation, while still providing immense numbers of, of, of like high-paying, you know, im-important jobs. ## 12:02 - 14:03 Mariana Minerals Utilizes AI and Full-Chain Autonomy for US Critical Mineral Sovereignty Mariana Minerals aims to help the US reclaim critical mineral sovereignty by focusing on the entire supply chain, from extraction to processing and refining. This approach addresses the current vulnerability where much of the processing capacity for rare earths and critical minerals resides overseas, particularly in geopolitical rivals like China. The company makes a significant bet on autonomy, using large language models to accelerate engineering and procurement lifecycles. They also implement autonomous control for construction operations, optimizing resource balancing between available materials, tasks, and personnel algorithmically. In refineries, Mariana Minerals employs reinforcement learning to remove humans from the loop in determining operational parameters. This is crucial because the earth's heterogeneous nature means feedstocks are highly variable, requiring constant tuning of temperatures, flow rates, chemical additions, and residence times in complex refining circuits. A key driver for this autonomous approach is the lack of a sufficient labor pool in the US with the embedded know-how to quickly operate and manage the variability of modern refineries. Autonomy also extends to mining operations, where thousands of decisions are made daily. > we don't have that labor pool here that has that embedded know-how, that can walk up to a refinery and quickly get it operating on spec and then also manage that variability. ## 14:03 - 16:04 Deep Software Integration and the Tesla Mindset for Industrial Innovation Software penetration in industrial operations, such as plants and mines, is often limited by operating teams' reliance on outdated tools like pen-and-paper or scattered spreadsheets. To truly accelerate the adoption of new technology, it is crucial to understand the core problems faced by these teams and design software specifically for their needs. The most effective approach involves embedding software engineers directly alongside operating teams, ensuring the tools are built for the actual users. This method, rather than a 'forward deploy engineer' strategy, yields better results for optimizing assets by closely aligning software development with operational realities. Drawing from the Tesla model, a key differentiator from traditional industrial companies is a strong techno-optimism and a deep belief that even long-standing, archaic industrial systems can be fundamentally innovated. This core conviction drives a different approach to building and optimizing operations. > The belief that you can innovate on systems that are old and archaic, is like at the core of the company. ## 16:04 - 18:04 Tesla's High-Stakes Environment and Purpose Attract Top Talent Tesla maintains a high appetite for risk, which promotes extremely fast decision-making and allows teams to move quickly without the fear of making wrong choices. This contrasts with other industries, such as autonomy in mining, where companies often abandon projects after initial failures. The company demonstrates a firm commitment to achieving desired outcomes, consistently fighting through challenges as long as the objective is deemed worthwhile. This relentless persistence, coupled with situations where the company's future and even paychecks depend on team execution, creates a uniquely focused and "do or die" reality for employees. This environment, combined with a clear vision of the company's purpose, acts as a powerful beacon for attracting top talent. People are drawn to a high-growth setting where they can contribute to a meaningful mission, despite the intense pressure and high stakes involved. > I hate to say do or die, but it's equivalent to that. ## 18:04 - 21:38 Creative Strategies for Building a Modern US Industrial Workforce As the US undergoes reindustrialization, new factories and facilities are creating thousands of jobs. For instance, initial lithium and copper projects are expected to add over 500 construction and full-time positions, with many more as operations scale. Similarly, a new large factory is anticipated to create around 500 jobs as part of an expanding industrial footprint. Building this modern industrial workforce requires creative hiring approaches. When specialized talent pools are scarce, companies look to 'analog industries' for transferable skills. One example is recruiting engineers and production associates from high-speed bottling plants or syringe manufacturing facilities for battery production, recognizing that the core skills in precision and volume manufacturing are highly adaptable. The mining industry faces a similar challenge, with significant labor pool attrition over the past 35 years. Companies are sourcing talent from the oil and gas sector and even the software industry for optimization algorithms, which share similarities with those used in common apps. This highlights the broad applicability of skills across different sectors within the US talent pool. Beyond finding transferable skills, some industries like mining must actively combat negative public perception. The industry's image as 'villains in every movie' necessitates efforts to rebrand and make it appealing to attract new talent, underscoring the importance of building a strong 'talent magnet' that goes beyond just job creation. > Make mining sexy again. ## 21:38 - 24:06 Proposing a Minerals Mandate, Durable Industrial Policy, and Federal Grid Investment To revitalize US industries, one speaker suggests implementing a 'minerals mandate' by drawing inspiration from the past fifty years of oil and gas energy mandates. The core idea is to establish stable incentive structures that attract and mobilize private capital, ensuring investors have confidence that long-term commitments will not be undermined, especially in sectors that have not seen significant domestic build-out in decades. Another key recommendation is for durable industrial policy that allows companies to plan long-term. This includes a concerted effort between federal and state governments to identify and develop co-located energy and manufacturing build-out zones within the United States. The aim is to create supportive environments where local jurisdictions actively collaborate rather than obstruct projects, fostering domestic manufacturing and resilient supply chains. A crucial enabler for this growth is the electricity sector, and a 'federal highway trust fund' for the grid is proposed. This fund would address the current fragmented, patchwork system by establishing a master plan for linear infrastructure build-out. Such a framework would connect manufacturing and energy zones, enhancing grid resilience, reducing costs, and advancing national progress. I like the idea of a federal highway trust fund for the, for the, for the grid. it never has existed. that's sort of why we have this patchwork. > I like the idea of a federal highway trust fund for the, for the, for the grid. it never has existed. that's sort of why we have this patchwork. --- Get podcast briefs for shows you follow: https://podbrew.app/