Jul 23, 2025 · 56m · a16z
The U.S. Can’t Build AI Without These Materials
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of The a16z Podcast, host Erik Torenberg and partners Ryan McEntush and Erin Price-Wright interview Mariana Minerals founder Turner Caldwell to discuss how technology-enabled, vertically integrated mining can secure Western critical mineral supply chains for AI, electric vehicles, and national defense.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The host as informed peer 3.3 Guest teaching 4.2 Guest disagreement 0.5 The host pushing back 0.4
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 38:52 Equipment Manufacturing Limits Banter
Turner resists overextending vertical integration into commodity equipment manufacturing while responding playfully to Erin's challenge.
Hardest push from the host ▶ 38:52 Challenging Over-VerticalizationErin directly challenges Turner on whether Mariana's vertical integration will overreach into manufacturing its own mining equipment.
Biggest teaching moment ▶ 33:00 RL Process Control AnalogyTurner educates the hosts on process refinery control loops with a thousand variables using Google DeepMind's data center optimization as an intuitive benchmark.
The host holds their own ▶ 40:30 Full-Stack Mining ThesisErin articulates a16z's comprehensive investment thesis on full-stack vertical integration in hard tech and geopolitical supply security.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The host as informed peer | Guest teaching | Guest disagreement | The host pushing back | Why |
|---|---|---|---|---|---|---|
| Episode Hook: The Geopolitical Urgency of Critical Minerals | 1 | 4 | 0 | 0 | The host opens with broad setup questions regarding critical minerals and battery production. Turner provides a comprehensive overview of the processing chain from exploration through refining and magnet manufacturing. | |
| Bespoke Ore Processing and Diverse Mining Skillsets | 2 | 4 | 0 | 0 | Ryan asks an informed question about site-specific flowsheets vs standardized processing. Turner explains how heterogeneous ore bodies dictate custom metallurgical unit operations. | |
| Lessons from Tesla and Misaligned Incentive Structures | 2 | 5 | 1 | 0 | Erin asks about Turner's career path from Tesla. Turner reframes the economics of mining versus manufacturing, noting how volume increases raise price expectations in mining due to supply constraints. | |
| Manufacturing Precision, Kaizen, and Vertical Integration | 3 | 4 | 0 | 0 | Erin draws parallels between Tesla's early vertical integration and Mariana's thesis. Turner details Asian battery manufacturing rigor and how vertical integration transfers partner risk profiles internally. | |
| Labor Contracting and Remote Mine Site Realities | 2 | 4 | 1 | 0 | Ryan probes global mine site realities. Turner clarifies that modern mine sites in Western nations are calmer than expected, contrasting them with higher labor intensity in developing nations. | |
| Applying Reinforcement Learning and Modern Software to Infrastructure | 2 | 5 | 0 | 0 | Erin asks about Turner's transition from Tesla to Mariana. Turner outlines the technological inflection point where reinforcement learning enables no-human-in-the-loop multivariable optimization for process plants. | |
| Incumbent Mining Giants and the Talent Retention Problem | 3 | 5 | 1 | 1 | Ryan and Erin ask about incumbent mining giants and their digital arms. Turner explains the heavy reliance on external consultants and EPC man-hour selling models, highlighting acute talent retention challenges. | |
| The Pilot Trap: Why Point Solutions Struggle in Mining | 5 | 3 | 1 | 1 | Erin offers a strong venture perspective on why point solutions get trapped in endless pilot loops with risk-averse mining incumbents. Turner agrees, breaking down the mathematical risk evaluation operators face. | |
| Chinese Industrial Scale and Talent Pool Superiority | 4 | 5 | 1 | 0 | Ryan asks why Chinese mining companies scaled rapidly. Turner highlights China's massive skilled talent pool and details the junior mining 'orphan period' asset dynamic that Mariana targets. | |
| Technical Deep-Dive: CapitalProjectOS and PlantOS | 4 | 6 | 1 | 1 | Erin engages on commissioning timelines and banter on civil construction versus deconstruction. Turner provides a deep technical breakdown of CapitalProjectOS and PlantOS, citing DeepMind's thermal optimization as a benchmark. | |
| Go-to-Market Strategy and Supply Base Bottlenecks | 5 | 4 | 2 | 3 | Erin directly pushes back on the limits of vertical integration, asking if Mariana plans to build its own mining equipment. Turner responds playfully while defending partner incentive alignment. | |
| a16z’s Thesis: Venture Capital Meets Hard Tech and National Security | 7 | 1 | 0 | 0 | Erin and Ryan lead the narrative, laying out a16z's full-stack thesis on vertical hard tech investments, geopolitical urgency, and shifting cultural mindsets around domestic mining. | |
| Evaluating Demand for Copper, Lithium, Nickel, and Rare Earths | 3 | 5 | 0 | 0 | Ryan prompts a breakdown of specific critical minerals. Turner delivers an extensive overview of market opportunities across copper, lithium, aluminum, and nickel, highlighting counter-cyclical investment strategies. | |
| Permitting Red Tape and Demand-Side Government Policy | 4 | 5 | 0 | 0 | Ryan and Erin ask about permitting bottlenecks and policy recommendations. Turner details BLM exploration permit hurdles and advocates for demand-side government support such as price floors. | |
| International Expansion and Mariana’s 10-Year Master Plan | 2 | 3 | 0 | 0 | The hosts ask about international strategy and long-term vision. Turner presents Mariana's 10-year goal to build 10 projects and restore Western industrial infrastructure building capability. |
Statements from this episode (20)
Caldwell: Raw metals are the primary cost driver in battery manufacturing
“When I was working on factory design and construction, The most expensive thing was actually the equipment that goes inside the factory. Then when we started working on cell manufacturing, kind of realized that the expensive part of making cells is the stuff t…”
Caldwell: Mining pricing dynamics invert traditional manufacturing economies of scale
“Like in the mining industry, more demand, this is totally different than manufacturing, more demand if you have a higher volume, the expectation is that the price goes up. And because it's a, you know, constrained supply, and so the pricing dynamics are totall…”
Caldwell: Tesla vertically integrated early because no subcomponent market existed
“Tesla had to vertically integrate early because people just weren't making the parts that were needed. Like that was a, it was a do or die, There was no incentive. There was no market for people to build the kind of like sub components that were required. And …”
Caldwell: Vertical integration transfers partner risk into your company's risk profile
“At the end of the day, when you vertically integrate, you are transferring the risk profile of your partner kind of into like your new expanded risk profile. And so you need to be really confident or at least believe that you can, you are better positioned to …”
Caldwell: Reinforcement learning enables fully automated control of industrial plants
“We're at this point where kind of like the compute and machine learning, reinforcement learning Do really enable, enable you to go no humans in the loop and how a lot of these plants are controlled.”
Caldwell: Humans are poorly suited for multivariable plant optimization compared to RL
“As you like build more of that large scale infrastructure and kind of like see the problems that humans have to solve on a daily basis, it becomes like pretty obvious that these are problems that humans don't, aren't best positioned to solve. Like these are la…”
Turner Caldwell: Long mining build cycles trap tech startups in pilots
“But you'll end up doing a lot of pilots and because they don't build enough plants kind of sequentially, like, they'll build one big mine every five years if that. And the, like, there aren't just, there just aren't a lot of opportunities to get into a commerc…”
Caldwell: First-of-a-kind facilities face major project financing hurdles
“It's hard to get project finance on a first of a kind facility where you're demonstrating a new unit operation for the first time.”
Caldwell: Mining pump lead times are 30 weeks in Australia versus days in China
“The supply chain for like a new pump in Australia could take, you know, 30 weeks and getting that exact same pump but with a mine in China, it shows up in, you know, a week or three days.”
Caldwell: Reinforcement learning will automate scaling of new chemical processes
“That like scale jump is actually something that we think that RL and will enable with like a pretty meaningful like pace adjustment where you don't need the humans to kind of like fine tune the process conditions around a new process chemistry because the, you…”
Price-Wright: Mining is one of the largest untapped tech markets
“Mining sort of screams one of the largest markets in the world. Very little adoption of technology.”
Price-Wright: Building a mining tech company requires end-to-end vertical integration
“We actually do believe you have to control every single piece of the entire journey, the entire life cycle of, you know, an atom of metal end to end to actually be able to build a tech company here. This is not about, you know, a point solution for one particu…”
Price-Wright: US supply chains are dangerously reliant on China
“Our supply chains are highly reliant on China. It's a huge problem. We have to figure out how to address it. And that means investing in mining in the US again.”
Caldwell: Aluminum is the most consumed metal in defense
“It's actually, like, the number one most consumed metal in defense applications.”
Caldwell: Global lithium production capacity must 4x within 10 years
“Lithium, you know, needs to four X in the next, in terms of production capacity in the next 10 years, roughly, in order for the, ah, batteries that we want to build to be built.”
Caldwell: Indonesia accounts for roughly 70% of global nickel supply
“Indonesian kind of like production capacity has scaled to the point where it's now something like 70% of global nickel comes out of Indonesia.”
Caldwell: Build mining infrastructure at the trough of commodity cycles
“You actually want to be building infrastructure at the bottom of commodity cycles, not at the top of commodity cycles.”
Caldwell: Mariana Minerals is targeting lithium at cycle lows
“Lithium is like exactly in this position right now. And that's why we're focused on lithium.”