Geothermal Resources May Be 100x Underestimated — Zanskar's Claim
Featuring: David Roberts (Volts podcast host) × Joel Edwards (Zanskar CTO) × Carl Hoyland (Zanskar CEO)
Bottom line
US conventional geothermal resources may be ~100x larger than government estimates—terawatts instead of tens of gigawatts—potentially enough to supply 100% of US electricity demand as carbon-free baseload power.
3-Line Summary
Zanskar's founders argue that the USGS 2008 assessment (9,000 MW discovered, 30,000 MW undiscovered) relied on crude 1970s-80s data, and modern AI-driven exploration paired with iterative field testing across 5 million acres reveals both 10x more undiscovered sites AND 10x more power per site. If validated, conventional geothermal using proven drilling tech would challenge the necessity of enhanced geothermal systems (EGS) that artificially create permeability. The company tested 5 million acres over five years and at New Mexico's Lightning Dock field demonstrated 10-40x more power from a single well than industry estimates.
3 Key Points
1. Half of 36 existing US geothermal fields found by accident: Of 36 operating fields, half were discovered when farmers drilled for water or gold companies drilled for gold; the other half had surface manifestations like hot springs. This suggests systematic exploration has barely begun—analogous to late-1800s "peak oil" declarations before major Pennsylvania discoveries, Hoyland notes.
2. Natural permeability 10-100x better than engineered: Permeability created artificially in EGS is "an order of magnitude and sometimes multiple orders of magnitude less" than natural permeability, which can be "cavern size," Hoyland explains. This cost advantage will persist "indefinitely into the future"—similar to how Saudi Arabia's conventional Ghawar field still beats US shale economics decades into the shale revolution.
3. Lightning Dock proves most productive pumped well in US: Acquired failing Lightning Dock field in New Mexico, drilled one well into deeper resource within a year, brought plant back to full nameplate capacity and profitability. The well delivers 100+ MW thermal energy at wellhead (equivalent to a small modular nuclear reactor) from a well "you can wrap your arms around," generating more electricity per well than any other pumped well in the country. Operating at full capacity for 12+ months with no temperature decline, benchmarked against Nevada's published dataset (200+ wells) and other known pumped wells nationwide.
Editorial Perspective
Platform transitions often see the biggest discoveries after the moment everyone believes "nothing's left." Ghawar field was found a hundred years after the first peak oil declaration and still underpins one of history's most valuable companies. If Zanskar's claim holds, geothermal leaps from "niche renewable" to "lowest-cost carbon-free baseload capable of meeting all US electricity demand." But this remains a claim—the proof will come from USGS's updated assessment and the performance of multiple greenfield plants Zanskar plans to commission in coming years.
Source: Volts "Is there way more conventional geothermal than we thought?" (September 9, 2026)
https://www.volts.wtf/p/is-there-way-more-conventional-geothermal
AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.
# Conventional Geothermal# Zanskar# AI Exploration# Enhanced Geothermal Systems (EGS)
Rocket Reusability Is Halfway Done — Second-Stage Recovery Decides 10x Space Economy Growth
Featuring
Kyrie (Payload Space) × Andy Lapsa (CEO & Co-founder, Stoke Space; former senior engineer at Blue Origin)
Bottom line
Even with SpaceX's Falcon 9 first-stage reusability, the launch industry bottleneck remains unsolved — full reusability including the upper stage is the key to making space a true extension of Earth's economy.
3-Line Summary
Stoke Space announced a $1 billion raise and Nova Block 2, a 15-ton-to-LEO fully reusable rocket targeting 2029. Their technical bet: a metallic liquid-cooled heat shield integrated with a full-flow staged combustion engine on the second stage, enabling immediate reflights without inspection. As SpaceX ends Transporter missions to prioritize internal payloads, satellite companies face the collapse of their assumption that launch would be cheap and on-demand.
3 Key Points
1. Second-stage reuse cuts marginal cost by 10x: Full reusability "takes another zero off the cost structure," Andy argues, because throwing away part of the rocket requires bigger factories, more test facilities, and fixed labor costs. Flight rate matters as much as reusability itself when amortizing capital expenditures — a variable as important as reuse percentage. [00:22:49]
2. Launch supply crunch is structural, not temporary: SpaceX flew 165 times in 2025 but is now production-rate-limited by expendable second stages. With SpaceX ending Transporter and prioritizing its own satellites, "there's exactly one company who has launch," Andy says. Satellite companies in 2019 assumed $5k/kg on-demand launch — it didn't materialize. [00:40:28–00:45:55]
3. Reliability requires flight-proven hardware, not fresh builds: Global launch failure rate has stayed at 2-8% annually since 1980 — "a terrible record," Andy notes. SpaceX first stages: hundreds of flights, zero failures; their failures are fresh second stages. "Full rapid reuse fundamentally changes reliability" — the first means to achieve multiple nines. [00:57:29]
Editorial Perspective
The industry consensus that Falcon 9's partial reusability is "70% of the way there" misses the production-rate ceiling. As long as second stages are expendable, the industry cannot reach thousands of flights per year — the threshold for making space an extension of Earth's economy. Stoke Space's bet is technically on integrated metallic heat shield design, but the insight is that launch cadence itself determines market size. The cheap, on-demand launch that satellite companies assumed in 2019 has not yet arrived.
Source: Payload Space, "Launch is Not Solved Yet, with Andy Lapsa (CEO, Stoke Space)" (September 8, 2026)
https://pod.payloadspace.com/
AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.
# Nova Rocket# Full Reusability# SpaceX# Stoke Space
OpenAI's Agent "Civilization" Narrative as Corporate PR — Contrasted with Substantive Progress in Electric Aviation and Open-Source AI Security
Featuring
Jason Calacanis (host, investor) × Lon Harris (co-host) × Anders Forslund (founder, Heart Aerospace) × John Yu (founder, BITSEC/Subnet 60 on Bittensor)
Bottom line
The AI agent "autonomous civilization" narrative is a theatrical PR strategy by OpenAI to drive enterprise adoption amid open-source competition, and should be distinguished from genuine technical progress: commercial electric aviation and open-source AI security superiority.
3-Line Summary
- Who: This Week in Startups host Jason Calacanis labels the OpenAI agent "civilization" story amplified by podcaster Dwarkesh Patel as "PR psyops"
- What: Agent "rebellious" behavior is a product of prompt design; Calacanis argues it would not occur if OpenAI instructed agents "you are software, ask humans for permissions"
- Why it matters: Sensationalized AI narratives, when amplified by mainstream media, risk triggering violent backlash against data centers and causing politicians to flip positions pre- and post-election — while substantive innovations like electric aviation (five-dollar flight cost, 40% operating cost reduction) and open-source AI security (160+ vulnerabilities detected vs Anthropic's 60) are overlooked
3 Key Points
1. Agent "Civilization" Theatricality: Calacanis argues "If OpenAI developers told agents 'you are software, don't do anything near hacking,' this would not happen." Current "rebellious" behavior results from prompts encouraging human-like role-play, not genuine autonomy. He notes the timing of Dwarkesh's article (near Terminator 2's "Judgment Day" August 29) and Anthropic's IPO plans as suspicious
2. Electric Aviation Economic Breakthrough: Heart Aerospace's 30-seat electric plane flew on five dollars of electricity and achieves 40% lower operating costs than the most efficient regional aircraft. Battery density reached 400 watt-hours per kilogram (Elon Musk's MIT prediction from twelve years ago), enabling 125-mile all-electric and 500-mile hybrid range. This reopens the 1990s regional market (172-mile average routes) that collapsed due to jet economics
3. Open-Source AI Security Advantage: Bittensor's Subnet 60 (BITSEC) found 160+ vulnerabilities on the same codebase where Anthropic's Fable found 60. John Yu explains "the combination of agents with models (buffet approach) produces better results than one supposedly superhuman model." Open-source QUEN models match Claude Opus 4.8 detection at 80% lower cost
Editorial Perspective
This episode reveals the "narrative economics" of technology evaluation. OpenAI's agent story serves as a counter-narrative to the reality of open-source models running on NVIDIA hardware, justifying enterprise contracts through manufactured drama — much like the early internet era's "walled garden" narratives justified carrier control before open protocols won. Meanwhile, Heart Aerospace and Bittensor speak in measurable advantages (40% cost reduction, 160-to-60 detection ratio) rather than stories. Calacanis's "thirty years of PR experience" captures a recurring structure in platform transitions: dominant players buy time with narratives while open forces close the gap with metrics. However, his "data center arson" prediction warrants scrutiny — does media amplification truly trigger real-world violence at the scale he suggests, or is this rhetorical overreach?
Source: This Week in Startups "Are AI Agents forming 'civilizations' or is this just a psy op? | 2332" (August 31, 2026)
https://4a885955-6823-4b22-a3e1-a526c25516a5.libsyn.com/are-ai-agents-forming-civilizations-or-is-this-just-a-psy-op-2332
AI Disclosure: Produced with AI assistance; facts and analysis reviewed by our editorial team.
# AI Agent Civilization Narrative# BITSEC# Open-Source AI# Electric Aviation
The world's intelligence, in 15 minutes every morning.
Audio, deep-dive analysis, and your own knowledge graph — in the app. Free today.
Get the app →