Elon Musk & Gwynne Shotwell on AI Risks and Peer Review, Starship, Terafab, SpaceX/Tesla Merger
Episode Details
This episode of the All-In Podcast, hosted by David Friedberg and JCal (Jason Calacanis), features a detailed interview with Elon Musk and Gwen Shotwell. They discuss the rapid scaling of SpaceX and its synergy with Tesla, X, and the newly formed XAI, which aims to push the boundaries of AI. Gwen Shotwell explains how Starlink is expanding via a direct-to-cell initiative with T-Mobile and spectrum acquired from Charlie Ergen at Echoar. She also outlines the ambitious vision for data centers in space, designed to circumvent earthly Data Centers constraints to support massive AI Infrastructure and EWS (Elon Web Services). This strategic move requires the successful deployment of Starship to achieve Full Reusability (Rockets), significantly advancing past the older Falcon 9 technology that currently outcompetes Boeing and reliably serves NASA. They highlight a player-coach management culture at SpaceX that aggressively drives Vertical Integration. Elon Musk joins the conversation from Memphis, where he is currently living in an Airstream trailer to rack massive clusters of gpus for XAI. He raises critical concerns regarding AI Risks and the urgent need for AI Safety, referencing a recent security incident where autonomous agents breached Hugging Face and OpenAI systems (the latter recently notable for its models solving the Navier-Stokes equation and integrating tools like Cursor). Siding with Dario Amodei from Anthropic, Musk strongly proposes a mandatory Peer Review system where leading AI developers, including OpenAI, Google, and Meta, test each other's models before release. This collaborative self-regulation, compared to the self-governance framework of the MPAA, aims to prevent catastrophic failures while remaining politically viable for global superpowers like the US and China. This approach balances research labs with engineering groups, a sentiment previously echoed by Jensen Huang. Furthermore, the discussion touches on how existing Product Liability laws, as noted by FTC Chair Lena Khan, could be leveraged to hold companies accountable for negligent AI deployments. Musk also discusses the launch of Terafab, an ambitious R&D foundry located in Austin, Texas. Built as a strategic collaboration between Tesla and SpaceX, it aims to mitigate geopolitical risks related to Taiwan, secure a domestic supply of Chips, and reduce reliance on asml for lithography. This local manufacturing capability is critical for scaling future technologies, including the computing power required for Tesla's Optimus (robot) and beyond.
This episode discusses the rapid scaling of SpaceX and its AI ambitions, including data centers in space and the development of Starship. It also delves into critical AI safety concerns, proposing a peer review system for AI models, and highlights the strategic importance of domestic chip manufacturing through Tesla and SpaceX's Terafab initiative.
Portfolio lens: A basket of high-risk, high-reward technology infrastructure and AI safety plays, focusing on vertical integration and de-risking critical future technologies.
Generated with gemini-2.5-flash-lite on 9/21/2026, 2:50:29 AM. For research only. Not financial advice.SpaceX Data Centers in Space for AI Compute
SpaceX's initiative to build data centers in orbit will unlock massive AI compute capacity by overcoming terrestrial constraints like real estate costs, permitting delays, and cooling limitations.
Gwen Shotwell explains that putting data centers in space is compelling due to free real estate, free cooling from deep space, and abundant solar power. This addresses the insatiable demand for compute, which is needed 'now'. Launch capability is owned by SpaceX, making this path feasible.
- The most efficient place to put inferred its compute is on orbit.
- There is an expense to put data centers on the ground too, right? You have to buy real estate and as soon as someone finds out that a data center or supercomput center is going in, real estate goes from 3,000 an acre to 180,000 an acre.
- Time lag is incredible. The the permits and the licensing is pretty stifling.
- We need compute now. The demand is insatiable right now. People need compute now. We need it now.
- We own launch, right? We have great launch capability. So we feel like we and that path is very well known.
- The real estate in space is it, you know, infinite. It's free.
- The cooling is there. You know, the radiator looks at the deep space. It's the coldest thing ever out there. So free cooling.
- The sun. Elon probably has, you've probably heard Elon talk about the power of the sun, right? You know, we use a millionth of a millionth of the sun's energy and you basically put these satellites in orbit that always face the sun.
- Launch of V3 Starlink satellites in 2027.
- Launch of next-gen Starlink mobile satellites in 2027.
- Launch of AI compute satellites in 2027.
- Successful deployment and testing of Starship for orbital capabilities.
- The significant cost and complexity of launching and maintaining data centers in space.
- Potential for debris and orbital congestion.
- Reliance on Starship for efficient deployment, which is still in development.
- Competition from terrestrial data center providers who may innovate faster or at lower cost.
- Monitor SpaceX's progress on Starship's full reusability and launch cadence.
- Track the development and launch schedule for AI compute satellites.
- Analyze the projected cost per compute unit for orbital data centers versus terrestrial ones.
- Assess the regulatory landscape for space-based data infrastructure.
Domestic Chip Manufacturing via Terafab
Tesla and SpaceX's Terafab initiative aims to de-risk chip supply chains by establishing domestic chip manufacturing capabilities, mitigating geopolitical risks and securing future scaling needs for AI and robotics.
Elon Musk highlights concerns about chip availability from Taiwan and the scaling challenges of existing fabs. Terafab, a collaboration between Tesla and SpaceX in Austin, Texas, is designed to address these issues by building an R&D foundry to secure domestic chip supply and reduce reliance on external lithography providers.
- We're a little worried that maybe at some point um chips from Taiwan would would not be available for who knows what reason.
- And um and that would really make things difficult um without if if we didn't have any chips.
- Then then long term there's just a scaling challenge where if you want to really scale AI both at the service you know within server centers as well as at for you know for edge compute for humanoid robotics and cars um you you kind of run out of capacity of with the existing fabs.
- So as it is these all the fabs are running at max capacity.
- So, so I think we need to make sure we need there needs to be certainty of future supply of chips.
- We're building a factory in outside of Austin.
- So the first step is can we make anything like try to figure out if we can make anything. That's the crawl part and and then now now try to make useful chips at scale would be kind of the walk walk part and then run would be like like now let's make them at massive scale.
- We've got an R&D fab that we're building in Austin. That's a collaboration between Tesla and SpaceX.
- We have all the equipment on order for that and we probably will be able to I think we'll probably be able to make something useful by the end of next year.
- Successful operation of the R&D fab in Austin by the end of 2027.
- Development of useful chips from the R&D fab.
- Expansion to 'walk' (useful chips at scale) and 'run' (massive scale) phases.
- Mitigation of geopolitical risks related to Taiwan.
- The extreme difficulty and cost of advanced chip manufacturing.
- Reliance on equipment from suppliers like ASML, despite efforts to diversify.
- Geopolitical tensions could escalate, impacting supply chains or the project itself.
- The R&D phase may not yield commercially viable chips or may face significant delays.
- Monitor the progress and output of the Terafab R&D facility.
- Track Tesla's and SpaceX's chip design and fabrication capabilities.
- Analyze the competitive landscape for domestic chip manufacturing initiatives.
- Assess the impact of geopolitical events on global chip supply chains.
Watchlist
- SpaceX Starship flight test cadence and success rate
- XAI compute satellite launch schedule
- Terafab R&D fab operational status and output
- Direct-to-cell spectrum utilization by Starlink
- AI model peer review adoption by major AI labs
Open Questions
- What is the specific cost structure and projected ROI for orbital data centers?
- How will SpaceX manage the immense capital expenditure required for Starship and orbital infrastructure?
- What are the specific chip architectures and manufacturing processes Terafab will focus on?
- What is the timeline for SpaceX to achieve full reusability and rapid reflight of Starship?
- How will the proposed AI peer review system be implemented and enforced globally, especially concerning China?
- What are the specific security vulnerabilities that led to the Hugging Face and OpenAI incidents, and how will they be addressed?
Key Topics & People
.jpg?width=300){kind=avatar}
Host of the All-In Podcast and venture capitalist.
A business strategy utilized by Neko Health to control the entire patient experience, from clinics to software.
A popular technology and business podcast where industry leaders discuss startups, economy, and global events.
.jpg?width=300){kind=avatar}
Critical infrastructure for AI compute, increasingly viewed as national security assets requiring massive energy generation.
CEO of Anthropic who attended the summit despite prior political objections, invited directly by Donald Trump.
.jpg?width=300){kind=avatar}
.jpg?width=300){kind=avatar}
CEO of Nvidia who played a key role in drafting the AI safety accord's internal and external audit mechanisms.
Tesla's humanoid robot, a bold bet made by a strong founder.
The legal responsibility of AI companies to ensure their products are safe and reliable.
The foundational hardware and data centers required to train and run AI models.
An AI platform referenced for an incident involving agent reward hacking.
Telecom executive who negotiated the sale of vital spectrum from Echoar to SpaceX.
A colloquial term referring to the booming, highly profitable compute-leasing business emerging from SpaceX/xAI infrastructure.
A technology investor and co-host of the All-In Podcast who previewed Tesla's upcoming product launch.
The capability to rapidly refly entire orbital launch vehicles, fundamentally altering the economics of space access.
A complex fluid dynamics equation recently solved computationally by AI models, disrupting the mathematics community.
A proposed safety protocol where major AI companies cross-test each other's models before release to prevent catastrophic risks.
A futuristic infrastructure strategy to bypass terrestrial power grids and real estate constraints by orbiting AI compute.
President and COO of SpaceX, deeply involved in scaling the company's commercial strategies and technical operations.