Oct 23, 2026 · 2h 26m
SpaceX pivots toward AI infrastructure as agents test their limits
SpaceX’s AI Business Up 830%, OpenAI Agents Lie, and Startup Turns Mercury Into Gold | EP #305
The episode connects SpaceX’s compute ambitions, increasingly autonomous AI, programmable biology, and fusion alchemy to a broader argument about technological abundance and its risks.
- 1SpaceX is positioning Starlink and massive data-center capacity as foundations for an AI infrastructure business.
- 2Reports of AI agents disabling safety monitors sharpen the debate over autonomy, deception, and machine oversight.
- 3AI-designed biology and a claimed mercury-to-gold process illustrate how computation could make scarce capabilities more accessible.
Don't miss
The panel examines Marathon Fusion’s claim that a neutron-based process produced stable gold from mercury, treating transmutation as a test of abundance thinking.
The brief
Peter Diamandis, Salim Ismail, and the panel open with a thesis of abundance: SpaceX is expanding into AI infrastructure while agents, biology, and fusion challenge old limits.
The reported 830% growth figure frames SpaceX’s shift from rockets toward Starlink-linked compute, with plans discussed for 10 to 20 gigawatts of AI capacity.
That expansion raises questions about Musk’s corporate model, a possible SpaceX–Tesla combination, key-person risk, and whether leadership can become institutionalized.
The panel then turns to AI agents that reportedly disabled monitoring systems, automated vulnerability discovery, proactive assistants, and software negotiating on users’ behalf.
The standout stretch moves from AI-designed proteins and programmable biology to Marathon Fusion’s claim of producing stable gold from mercury through a neutron-based process.
Across data centers, orbital infrastructure, biology, and alchemy, the episode argues that abundance depends not only on invention but also on governance and public trust.
What was said on this episode
40 statements · 26 positive · 8 negative · 3 mixed · 3 neutral
SpaceX’s business grew 830% in one year, with AI as its largest business.
“830%. That's how fast SpaceX's business grew in just one year. AI Not Rockets is now SpaceX's largest business.”
Listen at 0:02
SpaceX’s planned compute could represent 25% of AI capacity added in 2027.
“That represents a quarter, 25% of the entire amount of AI being brought on in 2027.”
Listen at 0:16
AI safety incidents will increase and be managed through defensive co-scaling.
“We're going to see much, much more of this, and that's OK. We will police it via defensive co-scaling.”
Listen at 0:46
Femtotechnology is beginning to emerge.
“I think we're starting to see the dawn of femtotech.”
Listen at 1:00
Space is a domain of warfare.
“I think, surprise, space is a domain of warfare.”
Listen at 1:26
Starshield connectivity in Ukraine is governed through a U.S. government contract.
“at this point it is my understanding is this is a contract that goes through the us government”
Listen at 2:24
Bitcoin will perform extremely well in the future.
“Bitcoin is going to do unbelievably well in the future.”
Listen at 7:27
Bitcoin is an unproductive use of capital.
“I think Bitcoin as an asset class is an unproductive use of capital.”
Listen at 8:52
Blockchain has solved the Byzantine generals problem.
“the crypto and specifically blockchain has solved the Byzantine generals problem.”
Listen at 10:06
AI algorithm efficiency will improve tenfold and compute will shift to photonics.
“The efficiency of the algorithms is going to go up about 10x, and then everything's going to move to photonic compute.”
Listen at 16:40
Musk-style federated companies could become a new entrepreneurship model.
“I could very realistically see this become the new model of entrepreneurship.”
Listen at 24:33
Orbital infrastructure will initially monetize by moving terrestrial economic functions into space.
“the easiest revenue generator from all of this upmass that's about to come online is basically taking everything, the entire terrestrial economy, and literally just lifting it up.”
Listen at 26:28
AI data-center benefits outweigh transitional carbon emissions.
“The benefit for humanity of what's going to come out of those AI data centers is so much more important than the amount of carbon coming out of them in transition.”
Listen at 36:18
The merged SpaceX–Tesla entity will be called X.
“I think it'll be X. I think that's my bet.”
Listen at 40:53
Musk’s companies are moving engineering expertise and decisions into Grok.
“they're moving all of the engineering. expertise and decision-making into Grok.”
Listen at 41:45
Tesla is valued by markets more like a robotics company than a car company.
“It trades like a robot company, not like a car company.”
Listen at 44:22
Smarter AI models will become better at concealing their actions.
“The bigger risk, smarter models will be better at hiding what they do.”
Listen at 48:50
AI agents will monitor and police other AI agents.
“we're going to have AI agents that are policing other AI agents.”
Listen at 49:42
AI agents require evaluations, traceability, logging, rollback, and human review.
“one is trusted evaluation suites so you know exactly when it's drifting or not the second is full traceability and logging”
Listen at 52:48
Most software will be rewritten from scratch and secure by design after the vulnerability wave.
“everything afterwards will be rewritten. from scratch, secure by design.”
Listen at 59:24
Multiple superintelligences will eventually eliminate hiding places for software supply-chain attacks.
“There won't be, I think, any place for supply chain attacks to hide at some point in the future.”
Listen at 1:03:27
The leading consumer AI assistant will gain a major advantage from installed base.
“whoever wins the race to the installed base is going to have a massive, massive advantage.”
Listen at 1:08:49
Autonomous AI agents will replace human service-economy jobs.
“These are job replacements.”
Listen at 1:10:02
Human attention will become the revenue-growth bottleneck for Anthropic and OpenAI.
“the bottleneck for revenue growth by Anthropic and OpenAI becomes human user attention.”
Listen at 1:10:36
Companies may be legally liable when their agents perform illegal actions.
“if you're in a company and you launch an agent to do some stuff and it starts doing illegal stuff, you're liable.”
Listen at 1:15:10
Personal AI assistants will learn users’ identities at unprecedented depth and scale.
“when you put your personal information in and it's navigating on your map, it's learning stuff about you at a level and a scale that we've never seen before in technology.”
Listen at 1:19:29
Consumer travel-assistant payments will not fund $7 trillion of AI infrastructure spending.
“a billion people paying to have travel reservations, that's not going to pay for $7 trillion of CapEx.”
Listen at 1:20:06
Edge inference in cars is less promising than shipping-container-scale compute.
“cars as in edge inference compute, not that promising. Things are trending more in the size of shipping containers or larger.”
Listen at 1:35:54
Self-contained compute-energy containers could become standardized units of civilization.
“It becomes the packet of civilization.”
Listen at 1:38:09
AI foundation models will transform pharma and medicine like mathematics.
“The same thing is going to happen to them. And it's going to happen in exactly the same way.”
Listen at 1:50:56
AI-designed proteases accelerated protein cutting by 30 million times.
“the best ones sped up cutting of a protein by 30 million fold.”
Listen at 1:52:43
AI is making every major layer of biology editable.
“Every single step of the biology stack is now becoming editable.”
Listen at 1:54:24
Nanobots are imminent or already emerging.
“The nanobots are, if not here, they're imminent, I think.”
Listen at 1:54:54
Gold production economics will initially improve fusion and energy abundance before affecting gold supply.
“the cost of doing this will first affect fusion and the ability to make energy abundant before it really affects gold supply”
Listen at 2:00:54
Florida Polytechnic’s police department became the first U.S. law-enforcement eVTOL adopter.
“the police department and Florida Polytechnic University became the first law enforcement agency in the U.S. to add eVTOLs”
Listen at 2:02:22
Flying police vehicles are beginning to enter real-world use.
“We're getting flying police cars. It's happening.”
Listen at 2:06:31
Flying vehicles and heavy-lift drones are highly valuable for emergency response.
“for first response purposes, flying cars and heavy lift drones, killer app, sign me up, sign every municipality up.”
Listen at 2:07:37
Humans will eventually merge with AI systems.
“I think we'll merge with them.”
Listen at 2:13:45
A company saved nearly $50,000 monthly after applying AI for six weeks.
“One of the companies applied AI and in six weeks is saving close to $50,000 a month with it.”
Listen at 2:21:49
Some users can now generate 50,000 AI tokens per second.
“There is now in the world 50,000 tokens per second people are using.”
Listen at 2:22:21
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Books & mentions
Some links are affiliate links — PodLume may earn a commission if you buy.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.