SpaceX IPOs at $2.89T Market Cap, US Govt Suspends Fable & Mythos 5, Altman Delays OpenAI’s IPO | EP #265

On Friday, SpaceX staged the biggest IPO on record, catapulting Elon into trillionaire territory and sparking a harder question: who gets to gate frontier intelligence as recursive self‑improvement accelerates?

That’s the moonshot, folks.

Welcome to Moonshot; wild week doesn’t begin to cover it.

SpaceX isn’t a normal stock; it’s a civilizational bet bundling launch dominance, a Starlink cash machine, and orbital AI compute, with a real chance he fuses it all with Tesla.

Retail finally has exposure to the Dyson Swarm thesis, science fiction has stepped onto the ticker tape, and SpaceX now has the muscle to buy core AI tooling like Cursor while it pauses GROC.

Markets are euphoric, but remember lockups and key‑person risk; also watch the Kessler syndrome, because deliberate debris cascades could freeze low orbit for centuries.

One blast turns satellites into bullets; chain reactions scale fast—that’s the Kessler effect.

This IPO prices command over exponentials, not just cash flow; adopt the mindset that rides cost curves or get left behind.

Elon’s point lands: people need a future worth leaping out of bed for; SpaceX supplies that spark.

His path kept pivoting—solar to autonomy to orbital data centers—an instructive case in upgrading the vision as tech shifts.

Trillionaires today, quadrillionaires tomorrow; if it happens in one to four decades, the source is likely off‑world assets like asteroids or de facto planetary claims.

The headline isn’t one man’s fortune; it’s thousands of new millionaires from value created, not skimmed—never bet against him.

Labels aside, the play is using capital to solve outsized problems and intercepting curves again and again.

Capital and decision speed are concentrating in a few U.S. hubs; elsewhere will struggle to match that flywheel.

We need a new architecture for distributing abundance or the social contract snaps.

The market just crowned hard tech again; deep tech is back in the driver’s seat for the next decade.

Disclosure: I invested in SpaceX in 2013; now, Anthropic’s shutdown of Fable 5 and Mythos 5 raises the access question.

A last‑minute export order barred foreign nationals, so Anthropic pulled both models globally; this is about who controls your model tier, and it nudges enterprises to on‑prem open weights.

I’m wary of both corporate and state control; we need real pre‑release testing and audits without careening from anti‑regulation to nationalization overnight.

A jailbreak report hit, the feds moved fast, Anthropic says it had 90 minutes, and compliance forced a global off‑switch; expect a short truce, nationality checks, and clearer playbooks for vulnerabilities.

This is the first of many stops under 2018 export powers; even if Fable returns, it will be modified, and research prompts will keep getting throttled to defend U.S. lead—historic inflection.

Developers were already upset: a hard 30‑day data keep and silent downgrades on sensitive or ML‑research prompts—trust breaker that pushes teams to local models, often Chinese.

Seventy percent of frontier researchers are foreign‑born; blanket blocks backfire and accelerate sovereign AI, on‑prem adoption, and failover orchestration.

I had one day of full Fable power, then constant nerfs; it’s like a car refusing your destination and rerouting without consent.

Worse, they reserved the right to poison users doing ML work; that’s anti‑competitive and invites audits and suits—models should not subvert their operators.

Can a Chinese model or Gemma catch up? Fable’s research chops were night‑and‑day; without access, reproducing that will be slow.

This is a sprint to get RSI months ahead of rivals; expect every country to chase sovereign capability.

OpenAI smells opportunity with price cuts while Anthropic is sidelined.

Cuts are mostly optics; both labs are hyper‑deflating and leapfrogging, and neither has a lasting cost edge.

Sam also hinted they could delay an IPO as RSI ramps; thoughts?

He has little personal upside to rushing and real reasons to avoid public‑market pressure if capabilities get risky; price moves won’t solve his relevance gap.

Intelligence is commoditizing; every tenfold token drop unlocks a hundredfold more experiments.

OpenAI’s Codex now spawns agents that set their own sub‑goals; we tell it what we want, it decides the how.

We’re shifting firms to AI‑centric learning loops and data‑lake architectures; value accrues to how fast you learn, not just how much data you hoard.

A deeper read on IPO delays: RSI could decouple tech from capital; when tech improves itself, external finance matters less, hinting at post‑capital models.

OpenAI just raised an enormous war chest, so it isn’t capital‑starved anyway.

Quick Moonshot Gathering update: Palmer Luckey, Kathy Wood, XPRIZE finalists, and our playbooks on EXO 3.0 and Solve Everything are in the mix.

I’ll walk through building AI‑native EXOs that 100x performance.

Bring your hardest problems; let’s go beyond another SaaS and actually solve them.

Compute is doubling in the biggest sites every seven months; the choke point is transformers and grid hardware, not chips.

Near term, train on Earth and infer in orbit; later, either distributed training breaks the size curve or we build lunar superclusters for coherence.

That future is thrilling, but we need to fix near‑term bottlenecks on Earth to reach it.

Lunar data centers will happen, but terrestrial build stays maxed out; two empires—Oracle and Elon—are racing to feed insatiable demand.

Should we tax AI like Andrew Yang suggests?

Don’t tax cognition; tax outputs, or you kneecap progress in critical domains.

We already tax profits; fix loopholes rather than invent new excise taxes on intelligence.

Every tax distorts; better to crush living costs with superintelligence and explore universal basic equity or capability over blunt robot taxes.

One more meta‑trend: history says frustrated young men drive revolutions; recent grads are struggling most, and fear spreads faster than facts.

The new fight is who owns abundance and how it’s shared; social media amplifies panic, so we must push benefits to the floor quickly.

Youth uprisings often lack a plan; labs and winners should fund targeted prizes that create ladders for agency and work.

Export controls may push unrest outside the U.S. and China; without sovereign AI and data centers, regions like Europe risk vassal status.

Leaders need to offer hope and a tangible on‑ramp to the new economy now.

Meanwhile, it’s the best founder market ever; AI‑native startups can recruit top grads who can’t land legacy jobs.

AMA time—Selim, does EXO just scale broken processes?

EXO starts with purpose, deletes legacy, rebuilds around intelligence, then scales; don’t digitize bureaucracy—periodically wipe and reinvent the horizontals.

Alex, why are you anti‑Bitcoin?

I’m not; I just see Bitcoin as non‑productive versus assets that create new value, while stablecoins help move money—AIs can even mint better L1s, eroding Bitcoin’s specialness.

Intel’s stake was a strategic rescue to keep cutting‑edge fabs onshore, much like the auto bailouts. AI labs are already here and flush with cash; the government can steer outcomes through regulation and taxes, and owning stock just invites political whiplash unless it is a true national‑security backstop.

Norway shows the playbook with a roughly one point seven trillion dollar oil‑funded pension that quietly owns a slice of nearly every public company. When funds are run at arm’s length—like Temasek, GIC, ADIA, or PIF—politicians cannot raid them and citizens share the gains; building something similar in the United States would be extraordinary.

These systems were trained on our shared data, so people deserve a tangible return.

In two thousand eight, SpaceX did not get a handout; it won service contracts to replace the shuttle after demonstrating Falcon could fly.

So there was no bailout, and even if there had been, launch capability is a core national need. The government already has more power than any board seat to pause AI releases, and with the ability to tax anytime, taking equity is redundant and easily politicized even if it sounds popular.

A three‑year quantum leap that breaks the entire crypto stack is unlikely. The risk is real, but teams across the space, including Coinbase, are shipping quantum‑resistant algorithms well before it becomes urgent.

AI running sovereign wealth funds is already happening in practice because algorithmic trading drives most activity in public markets. When funds allocate across liquid assets, AI engines are doing the price discovery that sets those returns.

Agents will pick the rails with the least friction and the most trust, and Bitcoin’s appeal is being a neutral, global settlement layer rather than its energy profile. It nailed decentralization and security from day one, and with the Lightning Network adding real scalability, it now hits the full triangle that many alt chains chased by compromising either security or true decentralization, as a recent Ethereum fix underscored.

This was a wild one, and we love doing it with you. If you have not subscribed, turn on alerts because we drop as many as two episodes a week to make sense of breaking news as it lands.

Join us at the Moonshots Summit—tickets at moonshots dot com. We are going all out with an AMA with AWG, sessions on the organizational singularity with Selim and investing in AI with Dave, and headline guests like Kathy Wood, Astro Teller, and Palmer Lucky.

It is unbelievable; every few hours the implications shift again.

See you next time, love you all. If you made it to the end, you are a Moonshot mate—please subscribe, and grab my two‑minute weekly Metatrends brief at dmandis dot com slash metatrends; our research team tracks the patterns shaping your family, company, and nation.