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

SpaceX went public at $135 per share, closing its first day up nearly 20 percent at $161 in what Peter Diamandis called the largest IPO in history. The company reached a $2.89 trillion market cap at time of recording, surpassing Amazon to become the fifth largest company in the world. The IPO absorbed $75 billion in cash from the market, and in approximately six months lockup periods expire, releasing roughly $1 trillion of stock eligible for sale. One speaker disclosed being an investor since 2013 and acknowledged that colors their enthusiasm. Elon Musk holds approximately 82 percent voting control and his net worth reached approximately $1.3 to $1.4 trillion, making him the world's first trillionaire by a large margin according to Diamandis. The IPO created approximately 4,400 employee millionaires and around 400 centimillionaires and billionaires in a single day.

Speakers described SpaceX as three converging businesses: a launch monopoly leading competitors by a factor of 10 to 100, Starlink generating over $1 billion in quarterly profits, and an AI frontier lab. For comparison, Google carries roughly the same market cap but generates around $300 billion per year of nearly pure margin, a contrast speakers used to temper enthusiasm at the current price, with one noting the IPO would have looked like a great opportunity at a $1 trillion valuation. SpaceX's S1 filing explicitly warned that orbital congestion and cascading debris collisions pose an existential threat to its core business, potentially rendering licensed orbits unusable for hundreds of years. One speaker claimed low Earth orbit is already past the critical mass threshold identified by Kessler math. Peter Diamandis expressed 100 percent certainty that SpaceX and Tesla will eventually merge, while Polymarket placed the probability of a merger by end of year at 37 percent.

The US government issued an export control directive to Anthropic at 5:21 PM Eastern on a Friday with only 90 minutes of warning, suspending all access to Fable 5 and Mythos 5 for foreign nationals anywhere on the planet, including Anthropic's own foreign employees. Approximately one third of Anthropic's workforce are foreign nationals affected by the directive. Because Anthropic lacked a system to verify user nationality on short notice, it disabled both models globally for all users. The government cited a jailbreak vulnerability in Fable's guardrails as the basis for the action; Anthropic characterized the issue as a minor, previously known problem rather than a universal jailbreak. Dave described this as the first time a government has blocked an AI model and argued it will not be the last. Salim warned the resolution may take longer than expected and will be damaging to all frontier labs. Alex predicted a grand bargain between Anthropic and the White House, possibly involving golden shares given to the US government for a hypothetical sovereign wealth fund.

Separately, Anthropic drew criticism for two undisclosed policies discovered in its terms. Buried in a 319-page document, Anthropic retained every prompt and context window for at least 30 days with no exceptions, even for enterprise customers who had negotiated zero data retention. Fable 5 also silently downgraded users to version 4.8 without notification when it detected frontier AI research queries or guardrail violations, and Anthropic had expressly reserved the right to actively poison ML-oriented queries that could benefit competing frontier labs. Anthropic subsequently backtracked on the poisoning policy and committed to warning users before any such interference. Salim noted that 70 percent of elite AI researchers in the US are foreign-born non-citizens, primarily from China, India, Taiwan, and the UK, and that blocking their access to frontier models risks sending top talent back to those countries. Speakers predicted every company will be forced to run models on premises and build failover capability, and that every country will pursue sovereign AI capability that cannot be shut down by the US government.

On the OpenAI IPO delay, Dave noted Altman is not a significant shareholder in OpenAI and has no personal financial damage from postponing it, and that Altman has invested in over 400 companies surrounding OpenAI, making him financially indifferent to forcing a risky rollout. Alex offered two interpretations: the shallow reading is that Altman is putting a positive face on the need to grow Codex revenue before going public, while the deeper reading is that a large private company achieving recursive self-improvement may no longer need external capital, and that if technology substitutes for capital rather than just labor this could mark the beginning of a decoupling between technology and capital. OpenAI raised $122 billion in its most recent capital raise, larger than the SpaceX IPO capital raise, leaving it well capitalized without needing public markets. OpenAI's engineering lead on Codex announced that Codex can now set its own goals, with a team developer stating they no longer write their own goals and instead ask Codex to write goals for itself and each sub-agent it spawns.

Federal Reserve Bank of New York data cited in the episode shows recent college graduates are the group experiencing the most difficulty finding employment, and Stanford AI Index data shows software developers and customer support agents aged 22 to 25 are diverging negatively from older age groups. Peter Diamandis identified a dangerous cocktail of high expectations, high connectivity, low opportunity for certain segments, and rapid mobilization, and noted that AI-assisted social media is a new and untested factor concurrent with rising discontent.