What’s Next For SpaceX?

The video recaps SpaceX's flawless IPO debut, which popped about 20% and gave retail investors a healthy gain with time to add shares, before pivoting to the two variables that will define the next year: terrestrial AI compute and Starship progress. Terrestrial AI compute is the near-term revenue driver, monetized through massive deals with Anthropic and Google that include 90-day termination clauses, proving SpaceX holds leverage to pivot compute back to its own models. Starship is the long-term enabler, cutting orbital data center costs roughly in half versus Earth and creating an unassailable moat, since competitors using Starship would face permanent cost disadvantages. The broader thesis positions SpaceX as an AWS-like AI compute platform spanning Earth and orbit, with the speaker calling it a once-in-a-career opportunity and potentially the most iconic company of all time. The segment closes by dismissing anti-Elon narratives as the work of an unreasonable 8–15% minority, arguing they cannot be reasoned with and should be ignored.

• ▶ 0:06 The SpaceX IPO went “flawlessly to a T,” popping about 20% in a smooth debut, with Goldman Sachs and Morgan Stanley credited for controlled execution.
• ▶ 1:02 Retail investors who got IPO allocations at $135 are sitting on a healthy but not outrageous gain, and because the stock didn’t take off immediately, they had a chance to add shares at reasonable levels.
• ▶ 3:34 The guest teases “two variables that are going to matter a lot over the next year” for SpaceX, leaving the specific factors as a cliffhanger.
• ▶ 3:57 Starship progress is a key focal point: it enables 20x more bandwidth per launch for Starlink Gen 3 and is the enabler of orbital AI compute, with Starship reliability being the real bottleneck.
• ▶ 6:31 Gavin's actual first focal point is terrestrial AI compute: scaling faster than anyone else, building excess capacity, and arranging deals with AI companies to create a monopoly on AI infrastructure.
• ▶ 7:11 Terrestrial compute drives short-term revenue (Anthropic ~$1.25B/month and Google ~$920M/month), while Starship progress is the long-term play for Starlink user growth and orbital compute.

• ▶ 7:52 AI compute scaling is the main near-term driver of quarterly financials; better models → more capacity → more users → more revenue, with SpaceX monetizing gigawatts at rates up to 2x average NeoCloud pricing and up to $50B per gigawatt on the Google deal.

• ▶ 9:18 Near-term attention centers on Colossus 3–10 and Cursor: Composer 2.5 is already "Pareto dominant" after RL and SFT on Colossus 2, raising the question of what happens with a bigger base model.

• ▶ 10:27 The host admits his long-term bias missed the immediate AI revenue case, while pointing out that once Starship flies regularly, it could massively boost Starlink revenue and enable orbital data centers.

• ▶ 12:12 SpaceX AI inserted a 90-day termination clause into both the Anthropic and Google compute deals, keeping maximum flexibility to reclaim compute for its own models or renegotiate with higher bidders—proving it holds the leverage.

• ▶ 13:32 Selling external compute is a bridge to monetize spare capacity while SpaceX AI scales, but the 90-day out means it can always pivot back to internal AI priorities when needed.

• ▶ 13:46 The long-term thesis is an AWS-like AI compute platform: SpaceX AI will build as much compute as possible on Earth and in orbit, and because demand for AI compute is near-infinite, its speed and coherence in scaling make it the dominant provider.

• ▶ 15:37 Orbital compute for a gig of capacity costs ~$30 billion versus ~$60 billion on Earth, because Starship's ~$5 billion launch cost replaces the ~$25 billion terrestrial power/cooling bill.
• ▶ 16:28 Starship creates a massive moat: orbital data centers require Starship, and competitors who pay SpaceX to launch would face a permanent, crippling cost disadvantage.
• ▶ 16:53 Terrestrial power/cooling costs feel "reasonably inflationary," which only strengthens the economic case for orbital data centers over time.
• ▶ 17:33 The speaker rejects "worried," reframing it as appreciation for a once-in-a-career opportunity.
• ▶ 17:52 He notes he owned 15% of a company and 10% of Tesla when each was below a $2B market cap.
• ▶ 18:04 He says SpaceX is special and has a relatively high chance of becoming the most iconic company of all time.
• ▶ 18:20 The speaker attributes "trillionaire hate from the left" toward Elon Musk to a reaction against extreme wealth, not his actual actions or achievements.
• ▶ 18:29 He cites Kevin Mahaffey's defense: Musk is treated like a "bond villain who's decarbonizing the world," a paradox given environmental goals are traditionally left-aligned.
• ▶ 18:38 Mahaffey's point is backed by concrete benefits — Starlink connecting low-income countries' schools and hospitals, and Neuralink helping blind people see — leading to the rhetorical question at ▶ 18:51, "That's a villain?"
• ▶ 19:00 A small but vocal minority on the left—not everyone on the left—claims Elon Musk is evil, greedy, and harmful, but these critics are "not serious."
• ▶ 19:16 Certain left-wing politicians deliberately push an anti-Elon narrative they don't personally believe, using it to manipulate "deranged" followers for political points and emotional exploitation.
• ▶ 20:08 The target audience of this narrative is unreasonable and "cannot be reasoned with," so the speaker argues they should simply be ignored—because "You cannot reason with unreasonable people."
• ▶ 20:24 The speaker argues that you "cannot reason with unreasonable people" and that the persistent "Elon bad" narrative will not disappear, prompting a request to Grok for a numerical estimate.
• ▶ 20:44 The speaker shares the full prompt for Grok, asking what percentage of the general population could be described as "[ __ ] in the head" — including traits like mentally ill, psychopathic, Machiavellian, hateful, and violent.
• ▶ 20:59 Grok responds with an estimate that roughly 8 to 15% of the general adult population fits that broad, informal category, which the speaker uses to explain why extreme anti-Elon narratives persist.
• ▶ 21:19 Grok estimates 8–15% of the population is seriously "deranged," and the speaker identifies this group as the source of extreme anti-Musk rhetoric like "Elon's a Nazi" and "somebody should make Elon disappear."
• ▶ 22:09 Grok's breakdown: severe mental illness 5–6%, personality disorders ~9% (including Cluster B), antisocial personality disorder 1–4%, and psychopathy 0.5–1% — the speaker says these, not the broader 23% with mental illness, drive the worst behavior.
• ▶ 22:48 The speaker concludes these people are "not reasonable" and cannot be reasoned with, citing firebombed Tesla dealerships and celebrations of assassination attempts as evidence.
• ▶ 23:18 SpaceX's legacy includes its blue-collar workers prospering significantly through the company's success.
• ▶ 23:27 For opponents of Earth-based data centers due to environmental concerns, SpaceX is presented as the ideal solution, with the argument cut off at "So, you..."
• ▶ 24:19 Musk's challenge to solve world hunger was a public "own goal" for critics when they couldn't explain how to do it despite his offer to contribute.
• ▶ 25:07 Seeing a rocket launch in person is highly recommended and far more visceral and emotional than expected—the sound and blast often make people cry.
• ▶ 26:09 SpaceX is likely to become the greatest company of all time for its civilizational impact, potentially becoming the "British East India Company of the solar system."

• ▶ 0:00 SpaceX IPO and Early Investors (0:00 - 3:47) - - Host and co-host discuss the successful SpaceX IPO, retail investor gains at the $135 price, and what the market will watch next.
• ▶ 3:47 Starship and AI Compute as Key Focal Points (3:47 - 7:54) - - They break down two focal points—Starship progress and terrestrial AI compute—and clarify short-term vs long-term investor focus.
• ▶ 7:54 AI Compute Scaling and Cursor (7:54 - 11:00) - - Predictions about AI compute scaling, Cursor's role as a big variable, and a reflection on the earlier prediction.
• ▶ 11:00 Cloud Business vs. Own Models and AI Compute Services (11:00 - 15:02) - - Tension between the cloud business and SpaceX AI's own applications, plus plans for an AWS-like AI compute service.
• ▶ 15:02 Orbital Data Centers and Starship's Moat (15:02 - 17:28) - - Economics of orbital data centers, Starship's moat for orbital AI compute, and a quick Anthropic/DC update.
• ▶ 17:28 SpaceX's Legacy and Defending Elon Musk (17:28 - 23:34) - - Reflections on SpaceX as a once-in-a-career company, trillionaire hate, and using Grok to analyze unreasonable anti-Musk narratives.
• ▶ 23:34 Social Media, Rocket Launches, and Final Thoughts (23:34 - 27:12) - - Posts on Musk's wealth, a recommendation to see a launch, SpaceX's civilizational impact, and closing remarks.