SpaceX's $2T IPO, Karpathy Joins Anthropic & the SaaSpocalypse — ft. Dr Dan | EP14
Cursor's $3B revenue run rate may actually cost them $6B to deliver — and Dr. Dan thinks every B2B SaaS database-with-an-interface is now a one-click clone.
God Mode Podcast
SpaceX's $2T IPO, Karpathy Joins Anthropic & the SaaSpocalypse — ft. Dr Dan | EP14
Cursor's $3B revenue run rate may actually cost them $6B to deliver — and Dr. Dan thinks every B2B SaaS database-with-an-interface is now a one-click clone.
TL;DR
Three tech enthusiasts — host Rik, co-host Ben, and guest Dr. Dan (ex-Google, now a vibe coder at Network School near Singapore) — dissect the week's biggest AI and tech stories. SpaceX's $1.75T IPO leads the conversation, followed by Andrej Karpathy's surprise move to Anthropic, the brutal Cursor vs. Codex vs. Claude Code debate, and why B2B SaaS margins are collapsing. The episode closes with a surprisingly deep breakdown of GLP-1 peptides, Ozempic, and retatrutide. Key takeaway: in the AI era, distribution and proprietary data are the only real moats left.
SpaceX filed the biggest IPO in history at a $1.75T valuation. Andrej Karpathy left OpenAI for Anthropic. The AI coding wars heated up with Cursor vs Codex vs Claude Code debates. Plus a full breakdown of GLP-1 peptides, Ozempic, and retatrutide.
- GLP-1
- Glucagon-like peptide-1, a messenger molecule produced in the gut that signals fullness to the brain and moderates insulin sensitivity; the target receptor for Ozempic and related weight-loss drugs.
- Tirzepatide
- A dual GLP-1/GIP receptor agonist sold as Mounjaro/Zepbound; the second-generation obesity drug after semaglutide (Ozempic), offering stronger weight-loss effects.
- Retatrutide
- A triple agonist targeting GLP-1, GIP, and glucagon receptors; the third-generation obesity drug showing ~28% body-weight loss in Phase 3 trials.
- Semaglutide
- The active ingredient in Ozempic/Wegovy; a GLP-1 receptor agonist engineered to have a long half-life so it only needs weekly injection.
- GIP
- Glucose-dependent insulinotropic polypeptide, a gastrointestinal hormone that works alongside GLP-1 to regulate insulin; targeted by tirzepatide and retatrutide.
- Compounding pharmacy
- A US pharmacy legally permitted to prepare custom drug formulations for individual patients; exploited as a loophole to sell cheaper, unbranded versions of Ozempic.
- Loss leader
- A product sold below cost to attract customers, with the intent of making profit on future purchases; described here as the pricing strategy for AI coding subscriptions.
- Moat
- A durable competitive advantage that protects a business from rivals; used repeatedly to assess whether Cursor, Anthropic, and SpaceX have defensible positions.
- SaaSPocalypse
- Coined term for the predicted collapse of B2B SaaS margins as AI enables businesses to clone or self-build software instead of purchasing subscriptions.
- Vibe coding
- AI-assisted software development where a user describes desired functionality in natural language and lets an AI agent generate and iterate on the code with minimal manual coding.
- Agentic
- Describes AI systems that autonomously plan and execute multi-step tasks without continuous human input; Meta and others are restructuring teams around agentic AI.
- Responses API
- OpenAI's stateful API that retains conversation history server-side so callers only need to send new turns, used as a lock-in mechanism.
- Prompt caching
- A technique where repeated portions of a prompt (e.g., system instructions) are stored and reused to reduce token processing costs on subsequent API calls.
- BPC-157
- Body Protection Compound 157, a synthetic peptide claimed to accelerate tissue repair; popular in biohacking circles but largely studied only in animal models.
- TB-500
- Thymosin Beta-4 fragment, a synthetic peptide used alongside BPC-157 for purported tissue regeneration; limited peer-reviewed human evidence.
- Half-life
- The time it takes for a substance's concentration in the blood to fall by half; natural GLP-1 has a half-life of about 30 minutes, driving the engineering of longer-lasting analogs.
- CapEx
- Capital expenditure — large upfront investment in physical assets; used here to describe the $2.3B average cost of bringing a drug to market.
- Solopreneur
- A solo entrepreneur who runs a business independently without employees, leveraging tools like AI to replace traditional team functions.
- Bioavailability
- The proportion of a drug that enters circulation and reaches its site of action; oral ingestion of peptides has lower bioavailability than direct injection.
- Gut motility
- The speed at which food moves through the digestive tract; GLP-1 agonists slow gut motility, contributing to feelings of fullness and reduced appetite.
Chapter 1 · 00:00
Intro & Guest Introduction: Dr. Dan at Network School
It's incredibly reckless in this world to not learn AI.
Chapter 2 · 02:50
SpaceX IPO: $1.75 Trillion Valuation Breakdown
Is the $1.75T SpaceX IPO the deal of the century — or a hype bubble with a lock-up cliff?
SpaceX is targeting $1.75 trillion at IPO — the largest in history. Dan calls it irrational hype driven by a master storyteller. Ben sees a monopoly in space launch, GPU-in-orbit economics, and a coming Taiwan-to-Texas chip migration. Rik bets retail FOMO sends it to $3–5T before the lock-up cliff.
SpaceX filed for its IPO targeting a $1.75 trillion valuation, raising $75 billion — making it the largest IPO in history.
Chapter 3 · 08:30
SpaceX Acquiring Cursor: The Coding Wars Strategic Play
Cursor's revenue might be $3B — but it probably costs them $6B to earn it.
Cursor fine-tunes Chinese models, has no frontier AI, and its IDE is just a VS Code fork. The only real assets are user data and brand awareness — and Elon is burning SpaceX money to buy both. Dan gives seven to eight times more value per dollar from Codex than Claude Code.
Chapter 4 · 12:40
AI Coding Tools: Cursor vs. Codex vs. Claude Code
He went to sleep. Codex built a full ERP system. He woke up to working software.
Dan gave Codex a detailed plan for a full ERP system and went to sleep. Seven hours and 25 minutes later, it was done — end-to-end tested, click-through verified. Claude Code can't do this. We are entering the era of fast fashion software.
Dr. Dan used OpenAI Codex to autonomously implement a full ERP system for a friend; it ran for 7 hours and 25 minutes overnight and finished end-to-end tested.
An entrepreneur Dr. Dan met at Network School fired 18 people — 15 customer support staff replaced by AI, and remaining developers replaced by Claude Code.
One entrepreneur attended a $34,000/year B2B SaaS demo, transcribed the call, fed it to Claude Code, and had a working clone by morning. The product used the same public APIs. This is now the template — your sales pitch is your product spec.
An entrepreneur cloned a $34,000/year B2B SaaS product by transcribing the sales call and feeding it to Claude Code, which rebuilt the product using the same APIs.
Chapter 5 · 20:30
OpenAI's $2M Credits for YC Startups: Lock-In or Generosity?
OpenAI isn't giving away $400M in tokens out of generosity — it's buying your startup's future.
OpenAI is giving $2M in credits to every YC startup — roughly $400M total. The real play: startups use those tokens to onboard users, burning through the credits and locking into OpenAI's ecosystem. When the credits run out, they keep paying. Sam gets a hand in every promising YC company.
OpenAI offered $2 million in credits to every YC startup; with ~200 companies in the batch, that's roughly $400 million in tokens given away.
Chapter 6 · 28:00
Andrej Karpathy Joins Anthropic: What It Means
Dan thought Anthropic was all hype — until Karpathy walked in the door.
Dan had dismissed Anthropic as marketing over substance. Then Andrej Karpathy — an OpenAI founding member — joined as an R&D researcher. Dan's take: Karpathy must know something the rest of us don't. One hire flipped the entire bull case.
Chapter 7 · 32:40
The SaaSPocalypse: B2B SaaS Is a One-Click Clone
Your $20/month AI subscription is losing money on purpose. Here's why.
Microsoft canceled its internal Claude Code licenses after token-based billing made the cost untenable, signaling the economics of frontier model subscriptions are breaking.
Coding subscriptions from Anthropic and OpenAI are deliberately priced below cost to capture market share. APIs carry the actual margins. Cursor is doing the same thing — just burning SpaceX's money instead of Microsoft's.
AI lab subscriptions are deliberately priced below cost (loss leaders) to gain market share, while API pricing carries actual margins.
Cursor reportedly hit a $3 billion annual revenue run rate ahead of the SpaceX acquisition deal, though Dan suggested it likely costs $6 billion to generate that revenue.
Chapter 8 · 37:40
Meta Layoffs & the AI-First Era
If your SaaS is a database with an interface, someone will clone it tonight.
B2B SaaS built on a glorified database is now a one-click clone. The industries with real staying power are those with physical assets, proprietary distribution, or deeply entrenched systems of record. Every other SaaS product will be commoditized like a cheap IKEA chair.
Chapter 9 · 46:00
China Tech Deep Dive: Shenzhen, EVs & Factory Tours
Her side hustle is a factory with 40 employees. China is on a different level.
A university lecturer Dan met in China casually mentioned her two side hustles — one of which is a factory with 40 employees exporting biological lacquer coatings. In China, 'side hustle' means real physical production. It completely reframes the Western obsession with digital micro-businesses.
In Shenzhen, Dan visited a company that manufactures factories — selling robotic arms and production lines to BYD, CATL, and global brands like KUKA. The twist: they OEM everything and forbid photos, because their customers don't want anyone to know who actually builds their equipment.
Chapter 10 · 51:50
GLP-1 Peptides Explained: From Ozempic to Retatrutide
Ozempic is just Charmander. Retatrutide is Charizard — and there's more coming.
Retatrutide Phase 3 trials showed 28.3% body weight loss on 12mg over 80 weeks, significantly outperforming earlier GLP-1 drugs.
Ozempic triggered a revolution by cutting injections from twice daily to once weekly. Retatrutide adds a third receptor key — GLP-1, GIP, and glucagon — delivering 28.3% body weight loss in Phase 3 trials. Think of each generation as a Pokemon evolution, and we're not close to the final form.
Natural GLP-1 has a half-life of only 30 minutes in the blood; drug engineers have extended this to enable weekly injections with sustained effect.
Ozempic unlocked mass market adoption by reducing injections from twice daily (liraglutide) to once weekly while doubling the therapeutic effect.
Taking a drug from early research to market approval costs an average of $2.3 billion and takes a median of 15 years.
Brand-name Ozempic costs around $400 per month in the US, while equivalent peptides sourced from China cost $10–20, a 20-40x price gap.
Chapter 11 · 1:00:30
Pharma Pricing, Compounding Pharmacies & Biohacking Culture
US pharmacies sell Ozempic for $50 using a loophole. Here's how.
Developing a drug costs $2.3 billion over 15 years. You get 8 years to recoup it before generics hit. Meanwhile, US compounding pharmacies exploit a loophole to sell unbranded Ozempic for $50 instead of $1,500. It's the AI subscription pricing model applied to pharmaceuticals.
Peptides like BPC-157 and TB-500 are popular at biohacking retreats, but nearly all the evidence comes from animal studies. Less than 1% of mouse findings transfer to humans. Most peptide users are running an uncontrolled personal experiment with anecdotal backing.
Dr. Dan cited that a very small percentage — potentially less than 1% — of positive results in mouse studies actually transfer to humans.
No indexed bits in this chapter.
Show stoppers
B2B SaaS built on a glorified database is now a one-click clone. The industries with real staying power are those with physical assets, proprietary distribution, or deeply entrenched systems of record. Every other SaaS product will be commoditized like a cheap IKEA chair.
Dan gave Codex a detailed plan for a full ERP system and went to sleep. Seven hours and 25 minutes later, it was done — end-to-end tested, click-through verified. Claude Code can't do this. We are entering the era of fast fashion software.
Snapshots ()
Key Quotes ()
This episode
Claims & Sources
Factual claims made this episode, and whether a source was named.
SpaceX is raising $75 billion at a $1.75 trillion valuation, targeting an IPO date of June 12, making it the largest IPO in history.
Cursor fine-tunes Chinese models rather than building its own frontier AI models.
Claude Code forcibly adds itself as a collaborator on GitHub commits even when explicitly told not to.
OpenAI's $2 million credits to each YC startup in the current batch amounts to roughly $400 million in total tokens given away across ~200 companies.
AI lab coding subscriptions are deliberately priced below cost as loss leaders to gain market share, while API pricing carries actual margins.
Microsoft canceled its internal Claude Code licenses after token-based billing made the cost untenable.
Taking a drug from early research to market approval costs an average of $2.3 billion and takes a median of 15 years.
Retatrutide Phase 3 trials showed 28.3% body weight loss on 12mg over 80 weeks.
Natural GLP-1 has a half-life of approximately 30 minutes in the blood.
Ozempic (semaglutide) requires only one injection per week versus two injections per day for its predecessor liraglutide, while delivering double the therapeutic effect.
Less than 1% of therapeutic findings proven in mouse studies successfully transfer to humans.
Cursor is at a $3 billion annual revenue run rate ahead of the SpaceX acquisition deal.
Anthropic has raised approximately $60–80 billion to date, while OpenAI's last funding round alone was $100 billion.
Codex ran autonomously for 7 hours and 25 minutes to build and fully test an ERP system based on a detailed plan document.
SpaceX is reportedly planning to acquire Cursor approximately 30 days after its IPO for $60 billion, with a $10 billion walk-away penalty.
This episode
Cast
-
Described as a master of hype; behind SpaceX's IPO, the Cursor acquisition deal, and providing compute to Anthropic.
-
OpenAI founding member who joined Anthropic for R&D; seen as the single most bullish signal for Anthropic's credibility.
-
Tech entrepreneur whose 'network state' concept inspired Network School, where Dr. Dan is currently living.
-
Offering $2M in credits to each YC startup; targeting an IPO in September; developing Codex as its agentic coding tool.
-
Filed the largest IPO in history at a $1.75T valuation; reportedly planning to acquire Cursor and provide compute to Anthropic.
-
AI lab reaching profitability; Andrej Karpathy joined for R&D; Dr. Dan calls it mostly hype and marketing but says Karpathy's hire is bullish.
-
Balaji Srinivasan's co-living and learning community near Singapore in Malaysia, where Dr. Dan is currently based.
-
Startup accelerator whose current batch of ~200 companies is being offered $2M in OpenAI credits each.
-
Track
Conducting AI-first restructuring layoffs; employees received termination emails at 4AM after being told to work from home.
-
Track
Canceled its internal Claude Code licenses after token-based billing became too expensive; cited as an example of SaaS cost pressure.
-
Chinese electric vehicle company that buys automated factory equipment from an unnamed Shenzhen manufacturer Dr. Dan visited.
-
Track
Pharmaceutical company developing an oral semaglutide pill formulation as a future alternative to injections.
-
Track
Taiwan Semiconductor Manufacturing Company; Ben cited speculation that Elon Musk would poach TSMC engineers to build a chip facility in Texas.
-
AI coding IDE reportedly being acquired by SpaceX for $60B; Dr. Dan argues it has no real moat and fine-tunes Chinese models.
-
OpenAI's agentic coding tool; Dr. Dan's preferred tool, claiming it delivers 7-8x more value than Claude Code per dollar.
-
Anthropic's agentic coding tool; Dr. Dan prefers Codex overall but acknowledges Claude Code is better at debugging multi-agent tasks.
-
GLP-1 weight-loss drug (semaglutide) that unlocked the mass market for obesity treatment; costs ~$400/month in the US vs $10-20 from China.
-
Third-generation triple GLP-1/GIP/glucagon agonist showing 28.3% body weight loss in Phase 3 trials over 80 weeks.
Stats
