← All episodes

SpaceX's Financials Leaked: Is it Worth $2TN | Meta Debuts Muse Spark: Are They Back in the AI Race?

| 29 products mentioned
20VC with Harry Stebbings 20VC with Harry Stebbings host
Jason Lemkin guest
Rory O'Driscoll guest
Meta Debuts Muse Spark guest
Are They Back guest
AI Race guest
Watch on YouTube ai adoption saas valuations enterprise software anthropic openai meta spacex

Stebbings and guests dissect the AI landscape's most consequential developments: Anthropic's Mythos model revealing massive cybersecurity vulnerabilities, the impending irrelevance of SaaS incumbents building 60% solutions, and Meta's pivot toward ownership of frontier AI models. The episode cuts through hype and forced positivity to deliver a clear thesis: builders must choose between becoming AI-native or becoming IBM—a slow-growth cash machine with no margin for error.

Key takeaways
  • • SaaS companies building "60% solutions" cannot charge for AI features and are trapped in a death spiral; only products genuinely better than standalone competitors (like Claude or Cursor) will command growth multiples and survive the transition.
  • • The enterprise AI market will be 2/3 of the total opportunity, not consumer; companies investing in the wrong channel (consumer ads for OpenAI vs. enterprise agents for Anthropic) are misallocating capital and will face valuation compression.
  • • Token budgeting by CIOs shifts the game from developer-led adoption to centralized procurement, favoring brands with traditional enterprise sales motions—meaning OpenAI's Microsoft relationship is now critical, not optional.
  • • Building AI agents will become as simple as prompting within 18 months, making the bar for "mediocre employees vs. mediocre agents" a literal hiring decision; companies must improve revenue per employee year-over-year or admit they're not adapting.
  • • SpaceX's $2 trillion IPO valuation requires 100% probability of success on speculative businesses (direct-to-cellular, space data centers) with zero discount rate—a bet few conservative investors should make without applying realistic probability and time-value assumptions.
  • • Mythos's ability to autonomously discover zero-days at scale ($20K compute cost) triggers an arms race where every website becomes instantly hackable; incumbents must now treat AI-powered vulnerability discovery as table stakes and rebuild entire security postures.

Recommendations (1)

Replit
Replit uses

"I built a for fun yesterday on Replit, I built a fully functional website in about six minutes that has video, audio, and everything"

Jason Lemkin · ▶ 1:09:54

Mentioned (28)

Anthropic Mythos "Anthropic unveils mythos, but withholds it from public release because it's too good at hacking" ▶ 1:21
Meta Muse Spark "Meta debuts Muse Spark. It's Alex Wang's first model from Meta's Super Intelligence Labs" ▶ 0:24
OpenAI
OpenAI "I'm pretty bullish actually on OpenAI in the enterprise" ▶ 0:35
Anthropic
Anthropic "Anthropic has the advantage of clarity and focus. OpenAI has the advantage of the consumer business" ▶ 0:38
Supabase
Supabase "It's an issue Supabase and others had to deal with" ▶ 4:52
Claude
Claude "the whole mythos run they said I think Claude said it took him $20,000 of credits" ▶ 5:31
ChatGPT
ChatGPT "the unwillingness to release ChatGPT-2 which in retrospect was overdone" ▶ 9:51
Joe Rogan
Joe Rogan "I heard you on Joe Rogan" ▶ 9:10
Airbnb
Airbnb "Remember the Airbnb was the sharing economy and Uber was the you remember the sharing economy" ▶ 12:26
Uber
Uber "Remember the Airbnb was the sharing economy and Uber was the you remember the sharing economy" ▶ 12:28
20VC "if you haven't heard the message that you got to like you got to catch up in AI, may maybe I'm no..." ▶ 15:56
Trainium "Mythos was trained entirely on Amazon's Trainium chips" ▶ 17:47
Nvidia
Nvidia "are we slightly seeing a loosening of Nvidia's stronghold on the market?" ▶ 18:25
AWS
AWS "Amazon instead of buying Nvidia chips is buying its own chips and then offering cloud hosting ser..." ▶ 18:49
Cursor
Cursor "Claude Code is clearly directly competitive with Cursor" ▶ 22:30
Replit
Replit "as the lovable Replit guru here" ▶ 22:27
Bolt
Bolt "Eric from Bolt on these screenshots. Eric from Bolt said, 'Oh, we all I was with I was at a dinne..." ▶ 22:47
Figma
Figma "those product teams could just use Figma make, right?" ▶ 24:59
Salesforce
Salesforce "Cursor does 4 million per employee, Salesforce does 700,000 per employee. Salesforce has 30% oper..." ▶ 1:11:26
Applovven
Applovven "AppLovin is an amazing business because they actually have fairly high gross margins and low empl..." ▶ 1:11:53
Trade Desk
Trade Desk "Given Trade Desk's downturn, it's the most successful ad network business by far" ▶ 1:12:10
Thoma Bravo
Thoma Bravo "Thoma Bravo shutting down the growth equity business and is this foreshadowing of a load of other..." ▶ 1:14:00
Coupa
Coupa "Challenge businesses like Coupa, like Anaplan, like Medallia" ▶ 1:16:29
Anaplan
Anaplan "Challenge businesses like Coupa, like Anaplan, like Medallia" ▶ 1:16:29
SpaceX "It appears that SpaceX has already filed and they're on track" ▶ 1:23:57

More from these creators

The Early Days of Anthropic & How 21 of 22 VCs Rejected It | The Four Bottlenecks in AI | Anj Midha

20VC with Harry Stebbings · Apr 14, 2026 · 3 recs

ElevenLabs: Building an AI Sales Machine & Why We Set a 20x Sales Quota

20VC with Harry Stebbings · Apr 11, 2026 · 9 recs

OpenAI Buys TBPN & Their Management Team Reboot | Mercor Hack & Why Now is the Time for Cyber

20VC with Harry Stebbings · Apr 09, 2026 · 4 recs

Demis Hassabis: Why AGI is Bigger than the Industrial Revolution & Where Are The Bottlenecks in AI

20VC with Harry Stebbings · Apr 07, 2026 · 1 rec

How Hims & Hers Reached a $4.3BN Market Cap on $2.3BN of Revenue | Andrew Dudum

20VC with Harry Stebbings · Apr 04, 2026 · 4 recs

OpenAI Kills Sora & Hits $100M ARR on Ads | Oura Going Public & Whoop Raises at $10BN

20VC with Harry Stebbings · Apr 02, 2026 · 2 recs