MurmurCast
DiscussionNews

20VC: SpaceX Buys Cursor for $60BN | Stripe's $8BN OpenRouter Bet | Anthropic's First Profit & The Math Behind Reaching $600BN in Revenue? | Lovable and Higgsfield Raise Mega Rounds

The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch1h 13m

This week's 20VC episode discusses major M&A deals in tech: SpaceX's $60B acquisition of Cursor, Stripe's $7B purchase of OpenRouter, Anthropic's first profitability at $11.5B revenue, and Silverlake's $43B bid for Workday. The hosts debate valuations, growth trajectories, market dynamics in AI, and the strategic implications of these deals.

Summary

The episode opens with discussion of SpaceX's $60 billion all-stock acquisition of Cursor, which represents a 10x return for early investors like Andreessen Horowitz. The hosts analyze how Cursor survived initial skepticism about negative gross margins by executing faster than competitors and adapting to market changes (particularly with multi-model support after Claude Code's launch). Elon Musk's strategic prowess is highlighted—he solved SpaceX's compute glut problem by acquiring the #2 coding product, creating synergies that justify the valuation. The discussion contrasts this with Meta's failure to acquire Cursor, attributing it partly to antitrust concerns and Elon's superior stock currency for the deal.

Stripe's $7 billion acquisition of OpenRouter is examined next. While OpenRouter raised at $1.3B just four months prior (representing a 5x increase), the hosts debate whether this valuation makes sense. Jason Lemkin argues OpenRouter is a niche product strong in developer tools and chatbots but weak for high-reasoning enterprise workflows requiring model consistency. Rory O'Driscoll counters that as long as frontier models extract value, enterprises will want multi-model options, though he acknowledges the tension between optionality and complexity. The acquisition reflects the broader pattern of acquirers buying infrastructure plays when markets move quickly rather than building internally.

Anthropic's announcement of Q2 profitability on $11.5B revenue is framed as inevitable given their gross margins and growth rate. The hosts explain that 40% gross margins applied to 12x revenue growth means the company is generating billions in gross profit faster than it can deploy expenses below the line. However, they note Anthropic's IPO filing will likely include complex footnotes about off-balance sheet compute commitments and extraordinary stock-based compensation. They project that only revenue growth matters for the IPO valuation; everything else is noise.

The conversation pivots to market sizing for AI in coding. The hosts debate whether Anthropic's $200B ARR projection for 2028 is achievable. Jason argues the realistic TAM is based on ~5-6M software-related workers in the US with ~$600B in annual salary spend. If AI represents $100K per engineer in annual token spend (with 30-40% team size reduction), the US market approaches $200B, with global markets reaching perhaps $350B. This contradicts the $600B projection, which the hosts find unrealistic without expanding beyond software.

Cursor, Lovable, and Higgs Field valuations are then discussed as the products have matured significantly since the show's inception. Lovable raised at a $13.3B valuation on ~$700M ARR, while Higgs Field raised at $5.5B valuation on similar revenue. The hosts note these products are no longer "little hacks"—they're developing genuine moats through feature richness, user retention, and talent acquisition. The Lovable valuation appears reasonable relative to Cursor's $60B price, suggesting an emerging competitive market with differentiated approaches.

Silverlake's $43B take-private bid for Workday prompts discussion of PE logic in mature SaaS. The hosts analyze this as precise financial engineering: buying at ~5.3x revenues, leveraging 4-5x EBITDA, maintaining 35% operating margins, and paying down debt with free cash flow to achieve ~20% IRR and 2x return over 5-6 years. They note Workday's advantage as a closed system of record (unlike Salesforce's open ecosystem) means it will retain customers even if growth slows. The challenge is ensuring the company grows—without growth, the PE math becomes unattractive. Jason suggests Silverlake benefits from founder Aneel Bhusri's return as CEO, potentially unlocking upside if he builds agentic versions of Workday.

The transcript concludes with discussions of Etched's $21B valuation (raised $700M just 3 weeks after closing at $10B, suggesting a strong market month) and the DOJ's investigation into Andreessen Horowitz board overlap at Databricks and Fivetran under the Clayton Act Section 8. The hosts frame this as a low-consequence regulatory matter likely resolved by board resignations, reflecting outdated antitrust law applied to modern venture scenarios.

About this episode

<p>AGENDA:</p> <p>04:20 Elon's Deal of the Decade: SpaceX Buys Cursor for $60BN<br /> 06:10 Why Cursor Was Surprisingly Cheap at $60BN<br /> 07:00 Why Zuckerberg Failed to Buy the AI Prize Elon Secured<br /> 12:00 Elon vs Zuck: Who Would You Rather Work For?<br /> 14:00 Will Microsoft or Amazon Now Race to Buy Cognition?<br /> 17:05 Stripe's $7BN OpenRouter Deal Creates Huge VC Winners<br /> 25:00 OpenRouter's Fatal Risk: Enterprises Don't Want 10 Models<br /> 28:15 Anthropic Turns Its First Profit on $11.5BN of Quarterly Revenue<br /> 32:15 Can Anthropic Really Reach $600BN in Revenue?<br /> 37:00 Why Every Elite Engineer Could Soon Get $100K in AI Tokens<br /> 39:30 Would Rory Buy Anthropic at a $2.5TN Valuation?<br /> 44:50 Silver Lake's $43BN Workday Bet: SaaS Isn't Dead, It's Mature<br /> 53:00 How Silver Lake Could Make $30BN From Workday<br /> 57:00 Lovable vs Higgsfield: Similar Revenue, Radically Different Valuations<br /> 58:00 Is Lovable's $13.3BN Price Actually Cheap?<br /> 63:30 Why the DOJ Is Coming After Andreessen Horowitz<br /> 69:00 Why A16Z Has "50 Legal Battles" Happening at Once</p> <p> </p>

Key Insights

  • SpaceX bought Cursor not just for the product but to fill its Colossus compute cluster with revenue-generating usage, making the marginal unit economics highly favorable for Elon despite the large headline price.
  • Cursor survived near-death experiences (perceived obsolescence after Claude Code launched) by executing faster than competitors and adapting product strategy, demonstrating that pessimists sound smart in the moment but optimists die rich in growing markets.
  • In hyper-growth markets with risk-on capital, buyers overlook negative gross margins and off-balance-sheet liabilities because the market is so large and execution is so strong that these issues get swamped by upside potential.
  • OpenRouter is a niche product that works well for low-stakes use cases (chatbots) and developer PLG but fails for high-reasoning enterprise workflows requiring model consistency, limiting its total addressable market despite strong product-market fit in specific segments.
  • Anthropic's profitability announcement will likely be accompanied by massive footnotes on stock-based compensation and compute commitments that make traditional financial metrics misleading; the only metric that matters for valuation is projected revenue growth rate.
  • A realistic US market for AI coding tools is approximately $200B ARR (based on 5-6M software workers earning $600B annually with $100K AI spend per employee), suggesting global markets of $300-350B rather than the $600B projections some are forecasting.
  • Lovable and Higgs Field have transformed from perceived niche toys into defensible platforms by accumulating feature richness, user data, and developer talent, demonstrating that moats accrete over time in new software markets rather than existing from day one.
  • The Silverlake Workday deal exemplifies how PE acquirers apply precise financial engineering to mature SaaS: they're buying at 5.3x revenues with the assumption they can maintain 35% operating margins and achieve 20% IRR through leverage and debt paydown, a strategy that only works if the business continues to grow.
  • Closed systems of record like Workday have structural advantages over open platforms like Salesforce because customers cannot easily abstract away via headless agents, providing a buffer against agentic competition and giving the acquirer more predictability for the 5-6 year PE hold period.
  • Founders' psychological preference matters in M&A outcomes: Michael Succarello preferred working for Elon over Mark Zuckerberg despite Meta's competence, suggesting that cultural and leadership fit can be a meaningful factor in deal selection beyond purely financial terms.
  • The DOJ investigation into Andreessen Horowitz's overlapping board seats (Databricks and Fivetran/dbt) under Clayton Act Section 8 reflects antiquated antitrust law designed to prevent 1900s-era industrial collusion being applied to modern venture capital with minimal real consumer harm risk.
  • Companies that do not finalize their 2027 roadmaps by August 2026 are already behind in the agentic world, indicating that AI execution velocity has become such a competitive differentiator that traditional quarterly planning cycles are functionally obsolete.

Topics

M&A in AI and SaaSSpaceX acquisition of CursorStripe acquisition of OpenRouterAnthropic profitability and IPO valuationMarket sizing for AI in software engineeringSaaS product maturation (Lovable, Higgs Field, Replit)Private equity LBO dynamics (Workday/Silverlake)Regulatory and antitrust issuesGross margins and unit economics in AIStartup valuation multiplesFounder incentives in M&AGrowth vs. profitability tradeoffs

Transcript

Your gross margin problem is my revenue opportunity for my Colossus cluster. Pessimists sound smart, optimists die rich. Only a fool denies that Elon Musk is wildly effective. I would much rather initially work for Elon than for Zuck. I think it'll be like the scale acquisition. I don't even think this product will exist in five years. You can't add expenses below the line fast enough to stop yourself making money. The only thing that matters will be the growth rate and the 27 and 28 projected revenue. Someone who was hired with a million dollar package in 23 ended up making 51 million four years later. A billion knowledge workers in the world, absolute bollocks. I think we'll…

Full transcript available for MurmurCast members

Sign Up to Access
View original source →

More from The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch

Get AI summaries like this delivered to your inbox daily

Get AI summaries delivered to your inbox

MurmurCast summarizes your YouTube channels, podcasts, and newsletters into one daily email digest.