20VC: NVIDIA Bonanza: Buys Poolside & Invests in Mercor and Perplexity | Anthropic's $30TRN Revenue Assumption & OpenAI Confirms IPO | Why Customer Service, Defence and Robotics are Overinflated
In this 20VC episode, Jason Lemkin, Rory O'Driscoll, and Harry Stebbings discuss major AI industry developments including NVIDIA's $12B Poolside acquisition, OpenAI's 2027 IPO announcement, and Anthropic's growing dominance. They analyze whether current valuations and market dynamics are sustainable, debating the emerging 'token addiction' in enterprises and which AI investment categories are overheated.
Summary
The discussion opens with NVIDIA's strategic acquisitions across the AI stack. NVIDIA is licensing Poolside's model factory for $6B plus $1B investment at a $12B pre-money valuation, acquiring engineering talent to strengthen Nemotron. The hosts debate whether this represents a success story or failure, concluding that even a 15x return on seed investment falls short of the 50x+ needed for seed fund economics in 2026. They note that VC funding has largely exited frontier model development, which is now dominated by hyperscalers (Microsoft, Google, Amazon) and NVIDIA itself. The conversation highlights that NVIDIA is systematically investing its massive free cash flow ($75B+ annually) across the ecosystem to expand the addressable market for chips.
The discussion then pivots to OpenAI's forced hand in announcing a 2027 IPO. Analysis of OpenAI's Q1-Q2 growth rates (18% quarter-over-quarter) revealed they were growing significantly slower than Anthropic (estimated at 60B+ annual run rate vs OpenAI's 30B estimate). This created existential competitive pressure, forcing OpenAI's CFO Sarah Fryer to publicly commit to going public in 2027. The hosts argue OpenAI faced a choice between appearing growth-constrained and communicating momentum. They assess that OpenAI is now positioned as number two in the market, which creates pressure from multiple angles: Anthropic is ahead on growth, open-source models are commoditizing inference, and the company must now go public on less favorable terms than if it maintained perceived momentum.
A critical theme emerges around 'token addiction' in enterprises. The hosts describe how companies initially over-spent on tokens without budget controls, then faced backlash, but now find themselves unable to retreat. Employees accustomed to running 10+ agents 24/7 won't accept constraints, creating organizational dependency. CFOs face a dual challenge: managing token budgets while retaining top talent who could leave for AI leaders like OpenAI and Anthropic. The Stripe investor letter's analogy of treating intelligence like capital allocation (rather than SaaS seat licensing) resonates as the framework for managing this transition.
On specific investments, they discuss NVIDIA funding Mercor ($20B round) and Perplexity ($30B round), noting that NVIDIA views ecosystem funding as cost of maintaining compute monopoly. They contrast this with Poolside (time expansion through acquisition) versus Mercor (pure ecosystem investing). The conversation highlights how even negative gross margin businesses like Cursor can become valuable if they achieve scale, but margins aren't guaranteed to improve. Stripe's 41% acceleration and 71% billing growth demonstrate how traditional companies are benefiting from AI tailwinds, positioning them ahead of legacy software companies in public markets.
The hosts conclude with 'dumbest investment categories' in AI: customer support software (becoming commoditized and absorbed into broader agent platforms), defense and robotics (will likely consolidate to 2-3 major players rather than distributed outcomes), and AI-driven professional services (law firms, accounting firms transformed by venture capital). They emphasize that the direction of travel feels right on these categories, but venture returns may be constrained by market structure, not product viability.
About this episode
<p><strong>AGENDA: </strong></p> <p>00:00 NVIDIA Launches a Three-Deal Blitz Across the AI Stack<br /> 04:31 NVIDIA's $12BN Poolside Deal Exposes Frontier AI's Capital Wall<br /> 15:39 NVIDIA Moves to Back Mercor at $20BN and Perplexity at $30BN<br /> 26:55 OpenAI Confirms 2027 IPO as Anthropic Seizes the Lead<br /> 40:42 Hugging Face Draws $13BN Takeover Interest as Open Models Boom<br /> 44:56 Citadel Cashes Out 80% of Its Leopold Aschenbrenner AI Trade<br /> 51:51 AI Boom Is "Less Than One-Third Over"—But Token Addiction Has a Price<br /> 01:04:07 Stripe Reaccelerates to 41% as AI Reshapes Software's Leaderboard<br /> 01:07:45 Instinct Security Scare Exposes Why AI Agents Still Cannot Be Trusted<br /> 01:23:55 VCs Name AI's Biggest Bubbles: Support, Defense, Humanoids and Roll-Ups</p> <p> </p>
Key Insights
- Only four or five companies globally (hyperscalers plus NVIDIA) have sufficient capital to finance frontier foundation models; VCs have been sidelined from funding this tier of AI development
- A $9B exit returning 15x on seed investment fails to clear the bar for modern seed fund economics, which require 50x+ returns due to later entry valuations and dilution patterns
- NVIDIA's ecosystem investments function as vendor financing to expand total addressable market for chips rather than pure financial returns, with an estimated $75B annual free cash flow budget allocated across portfolio companies
- OpenAI faced existential competitive pressure after Q2 growth rates (18% QoQ) revealed significant underperformance versus Anthropic's estimated 60B+ run rate, forcing premature IPO announcement
- Being number two in a market with abundant open-source alternatives creates compounding competitive pressure, forcing OpenAI to compete on brand and security rather than price or performance differentiation
- Enterprise token spending has evolved from unmanaged excessive consumption to constrained budget allocation, but employees now depend on continuous agent access for productivity and won't accept removal
- CFOs must simultaneously manage token budgets as constrained resources while retaining employees threatened by poaching from well-funded AI leaders, creating irreconcilable organizational tension
- Stripe's letter reframed intelligence allocation as analogous to capital allocation, requiring enterprises to make tradeoff decisions between different productivity investments rather than treating tokens as unlimited
- Customer support and CX software categories are collapsing as agent technology absorbs support functions into broader agentic platforms, reducing addressable market for standalone vendors
- Defense and robotics investments will likely produce returns, but consolidation patterns suggest 2-3 dominant players will emerge rather than distributed venture outcomes typical in software
- Cursor succeeded despite negative gross margins by achieving scale and improving margins over time, but not all negative-margin businesses follow this path, making margin trajectory improvement a critical but uncertain bet
- Anthropic's 30 trillion TAM claim (equivalent to US GDP) reflects necessary market sizing for justifying valuation but creates credibility concerns about overreach when compared to historical company outcomes
Topics
Transcript
9 billion doesn't clear the bar for seed investing in 2026. Let me tell you one thing I guarantee. If you get 15x on your failures in venture, you'll die a rich man. The VC money ran out on Entropic and OpenAI long ago, which is why no VC owns more than 1% or 2% of either of them. What if it all goes right? Entry price for any of these winners doesn't matter. They'll go out whatever price they get in 2027 because they can't wait any longer. If danger can be described as the absence of choice, they were now in danger. It's all about code. That's the only sentence that matters. Your time is the entire US…
Full transcript available for MurmurCast members
Sign Up to AccessMore from The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch
20VC: Is Seed Investing Dead Without a $1BN Fund? | Does Ownership and Price Matter When Companies Can Be $1TRN Exits | Are AI Revenue Numbers Real and What to Watch Out For with Venky Ganesan, Menlo Ventures
Venky Ganesan from Menlo Ventures discusses how venture capital has fundamentally changed in the AI era, arguing that seed investing still exists but founders must understand that venture is now focused on outlier companies with explosive growth rather than steady revenue growth. He emphasizes the importance of IRR over ownership percentages, the dangers of gaming metrics, and the need for VCs to play the current market game while maintaining authentic values.
20VC: The Future of Datacentres: What You Need to Know | Why Everyone Gets GPU Depreciation and AI's Energy Costs Wrong | Who Really Makes Money From AI & Why Most Moats Don't Exist with Chase Lochmiller
Chase Lochmiller, CEO of Crusoe Energy, discusses how data centers and AI infrastructure are becoming critical bottlenecks, arguing that energy-distributed computing rather than centralized hubs will power AI's future, and that vertical integration across data centers, GPUs, and inference services creates sustainable competitive advantages in the AI infrastructure space.
20VC: $1BN ARR in 18 Months; The Untold Story of Higgsfield | Spending $4M Per Month on Models | Why Moats in AI are BS | Scaling a Content Team to 150 People with Alex Mashrabov
Alex Mashrabov, founder of Higgsfield, discusses how his AI video generation startup achieved $1 billion ARR in 18 months—faster than Coursera—by focusing on product-market fit, strategic model routing between open-source and proprietary AI models, and building a 400-person team primarily based in Kazakhstan. He shares insights on AI moats being largely meaningless, the $4 million monthly spend on model inference, and his vision for Higgsfield to become bigger than Shopify by owning the distribution layer for AI-generated content.
20VC: Jensen Huang Declares AGI Has Arrived | GPT Astra and Fable 5.1 Accelerate the Model Race | Tesla Launches Cybercabs | Index Pulls Out of Town & Anthropic Pulls From Descartes Acquisition
In this 20VC episode, Rory O'Driscoll and Jason Lemkin discuss major tech news including Jensen Huang's AGI claims, AI agent risks and rule-breaking behavior, Elon's Cybercab launch, venture capital conflicts around AI assistants, and high-growth companies like Wonderful raising at massive valuations with significant secondaries.
20VC: The $100 Billion AI Assistant Race: Town vs Instinct vs GrokBot | We Spend $75K Per Engineer on AI Tools | Why the AI Assistant Market Is Not a Bubble & AI Assistants Will Replace Every App on Your Phone with JD, Founder of Town
JD, founder of Town, discusses how his AI assistant platform operates within email and calendar to automate work tasks, argues that network effects at the agent level will be the true moat in the AI assistant market, and explains why the speed of market development means companies can only learn at human speed while building at machine speed.