20VC: Sam Altman Offers Trump 5% of OpenAI: Fool or Genius? | Alex Karp Sounds the Alarm: Enterprises Fear Frontier Models & Questionable ROI of AI | The Rise of Chinese Open Source: Deepseek Building Own Chips
In this 20VC episode, Harry Stebbings, Rory O'Driscoll, and Jason Lemkin discuss major AI industry developments including government regulation of frontier models, Sam Altman's controversial 5% OpenAI stake offer to the US government, Chinese open-source models gaining dominance, and enterprise skepticism about AI ROI despite massive capex investments by hyperscalers.
Summary
The episode opens with discussion of Washington lifting the Fable 5 ban and implementing a pre-approval process for frontier models, marking a significant shift from the internet era's deregulatory approach. The hosts debate Sam Altman's proposal to give the US government 5% of OpenAI, with Rory viewing it as dangerous precedent-setting that could lead to Bernie Sanders demanding 50%, while Jason argues ownership stakes create meaningful alignment (citing the Klaviyo-Shopify example). They note this contradicts the founding narrative of tech companies seeking freedom from regulation.
The conversation shifts to Alex Karp's CNBC appearance highlighting two critical issues: enterprises express unprecedented skepticism toward frontier model providers despite massive spending, and there's serious questioning about AI ROI within organizations. The hosts discuss data security concerns, particularly HubSpot's failed attempt to share customer prospecting data and the broader risk of vendors pushing boundaries on data training and privacy.
On compute and infrastructure, Meta's announcement of a cloud business selling excess AI capacity receives mixed interpretation—the hosts debate whether this signals failed AI ambitions or smart hedging. They discuss NVIDIA's aggressive financing strategies (Compute Now Pay Later) and the tension between supply-side compute spending and demand-side revenue growth. The hosts emphasize that spending won't stop until demand-side signals weaken.
Chinese AI developments receive substantial coverage: DeepSeek, Kling, and other Chinese models are outperforming US alternatives in open-source rankings and video generation, with Kling valued at $18 billion despite only $500M in revenue. The hosts attribute this partly to US export restrictions forcing Chinese companies to build independently, noting Jensen Huang's earlier warnings about this consequence. They discuss potential Chinese restrictions on overseas access to their models as a mirror to US restrictions.
On the enterprise services front, Microsoft and Amazon's $2.5 billion commitment to embed engineers inside enterprise clients addresses the MIT finding that 95% of AI pilots deliver no measurable impact. Jason argues this will fail due to insufficient FDE talent depth, while Rory contends it mirrors IBM's successful transformation into a services company helping enterprises adopt new technologies. They discuss the need for both technical experts and domain experts in deployments.
The episode covers dilution trends in AI funding, noting that founders today are less sensitive to per-round dilution due to massive valuations, with examples like Anthropic's founder owning ~1.3% equity despite building one of the most valuable startups. They discuss how this differs from previous generations and whether massive outcomes justify dilution tolerance.
Final topics include 11 Labs' $22 billion secondary valuation and the broader question of employee incentives—Jason argues talented employees shouldn't join companies without clear paths to liquidity via tender offers within 24 months, while Rory emphasizes joining pre-tender offer companies with strong upside potential. The episode concludes with discussion of Ashton Kutcher leaving Sound Ventures to start a new deep-tech seed fund with Morgan Bella, which the hosts interpret as a natural evolution rather than conflict.
About this episode
<p>AGENDA:</p> <p>05:00 Washington Just Put Frontier AI on a Leash<br /> 06:30 Sam Altman's Wild 5% Government Stake Idea<br /> 19:00 The AI Funding Bubble: Why Founders No Longer Fear Dilution<br /> 28:00 Alex Karp's Brutal Warning: Enterprises Don't Trust Frontier AI<br /> 33:00 Meta's Shock Pivot: Has Zuck Accidentally Built the Next CoreWeave?<br /> 41:00 Nvidia's Dangerous New Game: "Compute Now, Pay Later"<br /> 45:00 Anthropic & DeepSeek Go After Nvidia's Crown<br /> 48:00 Kling vs Sora: Did China Just Win AI Video?<br /> 52:00 Is China Secretly Winning the Open Source AI War?<br /> 01:02:00 Microsoft & Amazon's $6B Bet: AI Still Needs Humans<br /> 01:11:00 Ashton Kutcher Walks Away From Sound Ventures<br /> 01:16:00 The New Startup Talent War: No Liquidity, No Chance<br /> 01:20:00 Final Thoughts: Who Wins the AI Endgame?</p>
Key Insights
- Sam Altman's proposal to give 5% of OpenAI to the US government sets a dangerous precedent that could invite demands for much larger stakes (Bernie Sanders' 50% proposal) if the company's own claims about AI causing massive job displacement prove true.
- The shift from internet-era deregulation (Telecom Act, Section 230, no sales tax) to current proactive regulation and government equity stakes represents a fundamental reversal in tech industry strategy toward Washington.
- Enterprise customers express unprecedented skepticism toward frontier models despite committing billions in capex, driven by concerns about ROI, data privacy, and whether their proprietary information is being used to train competing models.
- Chinese AI companies (DeepSeek, Kling, others) are filling a market vacuum created by US export restrictions, with Kling achieving $18B valuation on $500M revenue partly because US companies like OpenAI abandoned the consumer video generation market.
- Compute demand, not supply, will determine the success of hyperscaler capex spending, and as long as enterprise revenue growth rates (2X-3X) justify the spending, companies like Meta will continue investing regardless of usage uncertainty.
- NVIDIA's aggressive financing strategies (Compute Now Pay Later, revenue-sharing arrangements) recognize that downstream companies may lack cash to purchase GPUs upfront, effectively financing demand by recognizing revenue early while taking putback risk.
- The talent depth to successfully deploy AI in enterprises doesn't exist at the scale needed, with even top-tier companies lacking sufficient field deployment engineers, making Microsoft's $2.5B services bet risky.
- Founder equity ownership in mega-cap AI companies is dramatically lower than historical precedent (Dario at 1.3%, Sam at ~0%), yet founders remain less concerned about dilution because ultimate outcomes are potentially 10-100x larger than previous company exits.
- Employees making career decisions face a new dynamic where tender offers (not IPOs) have become the proxy for liquidity, making it rational to only join companies with clear paths to tender offers within 24 months.
- US restrictions on Chinese access to frontier AI models (preventing use of ChatGPT, Claude in China) have directly enabled Chinese companies to develop competitive alternatives that now rank top in open-source benchmarks.
- The 95% enterprise AI pilot failure rate reflects not technical problems but organizational change management and lack of integrated domain expertise, requiring both technical experts and subject matter specialists in deployments.
- Ashton Kutcher's departure from Sound Ventures to start a new deep-tech fund reflects modern venture dynamics where individual founder/investor brand matters more than firm brand, enabling talent to move without traditional corporate friction.
Topics
Transcript
It's like rewriting Atlas Shrugged, where John Galt goes to Washington and says, why don't you regulate me more? Why don't you take more? Why don't you take us, Mr. Mooch? Grab some of my stuff. What the fuck are these people thinking volunteering for this stuff? Madness. No one's worried about making their last round high-priced investors money anymore. Literally, no one is. Every technology company either goes bust or lives long enough to become next generation's IBM. As an employee today, why would you join something that you don't believe will have secondary options? This is 20VC with me, Harry Stebbings. Now, it's my favorite show of the week. Rory O'Driscoll, Jason Lemkin analyzing the biggest news in…
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.