The Twenty Minute VCAnthropic's Fable Banned by US Government | Wix & Adobe Hit All-Time Lows | Mistral Raising at $20BN
CHAPTERS
- 0:00 – 3:52
SpaceX IPO debuts at historic scale: pricing, pop, and early trading dynamics
The group breaks down SpaceX’s record-setting IPO, including the day-one pop, rapid market-cap expansion, and how Elon’s pricing approach avoided traditional price discovery. They also note why early trading can be misleading due to lockups and low float.
- •SpaceX IPO performance and the “designer pop” debate
- •Elon’s pricing power vs. traditional IPO price discovery
- •Why early market-cap headlines can distort reality during lockup
- •Float is tiny (~4%), setting up unusual volatility
- •Over/under predictions for where the stock trades in 6–12 months
- 3:52 – 9:11
Gamma squeeze 101: why options can mechanically drive SpaceX shares higher
Everett explains how a gamma squeeze works, why the launch of options trading matters, and why low float amplifies forced buying. The chapter frames how technical flows can dominate fundamentals shortly after IPO.
- •Call option buying forces market makers to buy shares to hedge
- •Feedback loop: rising stock → more calls → more hedging purchases
- •Low float makes the loop more extreme and faster-moving
- •Why volatility and option pricing can be prohibitive for short bets
- •Why the lockup expiration is the first true ‘real’ pricing moment
- 9:11 – 12:01
Elon’s long-dated call-option playbook: narrative, execution, and cost of capital
The conversation shifts to how Musk repeatedly funds ambitious, long-horizon bets by selling investors on ‘long-dated call options’—big future projects that expand perceived upside. They argue his credibility lowers his cost of capital and enables moves others can’t finance.
- •Everett’s Tesla short lesson and later SpaceX investment perspective
- •How Musk structures multi-year ‘optionality’ (FSD, Optimus, Starship, etc.)
- •Why markets often price Musk’s future claims with unusual trust
- •Cost of capital as a compounding competitive advantage
- •“Village of believers” who fund Elon repeatedly and win historically
- 12:01 – 18:33
Compute problem solved twice: Cursor acquisition and massive compute contracts
They discuss how Musk’s AI strategy filled compute utilization gaps through both acquisition and major commercial contracts. The point: speed of execution and monetization can reframe the company’s revenue mix, even relative to SpaceX/Starlink.
- •Cursor buy as a strategic compute utilization/retention move
- •Large Google and Anthropic contracts as revenue stabilizers
- •Claims that ‘AI business’ run-rate could surpass SpaceX/Starlink
- •Execution speed vs. rivals still ‘planning’ AI moves
- •How contracts and acquisitions function as operational hedges
- 18:33 – 19:51
Tesla–SpaceX merger speculation: incentives, governance simplicity, and market odds
Harry asks whether Musk will consolidate Tesla and X/SpaceX-related entities. Rory argues consolidation would simplify governance for Elon and likely faces limited shareholder resistance given recent gains, though prediction markets aren’t definitive.
- •Why one board/one shareholder base could be appealing
- •Market odds and the reliability of prediction markets
- •Shareholder incentives after massive value creation
- •Systemic risk: “single point of failure” if everything rolls together
- •Musk’s growing burden to keep delivering returns
- 19:51 – 23:39
Claude Fable banned: timeline, trust breakdown, and the cybersecurity trigger
The panel reconstructs the Anthropic ‘Claude Fable’ launch and rapid US government ban, emphasizing mistrust and communication failure. They distinguish Anthropic’s technical argument (scale and orchestration risk) from the government’s political need for a clear, decisive response.
- •Fable positioned as front-end to more dangerous Mythos model
- •Reported discovery of a jailbreak/cybersecurity capability
- •90-minute call ends with government threatening/using export powers
- •Anthropic’s argument: single instance ≠ scalable autonomous cyber threat
- •Politics and narrative overpower nuanced technical distinctions
- 23:39 – 31:09
Export Restriction Act and capability-based regulation: the new ‘Rubicon’
Everett frames this as the first major US move to regulate an AI model explicitly based on capabilities and foreign-adversary risk. They explore what happens if other labs reach similar capability levels and whether the government can apply rules consistently.
- •Shift from contract/ToS disputes to capability-based restriction
- •Potential precedent: gating access to intelligence by citizenship/country
- •Future implications for ASI and sovereign access frameworks
- •Consistency challenge when OpenAI/Google achieve comparable capabilities
- •Due process/equal-treatment questions if only one lab is targeted
- 31:09 – 38:02
Benchmarks vs. test-time compute: why ‘scorecards’ miss the real model behavior
They argue benchmark cards are increasingly misleading because models can improve materially with more inference-time compute. This points to exploding demand for inference infrastructure and blurs lines between frontier and open(-ish) models when ‘compute is the knob.’
- •Noam Brown’s critique: benchmark comparisons ignore scaling at test time
- •Throwing more inference compute can unlock better reasoning/capabilities
- •Implications for inference platforms and token-throughput demand
- •Open models may replicate ‘frontier-like’ outcomes given enough compute
- •Policy tension: restricting one model may be incoherent if others can simulate results
- 38:02 – 40:03
Anthropic IPO odds after the ban: near-term friction vs. strong market window
Harry presses on whether Anthropic will IPO this year. Both guests still lean ‘yes’ given the strong market reception to AI-linked stories, but acknowledge the ban is a material wrinkle requiring rapid resolution.
- •Rory’s Bayesian view: probability down, still >50%
- •Market window looks attractive post-SpaceX AI-reframing
- •Regulatory uncertainty becomes a prospectus/roadshow issue
- •Anecdotal demand: claims of dramatically higher ARR potential with Fable
- •Key variable: whether the situation stabilizes quickly
- 40:03 – 47:35
Mistral and sovereign AI: demand is real, but building top models is brutally hard
The discussion turns to Mistral’s reported raise at a $20B valuation and the broader push for national/European alternatives. They argue sovereignty concerns can support ‘second-best’ options, but the talent and capital requirements make true frontier competition rare.
- •Sovereignty trade-off: reliability of access vs. absolute model quality
- •Europe may fund alternatives but struggle to match $100B-scale investment
- •Scarcity of pretraining talent and operational excellence outside frontier labs
- •Outside China, few truly strong open-weight model producers
- •Likely end-state: a small oligopoly with regional affiliations
- 47:35 – 55:48
Salesforce acquires Fin (Intercom): the playbook for pre-AI SaaS survival
Rory (an investor) explains why Intercom/Fin’s transformation is a canonical example of an older SaaS company successfully pivoting to AI. They highlight the shift from seat-based pricing to outcome-based monetization and why Salesforce needs that DNA for its own AI ambitions.
- •Fin as proof old SaaS can ‘burn the boats’ and reinvent
- •Outcome-based pricing (pay-per-resolution) aligns vendor to customer value
- •Why this forces product quality, deployment, and customer intimacy
- •Salesforce rationale: acquire the team and the transition know-how
- •Board-level lesson: a viable liquidity path exists for transformed incumbents
- 55:48 – 1:06:08
Wix guidance cut and layoffs: when AI replicability crushes multiples—and what to do next
They dissect Wix’s guidance reduction, workforce cuts, and why markets punish products that appear easy to replicate with AI agents. Despite a cheap multiple (~1x revenue) and AI-adjacent acquisitions, they debate whether taking the company private could be the best path to execute a multi-year turnaround.
- •Public markets’ harsh filter: compelling AI story vs. ‘legacy SaaS’ exposure
- •Replicability risk from AI website-building tools (Lovable/Replit, etc.)
- •Why buybacks look especially painful when valuation compresses
- •At low multiples, ‘price becomes the forcing function’ for contrarian buyers
- •Private-market playbook: find a Silver Lake-style partner and grind it out
- 1:06:08 – 1:16:43
Adobe beats but drops: CFO exit, SaaS-to-semis rotation, and the trapped incumbent problem
Adobe’s stock falls despite strong results, driven by leadership transition and fear of AI disruption. They argue many hedge funds prefer semiconductors over SaaS due to clearer tailwinds, and incumbents like Adobe face a vicious cycle: shrinking currency limits their ability to buy AI talent/products.
- •CFO departure can overwhelm good earnings in a ‘turnaround’ narrative
- •Adobe exhibits multiple negatives: seat model, incumbency, share-to-lose, perceived replicability
- •Value signal: extremely low free-cash-flow multiple vs. deep uncertainty
- •Hedge fund rotation: SaaS underperforms while semis surge (SOXX vs cloud indices)
- •Why cheap incumbents struggle to acquire AI leaders without nuking their stock
- 1:16:43 – 1:28:09
Standard Bots and the robotics debate: humanoids vs. pragmatic industrial automation
In the closing segment, Rory highlights Standard Bots’ $200M raise and its argument against near-term humanoids in many industrial contexts. They discuss why robotics adoption is slower than software, but why LLM-driven flexibility could finally reduce brittleness and expand real-world deployments.
- •Standard Bots’ thesis: avoid expensive legs; focus on high-ROI industrial tasks
- •Middle path between old robotic arms and over-scoped humanoids
- •Real-world constraints: ROI sensitivity and ‘surprising detail’ edge cases
- •Benchmark’s Sunday Robotics and demonstrations of adaptive behavior
- •Long-term outlook: robotics as a potential trillion-dollar category, but on a slower curve
