The Twenty Minute VCDatabricks at $100BN, CoreWeave’s $11B Debt Bet & Nubank’s $2.5B Profit Shocker - Ep.19
CHAPTERS
- 0:00 – 2:58
Databricks at $100B: why the valuation no longer feels shocking
The group reacts to Databricks hitting a $100B private valuation and notes how the AI era has reset everyone’s sense of scale. They compare today’s mega-private valuations (Anthropic, OpenAI, SpaceX) to show why $100B now barely moves the needle.
- •$100B outcomes feel normalized in the current AI cycle
- •Private vs public markets can feel like “two different worlds”
- •Contextual comps: Anthropic, OpenAI, SpaceX and the new baseline for scale
- •Initial instinct: $100B no longer automatically signals ‘bubble’
- 2:58 – 5:36
Is Databricks actually undervalued? Growth-rate comps vs Snowflake
They dig into Databricks’ reported ~$4B ARR at ~50% growth and compare it to Snowflake’s ~$4B run-rate growing ~26%. At a stated ~25x revenue multiple, they argue Databricks can look surprisingly reasonable—if growth persists.
- •Databricks: ~4B ARR and ~50% growth vs Snowflake ~26% growth
- •25x revenue can look ‘cheap’ if growth durability holds
- •The real bet: growth persistence vs a sudden deceleration
- •AI infra positioning strengthens the case for sustained demand
- 5:36 – 13:50
Tech lead and “bubble + IPO wave” dynamics: Figma, Canva, Stripe as accelerants
Harry cites Databricks’ CRO claiming a multi-year tech lead over Snowflake, then Jason and Rory debate what a strong IPO pipeline could do to sentiment. They discuss whether IPO pops are scarcity-driven and what happens when more elite private companies list.
- •Claimed “five-year lead” over Snowflake (per Databricks CRO)
- •Jason’s thesis: best IPOs are still to come; returns could fuel more froth
- •Rory’s counter: scarcity premium may fade as more high-growth IPOs arrive
- •Figma’s frothy multiple and Canva as a potential ‘jaw-dropper’ comp
- 13:50 – 19:14
Andreessen’s Databricks bet: fund math, ownership, and the power-law payoff
They run the thought experiment: if Databricks IPOs at a much higher valuation, early backers like a16z could see massive absolute gains. The discussion highlights how decade-long compounding plus follow-on ownership can produce $10B–$30B outcomes for a single firm.
- •Early entry economics matter: old-vintage pricing + long hold periods
- •Potential ownership and follow-on strategy can create outsized absolute returns
- •Private-market downside: you ‘ride it down’ without public stop-loss liquidity
- •Secondary/SPV “spray” creates many small winners, not just big funds
- 19:14 – 25:02
Chamath’s SPAC return: structure critique, incentives, and ‘casino’ framing
The conversation shifts to the re-emergence of SPACs as a potential bubble signal. Rory critiques SPAC mechanics (adverse selection, misaligned incentives) and argues capital markets aren’t a casino; Jason questions why wealthy sponsors bother with the hassle.
- •You can’t know ‘peak bubble’ until after the turn, but SPAC resurgence correlates
- •SPAC incentive misalignment: promoters get paid for closing deals, not quality
- •Adverse selection increases in frothy markets
- •“Wall Street isn’t a casino” vs retail complexity and who bears the risk
- 25:02 – 29:24
Retail access to private markets: disclosure vs protection and valuation markups
Jason raises concerns about opaque PE/secondary markups and whether retail investors can truly understand them. Rory frames the trade-off between buyer-beware disclosure and regulatory protection—balancing access to upside with protection from complexity-driven losses.
- •Complexity often benefits promoters over less sophisticated investors
- •Disclosure-only model vs ‘nanny state’ protection model
- •Regulation can also block retail from elite upside (e.g., early Databricks)
- •The real question: how to design fair access without adverse selection
- 29:24 – 35:44
OpenAI staff secondaries: retention, Meta’s cash pressure, and ‘pseudo-public’ liquidity
They discuss OpenAI’s large employee secondary and how cash-outs intersect with competition for talent, especially from Meta. Jason argues late-stage private companies should allow more routine liquidity, while Rory notes money reveals who truly wants to keep building.
- •Secondaries as a competitive tool amid $10M–$100M talent offers
- •Making private companies ‘pseudo-public’ with periodic liquidity for employees
- •Weakened golden-handcuff dynamics: retention via illiquid equity is changing
- •Wealth effects: some people double down; others pursue new life goals
- 35:44 – 38:39
Founder raises $130M then leaves: why more ‘Hoppins’ moments are inevitable
The Story founder departure becomes a lens on late-stage venture loss rates and incentive drift in overheated markets. Jason predicts there should be many such cases in any boom, and Rory treats it as statistically normal rather than scandalous.
- •In frothy cycles, expect more founders to walk away with large payouts
- •Late-stage loss rates can be high; ‘tawdry’ outcomes are not surprising
- •Fast, weekend ‘hot deal’ dynamics can increase error rates
- •Reframing: it’s an expected cost of the venture model, not an anomaly
- 38:39 – 44:13
Nubank’s $2.5B profit and the neobank comparison: Nubank vs Revolut vs Chime
Rory steps back from the quarter to compare global neobanks and the market structure they attacked. He argues Nubank’s dominance and profitability reflect weaker incumbents in LatAm, while Chime’s smaller outcome reflects tougher U.S. banking competition; Revolut’s wedge came from cross-border FX.
- •Nubank: ‘all bank’ model (deposits + lending) and scale in LatAm
- •Chime: profitable via interchange/deposits without lending; smaller TAM capture
- •Revolut: started with cross-border FX/crypto, expanding into broader banking
- •Core pattern: find weak, high-friction banking niches and compound product breadth
- 44:13 – 49:49
Can Nubank be worth much more? Scaling upside vs ‘banks eventually do dumb stuff’
Harry pushes the bull case that Nubank could be dramatically larger; Rory agrees there’s room to run but flags the risk of eventually being valued like a traditional bank. The key tension is whether Nubank can keep growing and avoid regulatory/credit pitfalls that hit scaled banks.
- •Upside case: expansion beyond three countries and broader product suite
- •Valuation gravity: mature banks trade on book value, not tech multiples
- •Risks: regulation, credit cycles, and ‘bad loans’ as the business scales
- •If Nubank avoids classic bank mistakes, 2x+ outcomes are plausible
- 49:49 – 54:17
Revolut’s ‘venture lab’ operating system: building at scale with internal GMs
Harry describes Revolut’s internal pipeline of dozens of products run like venture teams, with rigorous weekly metric reviews. Jason highlights this as a repeatable playbook among elite founders—recruiting former CEOs as GMs to accelerate multi-product expansion.
- •Dozens of products incubated with testing gates and internal funding logic
- •Using ex-founders as GMs is a scalable company-building superpower
- •Multi-product compounding turns a wedge (‘snack’) into a primary bank (‘meal’)
- •Execution advantage: fast iteration + organizational design, not just idea quality
- 54:17 – 1:06:05
CoreWeave’s $11B+ debt: real estate-like model, mismatch risk, and ‘canary in the coal mine’
They interpret CoreWeave as a sophisticated financing vehicle for GPU infrastructure and argue debt is inherent given massive capex. The real risk is asset-liability mismatch—short-term demand against long-term obligations—and Jason frames CoreWeave as an early warning signal if AI capex cools.
- •Debt is expected when capex plans reach tens of billions
- •CoreWeave resembles leveraged infrastructure/real-estate with long-term leases
- •Key risk: contract strength (take-or-pay), duration, and customer concentration
- •CoreWeave (and imitators) may show stress before hyperscalers do
- 1:06:05 – 1:12:31
Will AI spend hit ‘trillions’? Metaphor vs math, and the stock market’s AI concentration
The discussion broadens to whether trillion-dollar capex claims are feasible and what it means that major indices are heavily driven by AI winners. Rory challenges the literal math and financing capacity, while acknowledging that if labor budgets shift to AI automation, the spend could be rational over time.
- •Altman’s ‘trillions’ framed as possibly metaphor rather than near-term plan
- •Hyperscaler capex is already straining balance sheets; financing has limits
- •Public markets are highly concentrated in a few AI names (index-level exposure)
- •Key unlock: shifting from tech budgets to labor budgets via real automation
- 1:12:31 – 1:20:16
AI agents in practice: labor replacement, budget sprawl, and the coming consolidation wave
Jason shares firsthand operational experience: multiple paid AI agents replacing human work, but also rapidly inflating software spend. They predict a fast consolidation cycle as enterprises balk at paying for dozens of overlapping agent tools, favoring platforms that bundle orchestration and breadth.
- •Early but real labor substitution at the ‘pointy edge’ of adoption
- •AI spend can balloon quickly (hundreds of thousands to millions annually)
- •Enterprises will consolidate agent stacks faster than classic SaaS consolidation
- •Winners will expand from wedge products to broad platforms with orchestration
- 1:20:16 – 1:26:43
Platform risk is back: vertical SaaS implications and Epic vs Abridge example
They discuss how platform incumbents (like Epic in healthcare) will inevitably build competing features, making platform dependency a core risk again. Rory argues vertical SaaS winners can still expand once trusted, while both note investors are currently ignoring classic risks because growth is so strong.
- •Epic’s move into transcription/scribe is predictable platform behavior
- •Partnering with the platform can be necessary even if it later competes
- •Vertical SaaS leaders can expand breadth once they win trust in a domain
- •Market mood: investors are downplaying traditional venture risks amid AI frenzy
- 1:26:43 – 1:34:58
Kalshi quick-fire predictions: Claude 5 timing, Mistral M&A, and Deel vs Rippling IPO
They close with rapid prediction-market style questions and probabilistic thinking. Jason argues Anthropic may prioritize coding improvements over a big consumer “Claude 5” moment, expects sub-scale foundation-model players may need strategic moves, and sees Deel’s profitability as enabling a sooner IPO than Rippling.
- •Claude 5 odds: incentives may favor Sonnet iterations vs consumer headline release
- •Mistral acquisition: ‘take the deal if you get it’ vs uncertainty of getting it
- •M&A pricing can be driven by narrative/round anchoring as much as revenue quality
- •IPO race: Deel’s profitability vs Rippling’s investment mode and capital strategy
