The Twenty Minute VCDatabricks at $100BN, CoreWeave’s $11B Debt Bet & Nubank’s $2.5B Profit Shocker - Ep.19
CHAPTERS
Holiday setup + why this episode is packed (Databricks, CoreWeave, Nubank)
Harry opens from Greece with Rory and Jason, setting the agenda: Databricks’ $100B valuation, Nubank’s profitability surprise, CoreWeave’s debt-fueled expansion, and broader AI-market froth. The tone is playful, but the frame is serious: are today’s prices justified or bubble behavior?
Databricks at $100B: “Doesn’t even feel big anymore”
The group reacts to Databricks hitting a $100B private valuation and how desensitized markets have become in the AI era. They compare it to other giant private marks (Anthropic, SpaceX, OpenAI) and ask whether $100B is now simply “normal.”
Is Databricks undervalued at ~25x revenue? Comparing to Snowflake
They ground the valuation debate in numbers: Snowflake’s ~$4B run-rate growth vs Databricks crossing ~$4B with ~50% growth. The conclusion: at similar scale but much faster growth, Databricks can plausibly justify (or even “look cheap” at) 25x—if growth persists.
IPO wave dynamics: froth, scarcity value, and what happens if Canva/Stripe list
Jason argues the “epic” IPOs may still be ahead, potentially amplifying returns and froth. Rory counters with a scarcity-value theory: public markets are starved of high-growth names, so more IPO supply could reduce pops and normalize pricing—while still unlocking huge LP liquidity.
Andreessen’s Databricks upside: could one deal return $30B+?
They explore how early entry economics (Seed/A-era pricing) can produce enormous distributions at modern valuations. The conversation highlights the compounding effect of following on over time and how a single generational infrastructure deal can reshape fund outcomes and LP returns.
Private-market risk: no stop-loss, big pricing risk, and the growth-persistence test
Rory emphasizes the structural difference between public and private markets: in private you can’t reduce exposure if sentiment turns. The group returns to the core underwriting question—can the company grow into a normalized multiple before growth decelerates?
Chamath’s SPAC return: bubble signal or just a flawed structure resurfacing?
They debate whether the return of high-profile SPACs is a reliable top indicator. Rory critiques SPAC incentives and adverse selection; Jason questions why a wealthy sponsor would bother, calling it potentially “edge of grift” relative to effort and reputational risk.
Retail access to PE/secondaries: disclosure vs protection
A tangent becomes a substantive discussion on PE in retail portfolios and opaque markups. Rory frames the regulatory tension: too much protection blocks retail from venture upside; too little invites exploitation through complexity and disclosure games.
OpenAI employee secondaries ($6B): retention, Meta competition, and ‘private markets acting public’
They largely endorse large-scale employee liquidity as rational in a world of intense AI talent bidding (e.g., Meta offers). Jason argues private companies should institutionalize regular liquidity (more like public markets) because the ‘4-year sprint’ is now a 10–15 year marathon.
Founder raises $130M then leaves: ‘expect more of these’ in hot cycles
Harry raises a high-profile founder departure after a big raise; Rory and Jason shrug: failure/turnover is an expected distribution outcome, especially in later-stage venture. Jason adds that hyper-competitive dealmaking (e.g., “Saturday closes”) increases the likelihood of mispriced risk.
Nubank’s $2.5B profit and neobank winners: Nubank vs Revolut vs Chime
Rory zooms out from the quarter to the structural playbook: find weak incumbents, start with a profitable wedge, then expand. Nubank stands out because LatAm incumbents were weakest and Nubank built a full-stack bank (deposits + lending), translating into bank-scale profitability and market cap.
Can Nubank be a $200–400B+ company—or will it ‘trade like a bank’ eventually?
They debate upside versus the gravitational pull of bank valuation frameworks (book value, regulation, credit cycles). Rory’s bullish case depends on continued execution without “dumb bank stuff,” while Harry argues product expansion headroom could be enormous relative to Revolut’s breadth.
CoreWeave’s $11B+ debt: a GPU real-estate financing machine—and the canary for AI capex
Rory calls the debt load unsurprising given massive capex needs; the model depends on matching long-term debt with long-term take-or-pay contracts. Jason frames CoreWeave as the early-warning indicator: if funding tightens or demand wobbles, the leveraged intermediaries break before hyperscalers do.
Will AI spend hit ‘trillions’? Metaphor vs math, and the key unlock: labor-budget substitution
They challenge Sam Altman’s “trillions” claim as potentially metaphorical given balance-sheet constraints—even for Big Tech. The group converges on a condition for trillions to make sense: AI must shift spend from tech budgets into massive human labor budgets; otherwise capex outpaces ROI and financing capacity.
AI tool sprawl → consolidation: why ‘too many $100K agents’ forces platforms and bundles
Jason shares a practical insight: AI tools quickly add up (hundreds of thousands to millions per year), triggering a consolidation cycle faster than SaaS. Rory ties it to a broader rule—multiple tools can’t all claim the same saved headcount—so winners must broaden surface area or get bundled into platforms.
Vertical SaaS vs platform risk: Epic vs Abridge and ‘ignoring classic venture risks’ in an AI boom
Rory argues vertical winners can expand product breadth once they earn trust; the wedge is just the start. On platform risk (e.g., Epic building competing features), Rory says it’s inevitable and must be assumed; Jason notes investors are currently ignoring classic platform-risk caution because AI growth is so compelling.
Kalshi quick-fire predictions: Claude 5 timing, Mistral acquisition odds, Deel vs Rippling IPO race
They close with rapid probability bets: Rory focuses on odds vs likelihood; Jason emphasizes product/revenue realities. The discussion highlights how naming/versioning can be marketing, why sub-scale frontier labs may need to sell, and how profitability/investment mode affects IPO timing.
