20VC

Harry Stebbings × Rory

SpaceX’s all-stock Cursor takeover, completed within a week and carrying a breakup fee, highlights AI M&A’s currency and clearance advantages while testing premium pricing against forward revenue.Stripe’s OpenRouter bet monetizes routing complexity but faces model drift, while Workday’s take-private underscores system-of-record durability without guaranteed upsell; AI-budget ceilings and agentic roadmaps remain key risks.

Harry Stebbings

Anthropic’s filing alongside a $65B raise and Cursor’s potential $60B exit reset startup scale and public-market timing.Token budgets are the battleground: Uber capped spend at $1,500/month, agent-attached software rose, and classic per-seat SaaS weakened.Watch whether tokens reach 33% of engineering compensation, or whether capex intensity and PE software returns constrain the upside.

Harry Stebbings × Sebastian Siemiatkowski

Siemiatkowski argues that near-zero software creation costs and one-click data migration could push threatened SaaS revenue multiples from 5–10x toward utility-like 1–2x, while enterprises reorganize around secure Lego-like components and agents.Klarna’s workforce fell from more than 7,000 to fewer than 3,000, with fewer than 2,000 possible by 2030, but whether AI-driven information compression reduces compute demand or consumer generation overwhelms those savings remains unresolved.

Harry Stebbings

Brex’s $5.15B sale to Capital One resets the comparison for Ramp’s roughly $32B mark: a real-money acquisition applied about seven times revenue to Brex’s roughly $700M, while Discover’s closed network gives Capital One a structural interchange advantage.Anthropic’s inference costs ran 23% above expectations even as gross margin improved to roughly 40%, leaving mid-tier SaaS economics and AI-capex sustainability unresolved.

Harry Stebbings

SpaceX combines exceptional rocket and Starlink businesses with roughly $15 billion of revenue growing about 30%, yet its $800 billion mark exceeds 40x run-rate revenue.The valuation embeds an Elon premium and leaves downside if growth decelerates, while 2026 IPOs from SpaceX, Anthropic, and Databricks could unlock liquidity without fixing private-market overpricing.

Harry Stebbings

OpenAI’s restructuring removes its corporate obstacle to raising another $100–200 billion and moves a potential IPO one enormous step closer, while Microsoft retains IP rights, revenue share, and prospective Azure business. a16z’s $10 billion scale supports optioning and follow-ons, but Mercor’s $350 million round at a $10 billion valuation depends on three to five more years of AI CapEx hypergrowth despite pass-through margins and two customers reportedly exceeding 50% of revenue.

Harry Stebbings

Databricks at $100 billion looks relatively inexpensive beside Snowflake because both have roughly a $4 billion run rate, but Databricks grows 50% versus Snowflake’s 26%.The valuation works only if growth persists for several years, while a coming IPO wave could release venture returns without preserving today’s scarcity premiums.Nubank’s $2.5 billion net income and 123 million customers show the upside of replacing the incumbent bank, whereas CoreWeave’s $11.2 billion debt remains dependent on matched customer commitments and AI demand.

Harry Stebbings × Vlad Tenev

Robinhood’s $35B-to-$85B re-rating reflects a strategy finally gaining credibility, with active-trader focus and near-100% engineering AI adoption reshaping its operating model.Tenev’s expansion thesis is tokenization without issuer opt-in, solving retail access’s adverse-selection problem as 200+ EU stock tokens move toward 24/7 trading, self-custody, and DeFi, with US regulation unresolved.

Harry Stebbings × Kieran Flanagan

NFDG's Meta deal shows AI talent repricing venture careers, while a $200M-ARR B2B company may now give each AI engineer 0.5–1% of the company.OpenAI's $4.4B stock compensation and Meta's hiring spree point to severe SBC dilution for venture returns, as CoreWeave uses equity to reduce leasing exposure and Olo's $2B take-private sets a tougher comp for non-AI unicorns.

Harry Stebbings × Kim Graves

Meta’s roughly $100 billion AI offensive is insurance against ChatGPT replacing social-media attention, while scarce researchers who were “in the room when the magic happened” command extraordinary prices as institutional loyalty weakens.Harvey’s $300 million round at a $5 billion valuation depends on capturing legal labor economics, not ordinary seat revenue, and Circle’s 57-times run-rate valuation and 46.4% five-day rise illustrate speculative price discovery even as cash-generative companies can fund growth privately.