The Twenty Minute VCOpenAI Restructuring: Who Wins and Who Loses & Mercor Raises $350M at a $10BN Valuation
CHAPTERS
- 0:00 – 1:11
Setting the stage: OpenAI, mega-funds, and what “returns” really mean
The hosts open with quick banter that tees up a packed agenda: OpenAI’s restructuring, mega-fund dynamics, breakout AI companies, and how investors should think about outcomes. Even in the jokes, the core tension emerges: hype vs. fundamentals, and governance vs. growth.
- •Framing the episode as a week-in-review for venture and AI markets
- •Early signals of debate on valuation step-ups vs. “real” up-rounds
- •Preview of topics: OpenAI structure, a16z funds, Mercor, spray-and-pray, Amazon, and IRR
- 1:11 – 5:04
OpenAI’s restructuring: what changed and why it matters
Rory explains OpenAI’s deal-making with Microsoft and state attorneys general to enable a cleaner corporate structure. The move removes the “donation-like” constraints of the old setup and makes future fundraising—and even an IPO—far more feasible.
- •Deals with Microsoft + Delaware/California AGs enable restructuring
- •New structure: a charitable foundation on top; operating company as a PBC
- •Restructuring removes the old “this could go to zero” donation framing
- •Major implication: OpenAI can now raise capital and plausibly go public
- 5:04 – 10:06
Who wins and who loses in the OpenAI deal (and Sam Altman’s no-equity anomaly)
The group breaks down the winners in the settlement and why outcomes largely reflect leverage and risk-taking. A standout controversy: Sam Altman reportedly still holds no shares, which is both unprecedented and strategically interesting.
- •Microsoft’s position: ~27% ownership, strong rights, strong return profile
- •Nonprofit/foundation becomes massively capitalized (framed as a “win”)
- •Employees and investors gain clearer liquidity paths
- •Altman holding no equity is highlighted as unusual and power-shaping
- •Limited “losers”: critics of for-profit shift and Elon Musk’s objections
- 10:06 – 14:50
IPO math, valuation euphoria, and SoftBank’s two-price reality
They explore how restructuring unlocks access to dramatically larger pools of capital, including public markets. Rory notes bizarre near-term pricing dynamics where the same security appears to be transacting at different implied valuations via primary vs. secondary activity.
- •Potential retail IPO demand could be historically massive
- •Valuation trajectory discussion: 500B now, possible trillion-plus on trajectory
- •SoftBank dynamics: primary investment vs. employee secondary at different prices
- •Instant markups create “IRR in an hour” optics for some investors
- •Broader implication: counterparties (Oracle, others) gain confidence OpenAI can fund commitments
- 14:50 – 16:12
Second-order OpenAI product bets: the browser and ‘VC-made’ narratives
They briefly touch on OpenAI’s rumored/expected browser and question whether users will truly change behavior. The discussion emphasizes how the tech narrative machine can overstate product impact before real adoption is visible.
- •Skepticism that a new browser is automatically a consumer game-changer
- •Comparison to other hyped launches that didn’t reshape behavior
- •Tension between land-grab strategy vs. genuine product breakout
- •Focus on what average consumers will actually do vs. what VCs amplify
- 16:12 – 18:38
Andreessen Horowitz raises $10B: the mega-platform model becomes normal
The hosts analyze a16z’s $10B raise and the broader trend toward large “platform” venture firms that resemble financial institutions. They debate whether the fund sizes are as intimidating as headlines suggest once broken into sub-funds.
- •$10B split: growth, AI apps, AI infra, defense; viewed as structurally intentional
- •Rory: this is the dominant venture modality today; could become ‘Goldman/JPM’ style market
- •Jason: sub-fund sizing makes competition at seed/A less hopeless than headlines imply
- •Human capital management: fund ‘fiefdoms’ help retain senior partners
- 18:38 – 26:22
How mega-funds win deals: option checks, media leverage, and LP bundling
They debate the mechanisms that large firms use to win—beyond just brand—including paying “option prices” early and creating constant momentum through media. The conversation also highlights LP “stapling” and bundling as a structural advantage for top firms.
- •Option-check strategy: overpay early to secure positioning for later rounds
- •Media presence (‘wall of sound/news’) as a real competitive moat
- •Disagreement on whether smaller players can replicate momentum via participation and coverage
- •LP bundling/stapling: access to early funds often requires commitments across multiple vehicles
- •Quantity as a weapon: AUM enables more shots, more services, more presence
- 26:22 – 37:18
Mercor at $10B: the new RLHF labor stack powering frontier models
Mercor’s explosive growth is framed as a consequence of frontier-model spending shifting from ‘labeling cats’ to high-skill human feedback. They explain the business: sourcing specialized experts to produce training feedback, and why buyer concentration is both a feature and a risk.
- •Mercor provides expert humans for RLHF (PhDs, specialists) to train models
- •Market evolution: from basic labeling to frontier-domain human expertise
- •Demand driven by massive model-company CapEx and urgency
- •Key risk: extreme customer concentration (a few buyers drive most revenue)
- •Key risk: margin profile constrained by high payouts to expert labor
- 37:18 – 44:48
Underwriting AI services at massive valuations: growth vs. durability (plus Ramp’s funding cadence)
They wrestle with what investors must believe to justify high valuations in fast-scaling AI-adjacent businesses. The conversation then expands to Ramp, debating whether frequent rounds are about capital need, signaling, or market dynamics—and whether small step-ups matter after dilution.
- •What a $10B price implies: need for enormous future revenue to justify venture returns
- •Two underwriting modes: long-term terminal value vs. short-term growth/multiple hold
- •‘AI CapEx hypergrowth’ as the core bet that makes high prices pencil
- •Ramp: capital intensity (financing balance sheet) may justify frequent raises
- •Jason: modest step-ups (e.g., 23B→30B) can be underwhelming after dilution
- 44:48 – 1:08:04
Spray-and-pray vs. concentrated picking: Carta data, options, SAFEs, and exit decisions
Using Carta’s distribution data on Series B outcomes, Rory argues the takeaway isn’t that spraying works—it’s that picking still dominates unless you have an options-based model at scale. The discussion turns practical: founder exit choices (Synthesia/Adobe), board burden, and why SAFEs feel ‘lighter’ for investors.
- •Carta dataset: outcome distribution for 2018 Series B deals (1X/2X/5X/10X tail)
- •Rory: the math only works if you avoid over-indexing into sub-2X outcomes
- •Three strategies: picking, spraying, and optioning (multi-stage funds can ‘option’ earlier)
- •Founder vs. VC incentives: 3B→10B matters more to funds than to founders personally
- •Board/late-stage burden and why some investors prefer SAFEs’ lower commitment
- 1:08:04 – 1:13:34
When regulators block outcomes: Roomba, M&A delays, and the cost of antitrust drag
Roomba becomes a case study in how a blocked acquisition can cascade into financial distress. They generalize the point: even “great” M&A multiples can erode if closing takes 12–24 months under antitrust review, changing founder calculus on whether to sell.
- •Roomba/Amazon deal blocked; debt bridge and deteriorating outlook highlight fragility
- •Argument that the decision harmed shareholders and company survival prospects
- •Antitrust timelines can turn a ‘30x revenue’ deal into a far lower effective multiple
- •Extended review risk makes IPO-or-bust more common once acquisition certainty disappears
- •Acquihire vs. full-product acquisition: different incentives and outcomes
- 1:13:34 – 1:19:06
Amazon’s AI moment: layoffs, AWS relevance, and whether founders need to return
They assess Amazon’s rough period—layoffs, AWS share pressure, outages—and debate whether Bezos stepping aside before the AI inflection was a strategic mistake. The core issue they land on: hyperscalers must be relevant in AI compute, either via partnerships or compelling owned offerings.
- •Amazon layoffs framed as part cost reset, part strategic pressure
- •AWS risk: new compute demand is AI-heavy; Amazon seen as behind vs. Microsoft/Google
- •Founder leadership debate: Bezos’ exit timing vs. Sergey Brin’s return at Google
- •Partnership gaps: Anthropic relationship noted, but compared against rivals’ positioning
- •Warning against overreacting: retail logistics still strong; cloud AI relevance is the key challenge
- 1:19:06 – 1:26:49
Lightning round: Brax vs. Ramp, a16z’s “best platform” case, and Anduril as a private-market darling
In agree/disagree, they debate value vs. growth tradeoffs (Brax vs. Ramp), whether a16z deserves top mega-platform honors, and which late-stage private companies are most desirable to own. The segment also surfaces personal investing preferences (interest alignment, social signaling, ethics).
- •Brax vs. Ramp framed as a valuation-multiple vs. growth-rate indifference problem
- •a16z praised for operational execution and scaling venture beyond old constraints
- •Discussion of partner churn vs. platform durability in mega-firms
- •Anduril prompts a candid split between “best brag” asset and personal values
- •Closing reflections on what it means to be a ‘top two’ choice for founders
