newsfilter
← Interviews
20VC with Harry Stebbings1h 15m

Will Jensen Return to NVIDIA | Kalshi, Mercor, Fractile: Mega-Rounds Continue

Will Jensen, Jason Lemkin, Rory O'Driscoll, Harry Stebbings

Roundtable, Podcast
  • NVIDIA Leadership Crisis & Return Prospects

    • Jensen Huang's resignation follows a $601 billion market cap loss, allegations of affairs with employees and customers, and emerging scrutiny over chip export control compliance (specifically regarding transshipment to China via Indonesia/Singapore).
    • Board members (including VCs who backed Huang for 30 years) made a "hasty" decision; the consensus is that a temporary leave of absence would have been a more prudent initial move.
    • Huang has signaled openness to returning, though a full reinstatement faces significant hurdles:
      • The board faces a credibility risk if they reverse the decision, given they are likely unaware of the full extent of the Department of Justice/Commerce investigation into NVIDIA's export practices.
      • Key executives (CFO, Head of Supply, CTO) are implicated in the same compliance issues, meaning their removal would be required alongside Huang's, complicating a potential return.
    • Rory asserts that if he owned 100% of NVIDIA, he would want Huang back, provided there is no threat of government imprisonment; however, the company's ecosystem is "maimed" and unlikely to recover to previous dynamics.
    • Forward-looking impact: A return is possible within days to weeks, but if it does not happen, the company may lose pricing power and aggression in deal-making, shifting the AI landscape toward closed-weight providers.
  • Anthropic IPO & Venture Capital Ecosystem Shifts

    • Anthropic's IPO is projected at $1.4–1.5 trillion, potentially 7x–10x oversubscribed, marking the largest IPO in history.
    • Efficiency Anomaly: Anthropic reached ~$100B revenue run-rate with fewer than 6,000 employees (approx. 1/12th of Google/Meta's headcount at similar revenue levels), signaling a drastic decoupling of revenue growth from human labor.
    • VC Fund Dynamics:
      • LPs are now grading GPs on the number of $20B+ portfolio companies rather than simple unicorns.
      • Early investors in Anthropic (Founders Fund, Menlo) will achieve 8x–9x returns on capital, creating "house money" for future deployment.
      • A trend is emerging of fewer deals (down 21% in Q3 2026) but significantly higher capital deployment (up 2.5x), with deals accelerating into late stages (Series C/D) as pre-revenue valuations skyrocket.
    • Ecosystem Consequence: The NVIDIA drama weakens the "liquidation preference" safety net for AI startups, as potential acquirers (like NVIDIA) may no longer backstop or buy out failing teams, increasing risk for open-weight alternatives like Hugging Face.
  • Investment Committee Decisions & Market Analysis

    • Deal Portfolio: The firm selected Mercor for investment ($200M) due to its leadership in the "dirty work" of data annotation and alignment with AI labs, despite uncertainty in the market.
    • Rejected/Considered Deals:
      • Kalshi: Viewed as a market leader in US sports betting/predictive markets but deemed "hubristic" at a $65B valuation with ongoing lawsuit risks.
      • Fractile: Rejected as a non-value play ($7.5B price) compared to public competitor Cerebras, which trades at a more attractive valuation relative to its revenue ($1B+ booked).
    • Market Trend: The "last round" strategy remains popular; firms are queuing to deploy capital quickly in Q1 2027 to avoid missing the "biggest funds" cycle, despite potential return compression due to rapid markups.
  • SaaS Sector Turnarounds & Founder Returns

    • HubSpot: Reinstated founder Brian Halligan amidst a critical decline in net new customers (dropping from 10,000 to <2,000/quarter). The "refounding" is seen as necessary but risky, as the company may struggle to re-accelerate growth to >30% in a saturated SMB market.
    • Pattern of Founder Returns: Historical precedents (Owen Thomas at Intercom, Dinesh at UiPath, Daniel at MongoDB) suggest returning founders can stabilize companies, but the "limit" of growth in mature SaaS applications may cap upside compared to new AI-native growth.
    • SMB Software Trend: Small businesses are increasingly bypassing traditional SaaS purchases (HubSpot, Monday.com) in favor of building custom internal tools via AI agents or using legacy spreadsheets, reducing the urgency to buy new software licenses.
  • Infrastructure Fragility & Security Risks

    • System Failures: Recent widespread outages (Microsoft/Teams/Outlook, FAA air traffic control) and cyber breaches (Fireworks, Notion, Cursor) highlight extreme fragility in critical infrastructure, driven by "DDoS 2.0" (agent-based load attacks) and configuration errors rather than traditional cyber warfare.
    • Security Implications: The era of "one-person billion-dollar startups" is clashing with the need for S-tier security teams; third-party vendor risks are now a primary concern, potentially consolidating market power among companies with massive security budgets (e.g., Stripe, AWS).
    • Microsoft Outage Context: The 36-hour outage was attributed to load-related bugs and network configuration issues, exacerbated by the decline in senior engineering talent, underscoring the fragility of "AI-reliant" systems.
  • AI Agent Adoption & Strategic Shifts

    • Horizontal Agents: Meta's "Muse" and Instinct are demonstrating the shift from vertical "AI apps" to horizontal, autonomous agents that can manage complex workflows (e.g., email triage, data analysis, booking) without human intervention.
    • Lock-in Risk: Users are reluctant to switch agents once workflows are established, creating high switching costs; "Muse" is noted for its superior headless browser capabilities and deep integration into existing ecosystems.
    • Business Model Impact: Agents may replace traditional marketing automation and CRM software by performing tasks (e.g., identifying and contacting leads) more cheaply than human teams or SaaS subscriptions.
    • Adoption Barrier: The primary challenge is workforce mindset; organizations must transition from "managing agents" to "agents managing humans," potentially leading to organic attrition in roles where AI performance exceeds human output.
  • SpaceX & Starship

    • Starship's successful landing/catching is deemed a "stunning technical achievement" critical for lowering orbital costs.
    • Investment Thesis: The primary driver for investing in SpaceX is not rocketry but the future compute value of Starlink; the stock is essentially a bet on what Anthropic (or competitors) will pay for compute capacity in one year.
    • Market Cap Context: Rocket launches now represent <10% of SpaceX's total market cap, with the majority of value derived from the compute and connectivity ecosystem.
  • Predictions

    • Jensen Huang is predicted to return to NVIDIA within the next three days, contingent on the immediate news cycle dying down before the Anthropic IPO.
    • The AI landscape will see 20+ "Odessa" (agent-based vertical) deals next year, marking the transition from "AI apps" to the "agentic era."
    • The SaaS market will continue to consolidate, with growth rates for legacy players capped while AI-native, compute-intensive firms capture the majority of venture returns.