“The best founders find a way to make it happen”: VC Roundtable with Bryan Kim and David Clark | E2222

Episode Summary

Executive Summary: The panel argued that AI is not in a classic bubble, though it is frothy, because demand, revenue, and adoption are rising fast enough to support high valuations. They discussed investing in AI learning startup Oboe, the Boom Supersonic pivot into data-center power, valuation and ownership in venture, exits and secondary liquidity, gross margins in AI apps, and why AI-native products and distribution models are still emerging.

Main Topics: Is AI in a bubble? (Priority: 5/5): David Clark argued AI doesn’t fit the historical definition of a bubble because valuations are supported by rapid revenue growth and adoption, while Brian Kim agreed demand is real and usage is massive. The group framed the market as early-cycle froth rather than a true bubble. Oboe investment thesis and AI learning (Priority: 5/5): Brian explained why Andreessen Horowitz led Oboe’s $16M Series A: AI can materially improve education, transform information into actual learning, and create a giant consumer learning platform. He highlighted founder quality, product speed, and a broad learning TAM. Valuation, ownership, and venture returns (Priority: 4/5): The panel debated how AI company growth affects valuation. David emphasized fund-level ownership and the need for one company to return a fund; Jason discussed target ownership levels and the power-law nature of venture outcomes. Exit timing, DPI, and secondary liquidity (Priority: 4/5): David argued managers should not be pressured to sell winners early and instead should manage liquidity through portfolio construction and growth funds. He noted rising LP secondary activity as a practical solution to venture illiquidity. Boom Supersonic’s data-center power pivot (Priority: 4/5): The group discussed Boom turning its jet-engine tech toward power generation for data centers. They treated the move as a pragmatic way to fund the original mission while addressing global power shortages for AI infrastructure. AI margins, go-to-market, and AI-native products (Priority: 5/5): Brian and David discussed that gross margins may be lower in percentage terms but still attractive in dollars, especially as AI replaces labor. They also argued that distribution and product velocity are now central moats, and true AI-native UI is still emerging. Future expectations: IPOs, M&A, and market reset (Priority: 3/5): The panel predicted one or more mega-cap private companies could go public in 2026, with potential IPOs serving as a reality check on private valuations. Jason also forecast a golden age of M&A and more startup formation due to weak hiring.

Key Arguments: AI is not matching the historical signs of a bubble; valuation is being driven by real adoption, revenue, and growth rather than pure speculation. The AI market is frothy, but the simultaneous buildout of infrastructure and deployment of products makes this cycle different from prior bubbles. Oboe is attractive because it uses AI to convert learning into a personalized, practical outcome rather than just delivering information. Strong founders plus fast product shipping and distribution are the real moats in AI, more than static product craft at this stage. Venture returns still depend on ownership and the ability of one company to return the fund, so valuation must be judged in that context. Managers should not force sales of top performers; liquidity should be managed through a mix of early-stage, growth, and secondary allocations. AI startups may have lower gross-margin percentages, but gross-margin dollars can still scale dramatically as products expand. The best AI companies are increasingly capital-efficient despite heavy AI and infrastructure spend because growth and margins compound together. Boom’s pivot reflects both necessity and opportunity, showing that hardware startups often need adjacent revenue lines to survive until the core vision matures. AI-native products will likely move beyond skewomorphic use cases into new interfaces and workflows that are not yet obvious. High private valuations will be better tested if major private companies like OpenAI, Anthropic, SpaceX, or Databricks go public. Weak hiring of new graduates may push more young people to start companies, increasing entrepreneurial formation.

Data Points: Oboe round size: $16 million - Brian Kim described the Series A lead investment into Oboe Oboe valuation context: Competitive round with at least 1-2 term sheets - Brian said the round was competitive and the founder had other options AI weekly usage: 800 million weekly active users - Brian cited ChatGPT usage as evidence of strong demand ChatGPT usage frequency: 25+ uses per week per weekly active user - Used to support the argument that AI adoption is intense and habitual Databricks revenue run rate: Over $4 billion - David referenced Databricks’ scale as proof that the best companies keep compounding Databricks AI revenue run rate: Greater than $1 billion - Highlighted as evidence that AI products are becoming material businesses Databricks growth rate: More than 50% year over year - Used to show sustained growth at large scale Figure valuation: $39 billion - Jason raised Figure as a debated example of a valuation that may be ahead of revenue Fal.ai Series D: $140 million - Example of a fast-growing AI company raising again soon after a prior round Fal.ai valuation: $4.5 billion - Reported valuation associated with the Series D raise Booi/Boom pre-orders: $1.25 billion - Jason cited pre-orders for Boom’s product as evidence of real demand Boom new funding: $300 million - Announced alongside the pivot into data-center power Launch accelerator ownership target: 7% - Jason said accelerators typically target this initial ownership level Expected dilution for accelerators: 50% to 70% - Jason described typical dilution through follow-on rounds Fund return target: 3x - David said they underwrite early-stage funds to roughly 3x Outlier fund performance: 5x+ once every 3 or 4 funds - David described the expected power-law pattern for top managers Estimated hold period for best companies: 15 years or more - David explained IPO timing plus post-IPO sell-down periods AI market size proxy: $50 trillion to $60 trillion annualized - David framed labor as the addressable market AI could potentially capture Probability of at least one top private company IPO in 2026: 80% to 95% - Panel estimates varied by speaker, with Jason the most bullish Probability of AI-native applications emerging: 80%+ - Brian’s estimate for meaningful AI-native UI/app emergence

Pivotal Quotes: "We are at the start of a major cycle here. We are, as things stand today, we are not in a bubble." — David Clark: David summarized the panel’s core view on the AI market "Every marketing channel sucks right now." — Brian Kim: Brian describing the need for new AI-native go-to-market approaches "The problem AI solves? It solves wages." — Jason Calacanis: Jason’s shorthand for AI’s economic impact on labor costs

Implications: Listeners should expect continued AI investment, higher scrutiny of valuations, more LP secondary activity, and eventual public-market tests of private AI leaders. The next wave likely rewards fast-shipping AI-native products, strong distribution, and founders who can convert labor replacement into durable margins.