Episode Summary
Executive Summary: The episode centers on AI investing strategy, venture discipline, and the changing geography of startup ecosystems. Michael Eisenberg and Jaime Mattis argue that proprietary data, vertical integration, and price discipline matter more than foundation models, while Jason Calacanis warns of overfunded AI washouts and echoes late-90s bubble parallels. The group also discusses VC hypocrisy, founder/board governance, and the rise of regional hubs beyond Silicon Valley.
Main Topics: AI investing: vertical apps, proprietary data, and skepticism of foundation models (Priority: 5/5): The panel agrees AI is a major VC wave, but argues the best opportunities are in vertical applications with proprietary data and deep workflow integration rather than commoditized foundation models. Bubble risk and valuation discipline in AI (Priority: 5/5): Jason repeatedly warns that many AI startups are overvalued, comparing the current environment to dot-com era 'eyeball' models and predicting a washout of weak companies. Venture capital role, governance, and the 'hypocrisy' critique (Priority: 4/5): The group debates a tweet criticizing VC behavior, defending board oversight, hard feedback, and the idea that venture is a high-stakes, home-run business rather than a lifestyle job. Emerging manager selection and fund-of-funds strategy (Priority: 4/5): Jaime explains how his firm evaluates managers based on unique deal flow, hustle, team quality, alignment, and fund size, preferring smaller, more aligned early-stage funds. Geography of startup ecosystems and regional hubs (Priority: 4/5): The conversation covers why the Bay Area still leads, why New York dominates fintech, and how Israel, Abu Dhabi, Dubai, and other regions are becoming more important global hubs. Work ethic, founder feedback, and the importance of in-person density (Priority: 3/5): Speakers stress intense effort, responsiveness, and physical proximity as key ingredients for building excellent companies and maintaining optimism in the ecosystem.
Key Arguments: AI value will accrue to companies with proprietary data, not just access to general models or compute. Vertical AI businesses win when they embed deeply into workflows and have high switching costs. Foundation models are becoming commoditized; startup moats will shift to data, product, and distribution. Many AI companies raising at huge valuations will not justify those prices and may collapse or be repriced. VCs should be judged by LPs and founders on work ethic, governance, and ability to provide candid feedback. Emerging managers succeed when they have unique sourcing, strong teams, and small enough funds to stay aligned with early-stage outcomes. Geography still matters because density, capital networks, and repeated interactions increase the odds of breakout companies. Remote work did not replace the value of in-person startup ecosystems; office proximity and local intensity still matter. The current venture environment has too much capital and too many funds, leading to weak discipline and weaker returns. Good founders often welcome hard feedback, and the best investors build trust before giving it.
Data Points: AI vertical applications capital raised in 2023: $27.8 billion - Pitchbook data cited at the start of the AI segment. AI horizontal platforms capital raised in 2023: $27.1 billion - Pitchbook data cited alongside vertical applications. Autonomous machines capital raised in 2023: $3 billion - Pitchbook breakdown of AI categories. AI and ML semiconductors capital raised in 2023: $2.4 billion - Pitchbook breakdown of AI categories. Conference attendance: ~125 people - The Napa LP/GP conference was intended for 100 but expanded due to late demand. Conference duration: 72 hours - The Napa event was described as a highly productive Sunday night through Wednesday brunch gathering. GP/LP ecosystem exposure: 600+ startups - Jaime described Invariantes' portfolio/fund-of-funds exposure. AI-focused exposure from manager strategy: 100-120 startups - Jaime estimated indirect/direct exposure to AI-related startups. West Coast share of investments: 65% - Jaime said most of his U.S. deployment goes to the West Coast. East Coast share of investments: 25% - Jaime’s geographic allocation. Rest of U.S. share of investments: 5%-10% - Jaime’s geographic allocation. U.S. exit unicorn concentration in California: 60% - Used to justify Bay Area concentration. U.S. exit unicorn concentration in California + Boston + New York: 77% - Used to support investing in top ecosystems. Managers analyzed per year: 200-300 - Jaime’s fund-of-funds manager selection pipeline. Preferred fund size: $30 million to $150 million - Jaime said smaller, aligned emerging managers are preferred over mega-funds. Athena investment: Largest investment of Jason's year - Jason cited Athena as a major bet on human-in-the-loop AI-enabled outsourcing. Athena virtual assistant talent pool: Top 0.1% in Manila - Jason described Athena’s sourcing model. OpenAI consumer price point referenced: $20/month - Jason argued such pricing may be unsustainable if AI becomes bundled by platforms. OpenAI customer example referenced: 350 paying customers - Used by Jason to critique valuation versus revenue in another company discussion. First AI investment year at Aleph: 2013-2014 - Michael said his firm began investing in AI early. War-related disruption to investing cadence: Post-Oct. 7 to mid-February - Michael described continuing investments in Israel during wartime disruption. Number of investments made after Oct. 7: 6 investments - Michael said Aleph invested despite the war.
Pivotal Quotes: "I think what is new in the world is actually not AI, but LLMs." — Michael Eisenberg: He distinguishes long-running applied AI from the newer LLM wave. "AI will be coming down to zero. And then it is a game of data and features and focus around the customer experience." — Jason Calacanis: Jason argues models will commoditize and startups need real moats. "Optimism is a weapon of mass creation." — Michael Eisenberg: He links in-person ecosystems and optimism to startup formation and success.
Implications: Investors should prioritize moats, pricing discipline, and real revenue over hype. Founders may still need to be near dense ecosystems, but capital is increasingly global. Expect AI consolidation, platform bundling, and a sharp reset for overvalued startups.
About This Week in Startups
Jason Calacanis covers startups, tech, markets, media, and all the hottest topics in business and technology. He also interviews the world’s greatest founders, operators, investors, and innovators.