Anthropic's Mythos: The AI Too Dangerous to Release

The Startup Exit Math Every Founder Should Know | Peter Walker, Head of Insights at Carta

Founders are constantly told to “think bigger,” raise more capital, and chase billion-dollar outcomes. But behind the headlines about AI unicorns and explosive growth lies a much harsher reality: most startups never IPO, most founders end up heavily diluted, and even successful venture-backed companies can take a decade or more to reach an outcome. In this episode of the Post Money Podcast, Nilanjana Bhowmik sits down with Peter Walker, Head of Insights at Carta, to unpack the actual numbers behind startup fundraising, dilution, exits, secondary sales, and the modern venture market. Drawing from data across tens of thousands of startups on Carta, Peter explains why AI companies are pulling away from traditional SaaS businesses, why many founders dramatically underestimate how long startup journeys really take, and why valuation alone often means far less than founders think. The conversation explores how venture capital has become increasingly concentrated around a tiny group of AI companies, why secondaries only benefit a very small slice of startups, and how most venture-backed exits still happen through acquisitions rather than IPOs. Inside the episode: * The real fundraising benchmarks founders need to hit in 2026 * Why “zero to $100M ARR” is distorting startup expectations * The reality behind startup valuations and dilution * The brutal math of venture capital ownership over time * Why down rounds are far more common than founders think * Why profitability no longer guarantees a successful exit * The truth about unicorns, secondary sales, and liquidity * Why most startups will never IPO * How raising too much capital can reduce exit opportunities * What founders should understand before taking venture capital at all Watch Full Episode on YouTube: About Peter Walker: Peter Walker leads the Insights team at Carta, where he analyzes one of the largest private startup datasets in the world to uncover trends shaping venture capital, fundraising, startup valuations, dilution, secondaries, exits, and the broader technology ecosystem. Over the past several years, Peter has become one of the most widely followed voices in venture capital data and startup market analysis. His research and visualizations are regularly referenced by founders, investors, operators, and financial media seeking a clearer understanding of how private markets actually function beneath the headlines. Before leading Insights at Carta, Peter worked across startup operations, venture analysis, and research roles focused on technology companies and private market ecosystems. Connect with Peter Walker: https://www.linkedin.com/in/peterjameswalker/ [https://www.linkedin.com/in/peterjameswalker/] Connect with Nilanjana Bhowmik: linkedin.com/in/nilanjanabhowmik [https://www.linkedin.com/in/nilanjanabhowmik] Converge VC website: converge.vc [https://converge.vc/] This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit postmoneypodcast.substack.com [https://postmoneypodcast.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

Investing in Civilization Scale problems at Lightspeed | Raviraj Jain

Inside top venture firms, investors are now funding frontier AI research labs before products exist, before revenue exists, and sometimes before there is even a clear path to monetization. The bet is no longer just on software companies, it’s on entirely new forms of intelligence. In this episode of the Post Money Podcast, Raviraj Jain from Lightspeed Venture Partners explains why some of Silicon Valley’s biggest capital pools are shifting toward self-learning systems, humanoid robotics, fusion energy, and AI infrastructure that could redefine how humans work, manufacture, discover new technologies, and generate power over the next decade. Raviraj breaks down why many investors now believe the next leap in AI will happen outside the digital world. Not in chat interfaces or productivity tools, but in machines that can operate in physical environments, understand motion, adapt to unfamiliar situations, and interact with the real world autonomously. He explains why robotics foundation models, reinforcement learning, and multimodal systems are advancing faster than most people realize, and why humanoid robots may arrive much sooner than the public expects. The conversation also gets into the hidden constraints behind the AI boom itself. As data centers become one of the fastest-growing consumers of global energy, technologies once considered unrealistic, including nuclear fusion, are suddenly becoming serious venture-backed bets again. The real question is no longer whether AI will change industries. It’s who will move fast enough to rebuild them before someone else does. Inside the episode: * Why venture firms are now funding pure AI research labs before monetization exists * The case for self-learning AI systems over human-trained models * Why humanoid robots may arrive faster than most people expect * The biggest bottleneck facing AI over the next decade: energy * Why nuclear fusion is suddenly investable again after decades of skepticism * The shift from consumer AI tools to enterprise-wide AI adoption * What early-stage investors now look for in founders during the AI era * Why domain expertise may soon outperform technical expertise again * How AI could completely reshape venture capital workflows and decision-making Watch Full Episode On YouTube About Raviraj Jain: Raviraj Jain is a Partner at Lightspeed Venture Partners, where he focuses on enterprise software, AI, cloud infrastructure, machine intelligence, and frontier technologies. Before joining Lightspeed in 2017, he worked across startups and large technology companies in both the US and India, including roles in product and strategy at LinkedIn and Fundbox. Having worked as an entrepreneur, operator, and investor, Raviraj brings a highly product-focused lens to early-stage investing. Raviraj holds an MBA from Harvard Business School and an engineering degree from IIT Bombay. Connect with Raviraj Jain: https://www.linkedin.com/in/ravirajjain/ [https://www.linkedin.com/in/ravirajjain/] Connect with Nilanjana Bhowmik: linkedin.com/in/nilanjanabhowmik [https://www.linkedin.com/in/nilanjanabhowmik] Converge VC website: converge.vc [https://converge.vc/] This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit postmoneypodcast.substack.com [https://postmoneypodcast.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

The Navy Veteran Building America's AI Ships | Rylan Hamilton

Autonomous ships, AI-powered naval systems, and unmanned fleets are quickly becoming one of the most important technology shifts inside the US defense ecosystem, driven by a reality that is reshaping how the US thinks about maritime power and industrial capacity. China now builds half of all the ships in the entire world. Their Navy is roughly the same size as the US Navy. The US cannot out-build them on destroyers and carriers, so the answer is out-innovating them with autonomous systems built fast, built cheap, and built in mass. The Navy has $2 billion to spend on this program this year alone. And they plan to buy every single year for the next 30 years. In this episode, Rylan Hamilton, Co-Founder & CEO of Blue Water Autonomy, explains why the future of maritime technology will look radically different from the last 50 years and why the United States is betting billions on autonomous systems that can operate for months at sea without crews onboard. Before starting Blue Water, Rylan served as a Navy surface warfare officer, helped scale Amazon Robotics after joining Kiva Systems, and later co-founded Six River Systems, which was acquired by Shopify for $450 million. Now, he is applying lessons from robotics, automation, manufacturing, and large-scale hardware systems to one of the hardest engineering environments imaginable: the open ocean. Inside the episode: * How a warehouse roboticist ends up building the Navy’s most important unmanned platform * What makes the Liberty class different from every other ship in the water today * The “Walmart analogy”, or why the Navy is the best customer Rylan has ever sold to * How Blue Water raised from Eclipse and Google Ventures and why they’re racing toward Series B * The supply chain reality: why the US already has everything it needs to build hundreds of these ships * Why autonomous ships are being treated as a strategic priority * How AI-powered vessels can operate for months without crews * The hidden engineering challenges of saltwater, sea spray, and autonomous navigation * Why the Navy is becoming one of the biggest buyers for venture-backed startups Watch Full Episode on YouTube: About Rylan Hamilton: Rylan Hamilton is the Co-Founder and CEO of Blue Water Autonomy, a defense tech company building autonomous surface vessels for the US Navy. After leaving the Navy, Rylan joined Kiva Systems, which became Amazon Robotics and remains one of the largest robotics deployments on the planet, before co-founding Six River Systems, a mobile robotics company for warehouse automation. Six River was backed by top-tier VCs and acquired by Shopify for $450 million four years after founding. Blue Water has raised from Eclipse Ventures and Google Ventures, and is currently building the Liberty class full-scale vessel at a shipyard in Louisiana, with a launch planned for end of 2026. Connect with Rylan Hamilton: linkedin.com/in/rylan-hamilton [https://www.linkedin.com/in/rylan-hamilton] Connect with Nilanjana Bhowmik: linkedin.com/in/nilanjanabhowmik [https://www.linkedin.com/in/nilanjanabhowmik] Converge VC website: converge.vc [https://converge.vc] This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit postmoneypodcast.substack.com [https://postmoneypodcast.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

Uber, PillPack, Coupang First-Check VC: The One Rule He Never Breaks

David Frankel wrote some of the earliest checks into companies like Uber, Coupang, PillPack, Whoop, Shield AI and Suno, yet his firm, Founder Collective, still refuses to lead Series A rounds or aggressively double down on portfolio companies. In this conversation, he explains why too much funding can quietly destroy startups, why founder-investor alignment matters more than ownership percentage, and how venture capital remains wildly inefficient even in the AI era. Before becoming one of the most respected seed investors in venture capital, David built the first commercial internet service provider in South Africa during the earliest days of the internet boom. What followed was a crash course in hypergrowth, acquisitions, global expansion, and eventually a massive exit that left him unexpectedly depressed and convinced he had sold too early. That experience became the foundation for how he thinks about founders, risk, ambition, and investing today. This conversation goes deep into the realities of building venture-backed companies: the psychology of founders, the hidden dangers of scaling too fast, the pressure of raising capital, and why startup life often looks far more glamorous from the outside than it feels from the inside. David breaks down Founder Collective’s famous “first-check only” strategy, why the firm never negotiates with founders twice, and how some of the biggest startup winners looked before the rest of the market noticed them. Inside the episode: * Why Founder Collective backed Uber, Coupang, PillPack, Shield AI and Suno while refusing to lead later rounds * The “first-check only” strategy that almost no VC firm copied * The difference between raising money for product-market fit vs market domination * Why some billion-dollar companies needed shockingly little capital in the early days * The founder test David uses before writing a check: “green button or red button?” * The real reason venture markets are still highly inefficient * What Founder Collective looks for in AI founders in 2026 * The only kind of failure founders truly cannot come back from * How secondary markets changed venture capital forever Watch Full Episode on YouTube: About David Frankel: David Frankel is the co-founder and Managing Partner of Founder Collective, a seed-stage VC fund whose mission is to build the most aligned fund for founders at the seed stage. An immigrant who got his entrepreneurial start selling airbrushes at a South African swap meet, David went on to co-found Internet Solutions, Africa's first and largest ISP, which he sold to NTT. In 2000, he was voted the South African Technology Achiever of the Century.Post a Fulbright Foreign Scholarship and an MBA with distinction from Harvard, David moved back to the USA. He provided the first capital for numerous companies, including Olo (NYSE: OLO), where he worked closely with founder Noah Glass through its IPO and acquisition in 2025 by Thoma Bravo. At Founder Collective, he is the lead investor in Coupang (NYSE: CPNG), SeatGeek, PillPack (acquired by Amazon), Shield AI, and more recently, Suno. Founder Collective's broader portfolio includes Uber (NYSE: UBER) and The Trade Desk (NYSE: TTD). David was ranked #2 on the 2025 Midas Seed List and has appeared on the Midas List of the world's best venture capital investors seven times.David is also the co-founder of NextUp, a youth employment accelerator in South Africa, which recently partnered with educator Taddy Blecher to launch the Maharishi NextUp Institute of Technology (MNIT) in Johannesburg, training underprivileged youth in AI, robotics, and cybersecurity.Connect with David Franklin: https://www.linkedin.com/in/davidafrankel1/Connect with Nilanjana Bhowmik: https://www.linkedin.com/in/nilanjanabhowmik/ Converge VC website: https://converge.vc/ This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit postmoneypodcast.substack.com [https://postmoneypodcast.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

Anthropic's Mythos: The AI Too Dangerous to Release

Anthropic just drew a line in the sand. Their latest frontier model MYTHOS is so capable at autonomously discovering and exploiting cybersecurity vulnerabilities that they’ve refused to release it to the public. Instead, it’s being quietly handed to a handful of giants like Microsoft and JP Morgan Chase and top government officials, including the Treasury Secretary and the Fed Chair have already been briefed. In this episode, James Falkoff joins Nilanjana Bhowmik to break down exactly what MYTHOS is, why it’s different from every AI model before it, and why the implications stretch far beyond cybersecurity. Inside the episode: * Why MYTHOS crossed the line that Opus 4.6 didn’t * Why MYTHOS can reverse-engineer closed-source code (including proprietary systems like Windows) to surface vulnerabilities no human ever caught * The shift from human-speed cyberattack to machine-speed and why defense must now match that pace * Why Anthropic chose to restrict access instead of releasing publicly * Why the government took the warning seriously and whether the market did too * The equity crisis hiding inside AI model access * What MYTHOS signals about where open-source models are heading * The growing concern: unequal access to powerful AI systems * Opus 4.7 and the “lazy model” debate * Compute scarcity as the hidden constraint shaping AI progress Watch Full Episode on YouTube About James Falkoff: James has over 16 years of experience investing in technology across public and private markets, including more than a decade in venture capital. He has led over 25 early-stage investments and is known for his thought leadership, strong exit track record, and hands-on work with portfolio companies to drive growth. Before joining Converge, he was a Principal at Longworth Venture Partners, where he partnered with Nilanjana Bhowmik to identify high-potential sectors and secure early access to top startups. His portfolio includes companies like Olapic (acquired by Monotype), RapidMiner (Altair), JIBE (iCIMS), and TrackVia (Primus Capital). Connect with James Falkoff:https://www.linkedin.com/in/jamesfalkoff/ [https://www.linkedin.com/in/jamesfalkoff/] Connect with Nilanjana Bhowmik:https://www.linkedin.com/in/nilanjanabhowmik/ [https://www.linkedin.com/in/nilanjanabhowmik/] Converge VC Website: https://converge.vc/ [https://converge.vc/] This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit postmoneypodcast.substack.com [https://postmoneypodcast.substack.com?utm_medium=podcast&utm_campaign=CTA_1]