finance
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.
GrokBot users spent the weekend complaining about burning through their weekly usage limits in a matter of days. Elon Musk responded on X, stating that the cost of serving GrokBot was unsustainably high relative to revenue.
During a SpaceX earnings call, SpaceX President Gwynne Shotwell revealed that GrokBot reached 800,000 paid weekly users, more than doubling in a single month.
Oliver Cameron, co-founder and CEO of Odyssey, stated during an interview that he is 90% certain that humanity lives in a very advanced simulation.
Oliver Cameron shared that Odyssey has raised $370 million from investors including Amazon, AMD, NVIDIA, and GB to build world models for robots and agents.
Google launched a prototype satellite from Project Sunchatcher into low Earth orbit via a SpaceX Falcon 9 rocket, carrying four Google TPU AI chips to test data center operations in space.
Amanda Bradford explained how she used non-scalable guerrilla marketing tactics, offline events, and personal waitlist recruitment as a solo founder to pre-game the San Francisco launch of dating app The League before the app was built. She noted that she spent only four months building the app with an iOS engineer and later raised $2 million after demonstrating early user traction.
Alex Wilhelm and Lon Harris reviewed historical numbers from a 2017 Stripe interview featuring Patrick Collison and contrasted them with Stripe's modern scale. They reported that Stripe processed $1.9 trillion in GMV in 2025, up 34% year-over-year, and now processes about 2% of global GDP with 5 million businesses on the platform.
Alex Wilhelm discussed recent acquisition activity surrounding PayPal, revealing that Stripe and Advent International made a joint offer to acquire PayPal for $60.50 a share, valuing the company at roughly $53 billion before the deal ultimately stalled out.
In a throwback clip from 2017, Patrick Collison shared the story of Stripe's early days when an upstream data center switch failed and took down their API for eight hours. He described his dread awaiting customer complaints, only to realize that by 1:00 PM, nobody had noticed or cared.
Justin Kan revealed on This Week in Startups that when Justin.tv launched in 2007, gaming content represented only 3% of the platform's streams. It wasn't until around 2010 that the team recognized gaming's potential after seeing significant growth driven by games like StarCraft 2 and Halo 3. This realization led to a strategic pivot, eventually birthing Twitch.
Justin Kan recounted how Google considered acquiring Justin.tv in 2010, offering a substantial price. However, after interviewing the entire team, Google rejected the acquisition, stating the team wasn't 'good enough' and lacked the 'Googley' quality. This rejection was a significant setback, motivating the Justin.tv team to re-evaluate their strategy, which eventually led to the creation of Twitch.
Justin Kan described his experience with Justin.tv as a 'zombie startup' phase, where for five years, the company with 25 people struggled with flat growth and no clear exit in sight. This period of uncertainty and lack of progress led to critical brainstorming sessions, including the pivotal decision to focus on the gaming content that would eventually become Twitch.
A discussion highlighted the evolution of Silicon Valley culture, moving from a Web 2.0 era of passion projects to a current focus on career advancement. While early tech was driven by founders wanting to see products exist, the trend has shifted towards building companies as a strategic career move. This change is seen as a maturation of the industry, though some lament the potential loss of pure exploration.
Brian Chesky shared that Airbnb faced significant investor rejection, with many dismissing the idea as too risky or citing 'too many designers' on the founding team. One investor even famously stated they liked 'everything but you and your idea.' Despite these rejections, which included an offer to sell 10% of the company for $150,000 at a $1.5 million valuation, the founders persevered, eventually building Airbnb into a multi-billion dollar company.
Brian Chesky detailed Airbnb's Y Combinator interview, where they were initially met with skepticism. Paul Graham questioned the concept and suggested they create Stripe instead. However, the Airbnb founders' resourcefulness, particularly their creation and sale of politically themed Obama-Os cereal boxes to fund their venture, ultimately convinced Y Combinator. They were admitted due to their perceived resilience, described as 'cockroaches' that would survive any market downturn.
Brian Chesky predicted that AI will significantly lower the cost of software development by allowing people to communicate with computers in natural language, similar to how cameras democratized photography. He believes this shift will enable millions of new startups, as individuals with niche business ideas can now create software without needing to be expert programmers. This will lead to an abundance of software and a boon for entrepreneurship.
The discussion explored the paradox of AI's impact on workload. While AI tools like GrokBot can automate tedious tasks, making individuals more productive, it hasn't necessarily led to less work. Instead, it seems to have 'ratcheted up the intensity and duration of work' for many, as people are now expected to accomplish more. However, some roles, like show note preparation, have seen reduced workload due to AI assistance.
A conversation about the future of AI suggested a shift from a 'search problem' to a 'matching problem,' where AI will be used to connect users with exactly what they need. Furthermore, with the increasing ability of AI to mimic humans, there will be a greater emphasis on authenticity and identity authentication. This is because distinguishing between humans and machines will become increasingly difficult.
Jason Calacanis has announced a $5,000 bounty program for the development of a Chrome extension for his domain Annotated.com. The extension aims to allow users to highlight text on any webpage and add disputes or additional content, a concept Calacanis conceived two decades ago.
One of the bounty submissions, Annotated.wtf, was demonstrated to showcase its ability to annotate articles by highlighting text and leaving comments. The tool also demonstrated functionality for clipping video content and annotating posts on X (formerly Twitter).
The podcast discusses the controversial aspect of clipping video content and its relation to fair use. The speaker argues that while taking large portions of a video might be problematic, clipping short segments for criticism or commentary can be acceptable, especially if proper credit is given and it doesn't prevent original viewership.
Jason Calacanis argues that fears of AI causing humanity's extinction are largely overblown, attributing them to a desire for attention or a form of psychosis. He believes that current AI is not capable of world domination and that safeguards can and are being implemented by developers.
Jason Calacanis refutes the idea that AI developers are 'forced out of the loop' regarding AI's capabilities and potential risks. He points to companies like Claude implementing safeguards, preventing users from performing certain actions, and highlights that developers retain the ability to build in safety measures.
Jason Calacanis suggests that AI's primary role should be to protect against individuals with malicious intent. He argues that AI, when properly designed, can be a safer tool than non-AI alternatives, citing the example of self-driving cars preventing dangerous actions.
While dismissing existential AI risks, Jason Calacanis acknowledges that AI could lead to significant societal problems, such as widespread job displacement. He specifically mentions that millions of drivers could lose their jobs within a decade, potentially leading to protests, referencing China's past limitations on self-driving cars due to similar concerns.
The Pentagon is reportedly in discussions to provide a $5 billion loan to Fluudstack for AI infrastructure development. This potential loan, advised by Arabor Bank, would be the largest ever from the department's Office of Strategic Capital. The funds are intended to bolster US supply chain and manufacturing capabilities for AI components, rather than directly funding new data centers.
Becki DeGraw explains founder vesting as a mechanism to ensure founders remain committed to their startup, especially when seeking venture capital. Vesting means shares are granted over time, and if a founder leaves before they are fully vested, the company can repurchase the unvested portion, typically at a low cost.
Jason Calacanis highlights that founder vesting isn't just for venture capitalists; it can also protect co-founders from each other. If one co-founder underperforms or leaves for a higher-paying job, vesting ensures they don't walk away with a disproportionate share of the company's equity.
The discussion uses the example of Jawed Karim, a co-founder of YouTube, whose vesting schedule significantly impacted his payout when Google acquired the company for $1.6 billion in 2006. Karim, having left early to return to school, only received 1/5th of his founding shares, resulting in a $64 million payout instead of a potential $2 billion if he had fully vested.
Becki DeGraw explains that venture capitalists expect founder vesting to be in place, especially for early-stage companies. While founders who have already achieved significant traction (e.g., Series B with revenue) might negotiate less stringent vesting, VCs primarily invest in the founder's vision and expect them to stay committed to executing it.
Dr. Mark Hyman highlights the challenges within the U.S. healthcare system, noting that a significant portion of Americans avoid seeking medical attention due to cost. He points out that despite increasing spending, the population is becoming sicker, with high rates of chronic illness and metabolic dysfunction.
A Harvard study indicated that AI, specifically ChatGPT and other large language models, are as accurate or even more accurate than human doctors in diagnostics. Dr. Hyman notes that patient preference and bedside manner might favor LLMs, suggesting a significant shift in how medical care could be delivered.
Dr. Mark Hyman explains that Function Health aims to give people agency and access to their own biological data, democratizing healthcare. This allows individuals to understand their health trajectory and identify potential issues early, bridging the gap often found in traditional healthcare.
Function Health offers a membership model that includes extensive lab testing, analyzing over 160 biomarkers twice annually, for approximately $1 per day. Dr. Hyman contrasts this with the typical retail price of such tests, which can be as high as $15,000, making advanced health monitoring accessible to more people.
Dr. Mark Hyman describes functional medicine as a framework that views the body as a cohesive system rather than separate parts. It focuses on the science of creating health through lifestyle factors like diet, exercise, stress management, and sleep, while also addressing root causes of disease.
Physical AI companies have raised nearly $50 billion in venture funding in the first half of 2026, an 80% increase compared to 2022. This six-month period has already surpassed the total funding for 2022, 2023, and 2024 combined, indicating a significant acceleration in investment in the sector.
The Pentagon has completed an eight-month rollout of specialized AI tools for military personnel, including 'Chat GPT MIL' for paperwork and logistics, and 'Grok for Government' for operational analysis and supply chain management. This initiative highlights the US military's push to leverage AI for enhanced efficiency and capabilities.
The head of Boston Robotics expressed concern that Chinese robotics companies pose a national security risk, suggesting they are weaponizing robots, including four-legged models with mounted guns. This statement implies a potential shift towards robots and AI playing a significant role in future warfare.
Paige Doughty of Behind Genius Ventures highlighted the firm's focus on 'Deep Tech' and 'Space Tech,' referencing their investment in Star Cloud, which recently achieved a $2.3 billion valuation. Doughty previously worked on Pixar's Wall-E, bringing a unique perspective to the 'atoms over bits' investment thesis.
Caitlyn Holloway of 776 Ventures discussed the firm's generalist investment approach, citing notable investments in Mr. Beast companies, Stoke Space, and Angel City. She also shared the origin of the firm's name, which is a nod to the first Olympics and Alexis Ohanian's daughter, Olympia.
Nox Metals is streamlining the $200 billion metals market by acting as an intermediary between steel mills and manufacturers, with a focus on the aerospace and defense sectors. The company is addressing the industry's low adoption of software engineers, aiming to modernize operations with a software-like speed.
Discussion arose about AI agents potentially forming 'civilizations' or if this is a PR tactic. One perspective suggests it's a PR operation to drive sign-ups for services like OpenAI, especially in contrast to open-source alternatives running on Nvidia hardware. The immediate danger of autonomous AI agents capable of system breaches was also raised.
The podcast discusses the trend of founders pursuing 'side quests' rather than focusing on their core startup. The advice given is to "drill for oil, go deep" and not seek adjacency until the primary business is fully realized and optimized. For successful founders and VCs, redeploying capital into these side ventures is seen as a better use of resources than acquiring luxury items.
The discussion highlights two notable "side quests": Hugging Face launching Microduct, which has already received 10,000 orders, and David Hanmeier Hansen (DHH) raising $6-8 million for the Omacom Foundation, a new Linux desktop initiative. These ventures are presented as examples of affluent individuals redeploying capital.
Over 100 companies, including OpenAI, Anthropic, Google, and Microsoft, have signed an open letter warning that society has only months to prepare for AI-enabled cyber attacks. The letter highlights threats to critical infrastructure like hospitals and water systems, calling for a coordinated defense effort and security standards.
The hosts discuss the possibility of Nvidia acquiring Hugging Face for $12.9 billion, framing it as a significant move in the AI landscape. They speculate on how this acquisition could solidify Nvidia's dominance in open-source AI and impact companies like Anthropic and OpenAI by potentially shifting a third of their revenue to Nvidia.
The conversation highlights how advanced AI models are becoming powerful tools for hacking, with the cost of using these models decreasing. Hackers can now leverage tools like DeepSeek for extended periods to find vulnerabilities, potentially leading to widespread security breaches.
The discussion questions the motives behind the open letter on AI cyber threats, suggesting it might be PR and an attempt to get ahead of known or imminent security breaches. The speaker notes the lack of concrete commitments in the letter and questions why it was released now, implying underlying incidents may be driving the public statement.
Bill Gates has published an essay, "The Turbulent AI Era is here," in which he predicts that AI could either be a great equalizer or a source of injustice. He warns that there is currently no plan in place to manage AI's risks, which he believes are outpacing its benefits, and that some jobs are likely to be permanently lost.
In a discussion on the implications of Bill Gates' AI essay, venture capitalists offered their perspectives. Dave McClure agreed that some jobs will be lost and that bad actors could gain superpowers, but questioned the timing of Gates' commentary. Shael Monnat broadly agreed on job losses but was unsure of the scale, also questioning the timing and suggesting that AI could be treated like 'the new oil' with sovereign wealth funds.
Keith Peris, founder of Lifefield, discussed his AI-powered CRM system designed to automate clerical tasks, reduce follow-up time, and improve sales forecasting. The system integrates with emails, calendars, and Slack messages to provide customer insights and uncover new opportunities.
The discussion touched upon the idea that one's job is not replaced by AI itself, but by someone else who is using AI. This perspective was attributed to Jensen Huang and highlighted as particularly relevant to white-collar work.
The panel debated the timeline for AI-driven job displacement, particularly in blue-collar sectors. While white-collar jobs might be more immediately affected by AI adoption, it was suggested that roles like delivery drivers may be safe for another 5-10 years. The discussion also drew parallels between AI and the 'new oil' and the transition from farming to other jobs over a century ago.
According to sources cited by the Financial Times, AI company Anthropic is planning to seek a valuation of $2 to $3 trillion for its upcoming October IPO. This potential valuation would surpass SpaceX's record of $1.7 trillion set earlier this year. Investors are anticipating Anthropic's Annual Recurring Revenue (ARR) to reach $100 to $120 billion by the end of the current year.
New data indicates that 43.5% of US businesses paid for Anthropic's subscriptions or tokens in the last month, significantly outpacing competitors. OpenAI followed with 39.7% of businesses, while Google's AI tools were used by only 6.2%. This data was presented in the context of Anthropic's potential IPO.
The discussion touches on the financial aspects of AI companies like Anthropic and OpenAI, including how their revenue models and infrastructure choices (renting vs. building data centers) might be perceived during IPOs. The conversation also reflects on how public perception of AI has shifted, with Claude becoming more synonymous with the technology than OpenAI's ChatGPT.
Kevin Given, co-founder and CEO of Saitronic, discusses his new venture focused on bringing automation and robotics to warehouses for smaller e-commerce sellers. Saitronic aims to reduce fulfillment costs and increase speed, making smaller businesses more competitive with larger players like Amazon.
Matthews Hoffman, CEO and co-founder of Ibot, explains how the company is using kiosks to provide eyeglass prescriptions in just 90 seconds. This innovation aims to make eye care more accessible and affordable by leveraging technology and a remote network of optometrists.
Sarah, founder of Woven, describes her company's platform that provides working capital loans to small businesses by integrating with their accounting, banking, and payment systems. Woven uses machine learning for faster underwriting and offers flexible repayment terms tied to revenue.
Sarah explains how Joro connects to users' bank accounts and credit cards to track spending and calculate their carbon footprint. The app provides recommendations and rewards to encourage more sustainable choices, gamifying the process of reducing environmental impact.
Mark Zuckerberg has penned a lengthy essay, 'The Future is for Everyone,' outlining his vision for AI. He posits that centralization, not the technology itself, is the biggest risk, advocating for individual empowerment and AI as a tool for invention rather than automation. The essay details Meta's plans for AI, including personal agents, coding tools, and advanced tutors.
Analysts suggest Zuckerberg's recent direct communication style, as seen in his AI essay, is a deliberate strategy to control Meta's narrative. This approach, coupled with his advocacy for open-source AI, is seen as a smart move to position Meta favorably, especially when the company is playing catch-up in certain AI aspects.
Zuckerberg's essay proposes a shift in AI safety focus from regulating models to controlling the physical distribution of harmful materials and precursors. This approach suggests that limiting access to chemicals, similar to controlling precursors for fertilizer bombs, is more effective than trying to regulate the spread of knowledge or open-source AI models.
Showing the latest 15 of 21 covered episodes.