Key Moments
OpenAI's Identity Crisis, Datacenter Wars, Market Up on Iran News, Mamdani's First Tax, Swalwell Out
All-In PodcastKey Moments
New York's 'pied-à-terre' tax on second homes over $5M could crash the luxury real estate market, mirroring London's experience with similar policies.
Key Insights
New York City is proposing a 'pied-à-terre' tax on second homes valued over $5 million, with speculation around a 3.9% annual tax rate.
OpenAI's chief revenue officer sent a leaked memo suggesting a pivot to enterprise customers and questioning Anthropic's $30 billion run rate, claiming it's inflated by $8 billion.
Anthropic's revenue growth is reportedly 10x annually, compared to OpenAI's 3-4x, with OpenAI's revenue around $30 billion in Q2, while Anthropic is projected to reach $80-100 billion by year-end.
Allbirds, a shoe company, saw its stock surge over 450% after announcing an AI pivot and purchasing $50 million in H100 GPUs via a convertible note.
Data center construction is facing significant opposition nationwide, with some towns voting down projects and Maine passing a bill to ban them entirely.
The market is currently shrugging off geopolitical tensions, with the S&P 500 hitting all-time highs, indicating confidence in a resolution of the current conflict.
New York's luxury real estate tax aims to curb 'pied-à-terre' culture
The discussion begins with New York City's proposed 'pied-à-terre' tax on second homes valued over $5 million, with some speculating a 3.9% annual tax. The panelists suggest this could significantly impact demand for second homes in NYC, potentially crashing the luxury market. This policy is compared to London's experience where similar taxes led to a hollowing out of certain neighborhoods, with properties becoming repositories for assets rather than lived-in spaces. The argument is made that this tax could stifle new development and that wealthy individuals parking assets in NYC are beneficial to the city's economy, contributing taxes without utilizing services. The discussion highlights that London saw capital flee after changes to its 'non-dom' tax status, redirecting to places like Zurich and Milan. San Francisco and Los Angeles have also implemented mansion taxes, leading to a significant drop in real estate transaction volumes due to high transfer costs.
OpenAI's identity crisis and the race for AI dominance
A leaked memo from OpenAI's chief revenue officer reveals an 'identity crisis,' suggesting a pivot towards enterprise customers and questioning Anthropic's $30 billion run rate, which is alleged to be inflated by $8 billion due to revenue sharing and accounting practices. Anthropic is described as building its narrative on 'fear, restriction, and the idea that a small group of elites should control AI.' OpenAI is also rumored to be focusing on the 'agent platform layer.' The memo implies a strategic shift, potentially due to declining ChatGPT market share against competitors like Gemini and Claude, and Apple's potential future entry. Investors are reportedly frustrated with OpenAI's lack of focus, with secondary markets now valuing Anthropic higher than OpenAI for the first time. The core debate revolves around whether OpenAI should focus on its massive consumer base or pivot to enterprise. Some argue that doubling down on consumer business could yield $3-4 trillion in enterprise value, while a dedicated enterprise focus, particularly on coding with 'Codeex,' could capture another $2-3 trillion.
Anthropic's exponential growth and the compute constraint challenge
A key point of contention is Anthropic's rumored 10x annual revenue growth, significantly outpacing OpenAI's 3-4x growth. While both companies were reportedly around $30 billion in revenue at the start of Q2, Anthropic is projected to reach $80-100 billion by year-end. This rapid growth is attributed to a strong focus on enterprise, particularly coding tasks, where businesses are willing to pay for metered usage. Consumer markets, which OpenAI prioritized, are seen as having lower willingness to pay and a smaller conversion rate to premium services. The discussion also touches upon the physical limitations of scaling AI, including compute power, electricity, and data centers. Concerns are raised about Anthropic hitting these limits, evidenced by user complaints of Claude 'thinking less' due to reduced thinking time. The panelists wonder if Anthropic's 'doomerism' stance, which aligned with getting compute from hyperscalers, will become a liability now that they may need to build their own data centers.
Big Tech's compute advantage and the datacenter 'disaster'
The conversation shifts to the 'datacenter wars' and compute dominance. Large tech companies like Google, Microsoft, and Meta are leveraging their existing infrastructure and profits to invest heavily in AI compute. The hyperscalers control a significant majority of all compute, leading to a strategic game theory where they could potentially throttle access for frontier labs like OpenAI and Anthropic to level the playing field. The panelists note that OpenAI has attempted to reduce its reliance on hyperscalers by investing in its own infrastructure, and Anthropic may face a similar necessity. The 'doomerism' narrative surrounding AI is questioned, with speculation that it might be tied to controlling compute capacity. The narrative around data centers becoming unpopular in the US is also discussed, with reasons ranging from power concerns to opposition from 'doomer groups' and environmental activists. Maine has even passed a bill banning data center construction.
Allbirds' AI pivot and the speculative market
The Allbirds story serves as an anecdote for peak speculative behavior in the market. The shoe company, previously valued at billions, saw its stock surge over 450% after pivoting to AI and reportedly purchasing eight H100 GPUs for $50 million. This is likened to the dot-com era, where simply adding '.com' to a company name could inflate valuations. The panelists discuss how the era of 'zero-interest-rate policy' (ZIRP) and massive capital influx, particularly during COVID, encouraged investors to look two to three years ahead based on current growth, leading to inflated valuations for companies without strong fundamentals like gross margins. Bird scooters are mentioned as another example of a company with a high valuation that struggled due to regulatory issues, highlighting the risk of cities controlling market dynamics and stifling innovation through 'regulatory capture.'
The 'Price is Wrong' game show and startup failures
A segment of the show features a game 'The Price is Wrong' where contestants guess the valuation of famously overvalued startups. Examples included a $13 billion valuation for a JPEG-selling NFT company (likely referring to Bored Ape Yacht Club) and Clubhouse, a $4 billion social audio app. Theranos, a $9 billion medical technology startup, and Quibi, a $1.7 billion short-form streaming platform that shut down quickly, are also highlighted as examples of major startup failures. This segment underscores the theme of collective delusion and the tendency of the market to sometimes reward hype over substance.
Geopolitical events and market resilience
The market's strong performance despite ongoing geopolitical conflicts, particularly in Iran, is discussed. The S&P 500 has hit all-time highs, recovering losses from the start of the war. This resilience is interpreted as the market pricing in a resolution or de-escalation of the conflict. Statements from the President suggesting the war is nearing completion and military objectives are being achieved contribute to this sentiment. The idea that Donald Trump's policy decisions are often influenced by the stock market's performance is also explored, with the theory that he aims to keep the market stable through his actions.
The future of AI and its impact on productivity
The panel discusses the transformative potential of AI, particularly in boosting productivity. While some, like Jal, express skepticism about tangible revenue and profit increases in large companies due to slow adoption and complex change management, others see significant upside. Travis highlights the productivity gains in tech companies that have embraced AI, with founders reporting faster development cycles and new feature rollouts. However, he also cautions that current agents are not yet 'smart' enough and still require human oversight, lacking taste and the ability to perform novel tasks independently. The debate touches upon whether startups or large enterprises will better harness AI's power, with Sax leaning towards Jal's view that change management is a major hurdle for larger corporations. The argument is made that the ROI is becoming evident at the model layer for AI, but the application layer's profitability is still being scrutinized.
Mentioned in This Episode
●Products
●Software & Apps
●Companies
●Organizations
●People Referenced
Common Questions
The proposed pied-à-terre tax in New York is an annual tax speculated to be around 3.9% on second homes valued over $5 million. This tax is expected to significantly impact demand for second homes in New York, potentially crashing the entire real estate market by effectively doubling the purchase price over a decade.
Topics
Mentioned in this video
Mentioned in the context of New York City's proposed pied-à-terre tax, which would affect homes over $5 million within 15 miles of this area.
Mentioned as a state with a pro-development approach to land use, in contrast to Democratic cities that restrict building.
Mentioned as a city that introduced a similar 'mansion tax' (5%) to San Francisco, leading to a dried up real estate transaction volume.
Mentioned as a destination for wealthy individuals and assets after London crippled 'non-dom status'.
Used as an example of a city where a similar stamp tax on high-end real estate led to a collapse in the market and parts of the city becoming 'unlived'.
Cited as a positive example where allowing substantial development led to decreased rents and housing prices despite increased net migration.
Mentioned as a destination for wealthy individuals and assets after London crippled 'non-dom status'.
Mentioned as a destination for wealthy individuals and assets after London crippled 'non-dom status'.
Cited as having a high transfer tax (6% over $25M, starts at $5M) on real estate, leading to high transaction costs and a dried-up market.
Referenced in contrast to the US regarding public sentiment towards AI, with China viewing AI incredibly positively.
A new AI model from OpenAI, with a 75% chance of being released soon according to a poly market prediction.
Mentioned as a computing company doing well, quadrupling its revenue.
A platform used by 6-7% of all accountants, demonstrating AI's efficiency and growth in specific professional fields.
OpenAI's consumer-facing AI product, noted for its 1 billion users but declining market share as competitors gain ground.
An AI model that is gaining significant market share, contributing to ChatGPT's declining share.
Anthropic's AI model, considered more reliable for day-to-day work tasks but less functional than Codeex for complicated tasks.
Anthropic's new model, noted for its rapid release cadence and supplanting other models like Cursor and Gemini in usage.
An AI coding assistant, mentioned as previously favored but now largely supplanted by Anthropic's Opus model.
An Anthropic model that was held back, speculated to be due to compute constraints and possibly genuine altruism in revealing coding vulnerabilities.
Identified as the overvalued startup (once valued at $4 billion) that allowed mobile-first audio communication during COVID and hosted venture capitalists.
Used as an example of technology that enhanced efficiency (30%), but not to the transformative 10x degree expected from current AI.
Mentioned referencing his previous podcast appearance, where he predicted Anthropic's revenue trajectory.
A politician who withdrew from the California governor's race and resigned from Congress amidst allegations of sending inappropriate pictures to employees.
A prominent politician, implicated as a central figure in orchestrating the exits of Swalwell and Biden from their respective races. Also noted for her exceptional stock trading returns.
Founder of Theranos, mentioned anonymously as the 'famously deep voiced founder CEO' currently serving an 11-year sentence.
A billionaire whose ownership of a New York property was mentioned in the context of the pied-à-terre tax and the debate around doxing wealthy individuals.
Referenced regarding his ability to raise significant funding for OpenAI ($122 billion round), despite financial uncertainties.
American politician, mentioned hypothetically as a target for doxing by a Republican, to illustrate a double standard.
OpenAI's Chief Revenue Officer, who sent a leaked memo criticizing Anthropic and outlining OpenAI's strategic pivot.
Mentioned for his XAI efforts and building compute infrastructure, including Colossus and plans for a fab.
Retweeted by Sax, suggesting that Anthropic held back Mythos due to lack of compute capacity.
Author of an article discussing Sam Altman, where sources allegedly called Altman a 'sociopath', used to criticize current AI leadership.
Suggested as a potential positive spokesperson for the AI industry, in contrast to current figures.
Suggested as a potential positive spokesperson for the AI industry, and mentioned for his projections on capex growth in data centers.
President during a debate performance with Joe Biden, and in the context of the stock market being his 'weather vane' for policy decisions.
A Democratic politician, seen by insiders as the preferred candidate for California governor.
His withdrawal from the presidential race was compared to Swalwell's situation, suggesting political maneuvering by party insiders.
A highly accomplished investor, mentioned alongside Warren Buffett and Nancy Pelosi for exceptional returns.
A congressman praised for his support of freedom of speech and a diplomatic track for Ukraine, also noted for his impressive stock trading performance.
A legendary investor, whose returns were compared to Nancy Pelosi's, with a controversial take on his performance pre- and post-Reg FD.
Founder of Quibi, implicitly identified as the 'Jeffrey' who garnered 'top investors across Silicon Valley' for the failed streaming platform.
Mentioned as not having an AI product in the market yet but possessing a large user base.
Mentioned as one of the companies that could experience phenomenal earnings by properly executing AI deployment.
Mentioned as profitable within its first few years during the mobile wave, used as a benchmark for AI companies needing scaled profits.
Identified as the mispriced startup that raised $1.7 billion and shut down in 6 months, a short-form mobile-first streaming platform.
A quantitative trading firm, mentioned as making a billion-dollar investment in a neo-hyperscaler and a $6 billion compute deal, highlighting compute constraint trends.
Incorrectly guessed as the startup for the 'JPEGs for fake internet money' question, demonstrating difficulty in recalling forgotten unicorns.
A market-making firm, used as a benchmark for frequent stock trading, implying Ro Khanna trades more often.
The discussion revolves around OpenAI's perceived identity crisis, its pivot to enterprise customers, its valuation, and competition with Anthropic.
OpenAI's competitor, discussed in terms of its valuation, growth rate, focus on enterprise/coding, and its stance on AI safety (doomerism).
An open-source project whose architect, Peter Steinberger, was hired by OpenAI, raising questions about OpenAI's strategy for open-source innovation.
Elon Musk's AI company, where new cool stuff and models are reportedly being developed.
An AI model, stated by Chamath to be generally better than Anthropic's models for complex, long-horizon coding tasks.
Travis Kalanick's former company, cited as an example of how network effects based on scale and volume can create a competitive advantage.
Mentioned as a competitor to Uber, illustrating the competitive landscape where network effects are crucial.
Mentioned as a competitor to Uber, illustrating the competitive landscape where network effects are crucial.
Google's AI division, stated to have an outstanding team and be a clear competitor in the AI race.
A company providing solutions for on-site power generation using natural gas, enabling faster data center deployment.
A consulting firm, cited for a study on transformation projects at large enterprises failing, highlighting challenges in AI implementation.
Identified as the mispriced startup ($900M raised at $9B valuation) that dissolved in 2018, with its founder serving an 11-year prison sentence.
An educational institution mentioned as an example of working on improving education with AI.
Identified as the overvalued startup (once valued at $270 million) that sold juice packages for a juice machine, branded as the 'iPhone of juice'.
Mentioned as a company with significant cash on hand, building infrastructure, and important for a potential IPO to de-lever and take chips off the table.
A private company creating dark pools of data for language models, using AI to identify contributors for corpuses, demonstrating AI's impact.
Mentioned as profitable within its first few years during the mobile wave, used as a benchmark for AI companies needing scaled profits.
Released its first proprietary AI model, indicating increased competition in the AI space.
Mentioned as one of the 'doomer groups' that oppose AI progress, potentially astro-turfing nimbyism against data centers.
Threatened to destroy data centers in Gulf states, which were built with American technology, highlighting geopolitical risks.
More from All-In Podcast
View all 402 summaries
90 minAnthropic’s $30B Ramp, Mythos Doomsday, OpenClaw Ankled, Iran War Ceasefire, Israel's Influence
62 minJosh Shapiro on Trump, Iran War Chaos, Israel's Failure, the Economy, and 2028 Race
70 minThe State of Modern War: Palantir & Anduril Execs on Drones, AI, and the End of Traditional Warfare
81 minSpaceX IPO, Iran War Fallout, Quantum Bitcoin Hack, The Space Opportunity
Found this useful? Build your knowledge library
Get AI-powered summaries of any YouTube video, podcast, or article in seconds. Save them to your personal pods and access them anytime.
Get Started Free