OpenAI's Identity Crisis, Datacenter Wars, Market Up on Iran News, Mamdani's First Tax, Swalwell Out
The All-In podcast hosts discuss OpenAI's strategic identity crisis versus Anthropic's explosive growth, the political backlash against data center construction across the US, and New York Mayor Mamdani's proposed pied-à-terre tax on second homes. They also cover Eric Swalwell's resignation from Congress amid allegations, and debate whether AI productivity gains are yet showing up in enterprise bottom lines.
Summary
The episode opens with a discussion of New York City Mayor Mamdani's proposed pied-à-terre tax, estimated at around 3.9% annually on second homes valued over $5 million within 15 miles of Midtown Manhattan. The hosts argue this will devastate demand for luxury second homes, dry up real estate development, and drive away wealthy international investors who pay property taxes while consuming minimal city services. They draw comparisons to London's stamp duty and non-dom tax changes, which they say hollowed out high-end neighborhoods and redirected wealthy buyers to Zurich, Lugano, and Milan. The contrast with Austin, Texas is highlighted — where permissive building policies have allowed housing supply to keep pace with population growth, actually reducing rents despite a doubling of the city's population over a decade.
The main tech discussion centers on OpenAI's internal memo from Chief Revenue Officer Denise Dresser, which attacked Anthropic's reported $30 billion revenue run rate as inflated by roughly $8 billion due to revenue-sharing accounting with channel partners. The hosts interpret the memo as a deliberate leak designed to undercut Anthropic's valuation ahead of fundraising. Anonymous OpenAI investors cited by the Financial Times expressed frustration with the company's lack of focus, and secondary market pricing now puts Anthropic ahead of OpenAI for the first time. The panel debates the growth rate disparity — OpenAI growing roughly 3-4x annually versus Anthropic's approximately 10x — and argues that Anthropic's focus on enterprise coding use cases, where businesses pay on a token-metered basis like electricity, has created a far more scalable revenue model than OpenAI's consumer-oriented $20/month subscriptions. David Sacks argues that if Anthropic reaches $80-100 billion ARR by end of year, OpenAI may face an insurmountable lead within one to two years unless it pivots decisively to enterprise coding.
A significant portion of the episode covers the growing political and populist backlash against data center construction across the United States. Chamath reports that approximately 100 data centers are currently being contested, with about 40% getting cancelled, representing roughly $162 billion in economic value at risk. The hosts identify multiple forces driving this opposition: genuine community concerns about electricity rate increases, well-funded 'doomer' groups using data center opposition as a proxy to slow AI development, and ironically Anthropic itself, which allied politically with anti-data-center groups while relying on hyperscaler compute — a strategy that has now backfired as Anthropic hits compute capacity limits and needs to build its own infrastructure. Freeberg argues the deeper cause is broad populist resentment, with data centers serving as the physical symbol of tech-driven wealth inequality that ordinary Americans feel left behind by.
The hosts play a game show segment called 'The Price Is Wrong,' quizzing each other on famously overvalued startups including OpenSea, Clubhouse, Allbirds, Theranos, and Quibi. The segment transitions into a discussion of the broader 2021 valuation bubble, where physical-world companies were valued like software companies during a period of zero-interest-rate policy and COVID-era monetary expansion.
On politics, Freeberg reveals that as early as December, multiple independent sources told him about allegations against Eric Swalwell involving inappropriate conduct with employees — information that was widely known within Democratic Party circles but deliberately held back and then released in a coordinated fashion. David Sacks draws parallels to how Joe Biden was forced out of the 2024 presidential race, suggesting Nancy Pelosi — Swalwell's longtime mentor — is again the central figure orchestrating the removal of a candidate who had become a liability, this time to consolidate the Democratic field in California's jungle primary where two Republicans were polling highest.
The episode closes with a market analysis discussion. Travis Kalanick offers the thesis that the stock market is Trump's primary 'weather vane' — that Trump consistently pulls back from policies that cause the S&P 500 to drop too far, and markets have learned to price in this behavioral pattern. Chamath notes that while valuation metrics like the Shiller PE and Buffett Index are at or near all-time highs suggesting a risk-off posture, historical data shows that when markets are up 5% in the first half of April, they tend to be up ~32% for the remainder of the year on average. The hosts debate whether AI productivity gains have yet translated into meaningful enterprise profit improvements, with Chamath skeptical and Jason Calacanis arguing that companies deploying AI correctly are already running the table on competitors.
Key Insights
- David Sacks argues that Anthropic's approximately 10x annual growth rate versus OpenAI's 3-4x means Anthropic could reach $80-100 billion ARR by end of 2025, and that this growth disparity — driven by Anthropic's focus on enterprise coding where businesses pay on a scalable token-metered basis — could produce an insurmountable lead within one to two years if OpenAI doesn't pivot decisively to enterprise.
- Chamath argues that Anthropic's anti-data-center lobbying and alliance with doomer groups was a strategic blunder — it made sense when Anthropic relied entirely on hyperscaler compute and was effectively throwing sand in competitors' gears, but has now backfired catastrophically because Anthropic has hit the limits of third-party compute availability and must now build its own data centers in an environment it helped poison.
- Freeberg reveals that multiple independent sources told him as early as December 2024 about allegations of inappropriate conduct by Eric Swalwell with employees — knowledge that was broadly held within Democratic circles but deliberately withheld and then released in a coordinated fashion, which he found more striking than the allegations themselves.
- Chamath contends that the data center backlash is not primarily about electricity rates or AI risk, but rather that data centers have become the physical symbol of tech-driven wealth inequality — the 'temple of the wealthy' — and that most ordinary Americans don't yet feel meaningful benefit from AI in their daily lives, making data centers a natural populist target.
- Travis Kalanick argues that Trump's primary behavioral constraint is the S&P 500 — that Trump consistently retreats from policies that push the market down too far, and that traders have now internalized this pattern, effectively using Trump's market sensitivity as a predictive model for when policy escalations will be walked back.
Topics
Full transcript available for MurmurCast members
Sign Up to Access