Anthropic S-1 Shocker, Greatest Business Ever? Google's $500B Wipeout, Midterm Predictions

Anthropic’s reportedly targeted November 16 IPO would be led by Goldman Sachs and Morgan Stanley, use ticker ANT, and seek $100 billion—more than the $85 billion SpaceX raise cited on the show. Polymarket put the odds of a raise of at least $95 billion at 78%, but the odds of a $1.75 trillion–$2 trillion valuation at 31%. Anthropic reported $16.3 billion in first-half 2026 revenue, including $11.6 billion in Q2, and 14-fold year-over-year growth. Its year-end target is $100 billion in ARR, though a host questioned whether that means annual recurring revenue or annual revenue run rate, and also doubted the company’s $40 trillion addressable-market estimate.

Saks called Anthropic potentially the biggest venture outcome ever and argued that its roughly $80 billion reported loss was a non-cash accounting effect tied to convertible notes from infrastructure partners. He said the company had raised about $110 billion, spent roughly $9 billion–$10 billion, and retained around $100 billion in cash or equivalents; other estimates put cash above $100 billion, including one projection of $125 billion by June. Saks cited gross margins of about 85% and roughly tenfold annual improvement in cost effectiveness for a given task. Speakers said algorithmic, data-curation, and architecture gains can make the same answer cheaper or enable more capable models, allowing lower prices alongside improving margins.

Customer growth and funding capacity underpin the bullish case, but the panel stressed execution risks. Anthropic customers spending at least $100,000 grew from 200 at the end of 2024 to 1,500 at the end of 2025 and 11,900 by the end of September. Chamath estimated about 40 million individual and small-business customers below that spending level and argued growth could come from both deeper spending and new users; he said Anthropic might need only 25%–30% of its potential market. He also said the company must demonstrate enough return on investment to move its 10,000 customers toward multi-million-dollar commitments. Speakers described a prospective $100 billion IPO, more than $100 billion in cash, and an $80 billion debt facility as potentially providing over $300 billion in funding capacity for compute commitments. One estimated a $1.3 trillion–$1.7 trillion IPO valuation, with a midpoint near $1.5 trillion, and argued that needing less capital would strengthen Anthropic’s pricing power and control over the float. Chamath considered “14/15/16” fair, not cheap; Jason argued the company could become one of history’s greatest business franchises, projecting a path from about $100 billion in revenue this year to $300 billion and eventually $600 billion.

Compute, financing, and competition temper that view. Chamath identified data centers, financing commitments, and power as the S-1’s main compute constraints, followed by regulation, data retention, privacy and intellectual property, AI safety, monetization, competitors, and governance. He warned that AI firms could compete with U.S. Treasuries for capital, raising borrowing costs and bringing defaults forward; he estimated the sector could absorb 40%–50% of capital-market equity and debt flows and 30%–40% of Treasury issuance. One speaker said Anthropic was leading the race toward recursive self-improvement, citing a company blog’s report that 26% of tasks involved in developing the next generation of models used it as of August, while speculating the share was higher by October. Speakers also said cheaper models can spur demand: agents and continuous cyber defense might use 50–100 times as many tokens as older applications. Open-source models, particularly for coding, remain a competitive risk, while workplace adoption is constrained by cyber concerns. A cited Gallup survey found roughly 10% of workers use AI daily and about 20% frequently; speakers suggested daily use could reach 60%–80% if access to email and company knowledge bases becomes safer.

Google’s shares fell 10%–15% on Thursday after Q3 results, wiping out about $500 billion in market value. Search revenue rose 12%, to $63 billion from $56 billion, slower than the prior 17% growth rate, and ad clicks did not keep pace with searches. Overall revenue rose 25%, while Google Cloud nearly doubled year over year to $30.7 billion. Freeberg said Google traded at roughly 21 times operating income and that Cloud operating income had climbed from $3.5 billion a year earlier to more than $11 billion. Sacks saw the growth in Cloud and other businesses as positive and the stock as somewhat oversold; others warned that AI agents could threaten the cost-per-click advertising business that funds Google. A widely circulated post alleged Google had changed its ad system and moved away from a Vickrey auction, potentially increasing revenue, but a speaker said the claims were unverified and urged Google to respond. The panel warned that investors could react badly if they conclude ad revenue is being pulled forward or consumer AI cannot offset slowing search. One speaker bought more shares during the show and remained long Google.

The panel’s midterm forecasts ranged from a normal split result to a Democratic sweep. One speaker expected Republicans to keep the Senate and Democrats to win a narrow House majority, rejecting predictions of a “tsunami” or “bloodbath” and urging people to vote. Another predicted a sweep, citing Polymarket odds of 68% for Democrats to win the Senate and 92% for the House; they expected investigations, subpoenas, an impeachment process against Trump, and two chaotic years. That speaker attributed the forecast to Trump’s low approval, anti-Trump turnout, and dissatisfaction among some Trump supporters over the economy, inflation, and wars, while also arguing Democrats had grown more extreme and had not adjusted since losing in 2024.