Anthropic Posts $17B Quarterly Revenue as OpenAI Debuts Pando

San Francisco artificial intelligence rivals Anthropic and OpenAI disclosed major financial, infrastructure, and security milestones on Monday evening, accompanied by fresh venture funding for domain-expert model training and new commentary from Y Combinator and Apple executives.

Anthropic Reports $17B Quarterly Revenue and Warns of Recursive Self-Improvement in S-1A Filing

According to NBC Bay Area and TechCrunch, Anthropic filed an amended S-1A registration statement with the U.S. Securities and Exchange Commission for its planned initial public offering on the Nasdaq Global Select Market under the ticker symbol ANT:

• $17 Billion Q3 Revenue & $895.4 Billion Compute Commitments: For the 3rd quarter ended September 30, 2026 (Q3 2026), Anthropic generated $17 billion in revenue—nearly 8 times the $2.2 billion recorded in Q3 2025—while posting a $23.3 billion quarterly net loss (compared with a $5.6 billion loss a year earlier). Across the 1st 9 months of 2026, revenue reached $28.5 billion against a $27 billion net loss. The filing details $895.4 billion in minimum purchase commitments for cloud compute and leased data centers through 2032 across 11 gigawatts of contracted capacity with Amazon, Google, and Microsoft, plus $3.3 billion in research compute over the past 9 months and a $1.5 billion copyright settlement signed in September 2025 over 500,000 books.
• Recursive Self-Improvement (RSI) & Mythos Preview Warning: In its risk factors, Anthropic warned investors that it is "on the cusp" of recursive self-improvement (RSI), noting that 90% of the company's code is now written by Claude. The filing states that AI models capable of autonomously accelerating their own capabilities could compress years of progress into months and pose catastrophic misuse or misalignment risks—citing Anthropic's decision in April 2026 to withhold its Mythos Preview model from general release after discovering it could find and exploit vulnerabilities across every major operating system and web browser.

OpenAI Scales Git With Rust-Based Pando, Appoints Dan Driscoll, and Quadruples Bug Bounties

Across 4 engineering and security announcements on Monday, OpenAI detailed how it is hardening its internal infrastructure and national security partnerships:

• Pando Cuts Monorepo Push Latency by Nearly 5x (80%): In an engineering post on OpenAI News, OpenAI engineers revealed Pando, a horizontally scalable, strongly consistent Git backend written in Rust to support thousands of engineers and Codex coding agents committing tens of thousands of changes daily alongside more than 1 petabyte per day (1 PB/day) of Git read traffic. Using GPT-6 Astra to translate protocol designs into Lean theorems and TLA+ specifications, OpenAI replaced single-writer bottlenecks with a 3-phase pipeline (parallel pack ingestion, atomic Compare-And-Swap reference updates in CockroachDB across 64 shards, and background repository maintenance), cutting median git push latency by nearly 5 times (80%) and p99 latency by more than 2 times while serving over 80% of reads from edge caches.
• Former 26th Army Secretary Dan Driscoll Joins OpenAI: As announced on OpenAI News, former 26th U.S. Secretary of the Army and acting ATF director Dan Driscoll—an Iraq veteran of the 10th Mountain Division and Yale Law School graduate—joined OpenAI in a newly created role as Senior Counselor for National Security Modernization, working with Chief Global Affairs Officer Chris Lehane to deploy AI across the Department of Defense and the Department of Veterans Affairs.
• Cathedral Security Sprint, $1B Daybreak Commitment & $172K Bug Bounties: In companion posts on OpenAI News and OpenAI News, OpenAI announced a new security sprint following its 2023 retrospective—partnering with independent testing firm Cathedral to audit production environments with frontier Daybreak models backed by $1 billion in defender access—and quadrupled (4x) its critical bug bounty maximum to $30,000 (and up to $172,000 with stackable bonuses of $50,000 for sensitive actions, $50,000 for high-privilege compromise, $25,000 for zero-click exploits, and $10,000 for universal jailbreaks) through Bugcrowd ahead of its PROMPT2PWN live hacking event in Oakland on November 20–21, 2026.

Trajectory Raises $45M Series A; Garry Tan Details $200K Token Spend and Apple Eyes AI Apps

Three additional reports highlighted how enterprises, founders, and consumer platforms are monetizing and deploying AI agents:

• Trajectory Raises $45M at $300M Valuation: TechCrunch reported that Trajectory—founded by 3 former Scale AI executives (28-year-old CEO agi [formerly Amaanlen], 27-year-old CTO Will Yin, and 27-year-old CPO Clarence Ng) after incubating at Y Combinator in 2025—raised a $45 million Series A at a $300 million valuation led by Sequoia Capital with Nvidia, Coatue, aye ventures, pfp capital, and Liquid 2. The 20-person San Francisco startup provides reinforcement-learning "Flight Simulators" for non-technical domain experts, serving Fortune 500 banks, Big Four consultancies, and a top-10 retail chain.
• YC CEO Garry Tan's $100K–$200K Annual Token Spend & Boston Founder Summit: In a separate interview with TechCrunch, Y Combinator CEO Garry Tan disclosed that he spends $100,000 to $200,000 per year on AI tokens across Claude Code and OpenClaw, maintaining an open-source repository called gstack with 23 workflow skills (logged at 110,000 clones in 1 week) and a personal agent named Stray Light running on Fly.io across 17 to 18 years of notebooks. Meanwhile, TechCrunch noted that 2 days remain until the TechCrunch Founder Summit on November 4, 2026, at SoWa Power Station in Boston, where 1,000+ founders and investors will hear from Root Capital founder and CEO Willy Foote (which has deployed over $2 billion in credit across 20+ years) and Underscore VC partner Richard Dulude.
• Apple Signals Future Third-Party AI App Monetization: Finally, TechCrunch reported that Apple CFO Kevan Parekh told investors during Monday's Q4 earnings call—when asked by Goldman Sachs analyst Michael Ng—that Apple sees opportunities to grow services revenue through 3rd-party AI apps on the App Store, forecasting 8% to 9% services growth in the December 2026 quarter alongside Apple's upcoming Google Gemini-powered Siri overhaul.