AI: Mega-IPO Preps, Stripe M&A, Open Source Steam, & More. AI-RTZ #1186

1. Anthropic’s mega-IPO prep gathers steam: Anthropic is building revenue momentum with its class leading AI Coding and enterprise products, at a run rate of $65+ billion, with second quarter revenue surging more than 14-fold year over year. The company has filed confidentially with the SEC, was last privately valued around $965 billion after its May round, the first time it eclipsed arch-rival OpenAI, and is shoring up investor confidence ahead of an expected fall debut, warming up institutions on the roadshow story while reportedly shopping along the way, in talks to buy Decart for about $6 billion. This week brought the governance piece: Anthropic plans supervoting shares for CEO Dario Amodei and its co-founders, the same founder-control structure Meta’s Zuckerberg and Snap’s Spiegel used. Add the trial-balloon valuations floating in the $2 trillion range, and the commitments both frontier labs are making on the way to the public markets, and the fall calendar is shaping up as the largest IPO story in market history. More here.

   

2. OpenAI not far behind: OpenAI is not far behind Anthropic in its revenue run rate and follow-up mega-AI IPO, also confidentially filed and expected to follow Anthropic’s float. The notable shift this week: CFO Sarah Friar told investors that enterprise is now a bigger business than consumer for the company best known for ChatGPT’s billion-plus users, a deliberate repositioning for the public markets on a run rate past $40 billion that has roughly doubled since year-end. The revenue engine rebuild continues underneath, with a new chief revenue officer in under a year, senior leadership churn, aggressive price and product moves in AI Coding to close the gap with Anthropic, and $500 billion of Ohio data center fine print worth reading closely. Two mega-AI IPOs pre-marketing in parallel, each striking its safety pose along the way, is itself the story: the public markets are about to get their first direct read on frontier AI economics. More here.

   

3. Stripe’s OpenRouter AI Developer Strategy: Private payment behemoth Stripe has a strategy to leverage its largest acquisition to date, the reported $7.5 billion purchase of OpenRouter, the three-year-old AI model gateway its founder long called ‘the Stripe of AI’. OpenRouter routes developer requests across 400+ AI models from more than 80 providers, open and closed, picking the optimal model per task, and serves the same developers who run close to $2 trillion in payments through Stripe’s infrastructure, OpenAI and Anthropic included. The $7.5 billion price values a company that raised just $164 million and was marked at $1.3 billion as recently as May, from a payments giant telling investors ‘the singularity’ has begun. The strategic frame: developers managing token payments in and out is the same motion as managing cash payments in and out, making this a platform-of-platforms play on the AI token economy itself, one of the hidden figures beneath the week’s headlines. More here.

   

4. Open Source AIs build global steam: Hugging Face’s latest global open source AI report is worth a discussion, a census of where the open model world actually stands this summer. It shows momentum in China and beyond, with Alibaba’s Qwen family crossing three billion downloads to pass Meta’s Llama and Google’s open models as the most downloaded in the world, while China’s open-weight labs ship models nipping at the closed frontier at a fraction of the price, DeepSeek’s new multimodal test model chasing Anthropic’s Opus 4.8 being Friday’s example, from the academic-founder ecosystem executing through conditions way harder than they look. It is also why I keep arguing Nvidia and Apple should be the US open source champions in response. The two-part arc this week paired the China zoom-in, downloads and those white-hot IPO valuations, with the global zoom-out from the Hugging Face data. Open source AI is no longer a sideshow to the frontier labs; it is the distribution layer of this AI Tech Wave, and its center of gravity is shifting east. More here and here.

   

5. Timely Anthropic AI Research on multi-agent behavior: The latest Anthropic AI Research on AI Agents interacting with each other is worth a closer look, arriving just as the industry pivots from single chatbots to fleets of agents doing real work. The research studies how populations of AI agents behave together, where they cooperate, where they fail in correlated ways, and what coordination actually requires, useful grounding against both the utopian and doomer framings, and timely as long-running AI agents with metered pricing go mainstream. My take remains that the future is not one monolithic ‘Skynet’, but billions of AI agents from many makers, interacting with each other and with us, which makes agent-to-agent behavior, trust and coordination infrastructure the practical frontier to watch, with the ‘Forever Problems’ like Prompt Injections still unsolved underneath it all. More here.

   

Additional Items.

1. Takeaways from Anthropic’s AI Watermarking efforts: statistical watermarks woven into Claude’s text outputs, detectable at scale without changing what users see. Provenance infrastructure is quietly arriving ahead of the regulation that will eventually demand it. More here.

2. AI Industry’s $3+ trillion ‘off-book’ AI Capex has details worth noting: SPVs, leases and financing vehicles put Big Tech’s true AI buildout trillions above the reported capex lines. Worth keeping in mind as the AI infrastructure debate plays out on headline numbers. More here.

(Additional Note: AI Ramblings is now a weekday Daily podcast called AI Ramblings Daily (ARD). Different content than AI-Reset to Zero (AI-RTZ), which remains a daily morning substack with now over 1,180+ ‘MY TAKES’ on key AI events and issues turbulently flowing by. AI Ramblings Daily is typically a 20 minute afternoon podcast with my take on additional AI developments of the day. Both daily substack and podcasts typically discuss different AI issues and items. And are free to subscribe. Try this week’s series below:)

ARD 142, ‘Hidden Figures’ in AI. Big Tech, Stripe & Anthropic: $3 trillion of AI capex off the books, watermarks inside Claude’s math, & Stripe’s $7bn OpenRouter buy.

ARD 143, ‘Devil really is in the Details’ with AI. OpenAI, Jane Street & YouTube: the $500B Ohio deal fine print, Leopold’s $15B hit to Jane Street, & YouTube’s AI ‘Slop’-era creator math.

ARD 144, ‘Strike a Pose’ in AI. OpenAI, Anthropic & Apple: OpenAI strikes the safety pose, Anthropic eases its own, & Apple settles up with the EU.

ARD 145, ‘Wheels within Wheels’ in AI. Stripe, Nvidia & Unitree: Stripe buys OpenRouter for $7.5B, Nvidia funds Mercor at $20B, & China Unitree’s white-hot IPO wheels.

ARD 146, ‘Way Harder than it Looks’ in China AI. Nvidia, DeepSeek & YMTC: Nvidia’s denied-but-reported China comeback chip, DeepSeek’s frontier push, & memory maker YMTC’s IPO after CXMT:

(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here.)