Week of Friday October 23, 2026
1. Anthropic readying AI Compute & mega-AI IPO: Anthropic’s draft S-1 shows $750+ billion AI Compute Plans, Risk Factors, and AI enterprise growth. First-half revenue rose to about $16.3 billion, from $1.1 billion a year earlier, while compute spending rose to nearly $9 billion. By my math, Anthropic now takes in about $1.81 of revenue for every dollar of compute, up from 63 cents last year. The commitments behind that: about $450 billion of mostly noncancelable cloud contracts with Google, AWS, Microsoft and neoclouds, more than $160 billion for Google TPUs through Broadcom, and up to $150 billion more in talks, per Bloomberg. The $80.7 billion first-half net loss is mostly accounting charges on its fundraising instruments, and about a third of the draft is risk factors, some of them likely uninsurable. My take, from AI-RTZ #1246: of the mega-AI IPOs, Anthropic’s story requires the fewest leaps of AI faith, with revenue now outrunning compute. But the draft puts real numbers on the leaps that remain: a compute bill locked in for years while the builds behind it can slip, concentrated customers, and AI risks the insurers won’t touch, as I laid out in AI-RTZ #1243. Mr. Market sets the final price before Thanksgiving. More here and here.
2. OpenAI’s penultimate private raise: OpenAI is finalizing a $30 billion plus private bridge round at a revised $1.25 trillion valuation, down from an earlier $1.4 trillion number ahead of its expected 2027 mega-AI IPO. The round is led by investors from the UAE and Coatue, with commitments due by October 30. Demand from wealthy private clients came in softer than pitched, per the FT, but even the trimmed price is about 50 percent above OpenAI’s last round in March, at about $852 billion. Alongside it, a big UAE compute deal is under negotiation, even as the war with Iran, including Houthi strikes on Riyadh, led Saudi Arabia to postpone its ‘Davos in the Desert’. My take, from ARD #191: OpenAI has been committed and resilient enough to do what it needs to get things done, and this is no exception. Investors are showing some fatigue after hundreds of billions raised this year, from Google’s $80 billion plus to Meta’s new share class. Expect OpenAI to keep adjusting as needed, especially as new data points arrive from Anthropic’s IPO. In September I called OpenAI the Pepsi to Anthropic’s Coke, in ARD #166. A strong number two, raising tens of billions at a trillion and a quarter. More here and here.
3. Nvidia’s Neocloud backup plan: Nvidia is committing over $170 billion to support Neoclouds like CoreWeave and others, about $164.5 billion in lease guarantees and capacity commitments, plus more than $7 billion invested directly in AI cloud providers. The stress is real: CoreWeave’s shares fell about 20% in a week and its bond yields topped 15%, as rising rates raise the cost of money for the most leveraged builders, per The Information in ‘How CoreWeave’s Debt Binge Brought a Moment of Reckoning’. Firmus pulled its IPO, and Nvidia is now in talks to back it with hundreds of millions more. Neoclouds and enterprises were 45% of Nvidia’s data center revenue last quarter, about $40 billion. My take, from AI-RTZ #1248: the Kingmaker became the AI banker, and now looks more like the neoclouds’ central bank. Nvidia can afford to be the backstop, and it hedges the ‘Frenemies’ building their own chips, as I laid out in AI-RTZ #1234. This is still a financing cycle more than a demand cycle. But those of us who lived through the telecom and fiber boom remember vendors financing their customers. Expect consolidation rather than collapse. More here and here.
4. Meta’s Equity Issuance & AI plans: Meta’s Mark Zuckerberg restructured Meta’s capital structure to accommodate California’s billionaire tax, and Meta’s growing AI compute ambitions. Starting in January, every Meta shareholder gets a new non-voting Class C share for each share owned, effectively a 2-for-1 split. Zuckerberg can sell 20% of his new shares right away, including to pay California’s proposed one-time 5% wealth tax, which would otherwise cut his voting control from about 61% to about 53%, per the NYT. He commits to staying where Meta is headquartered for eight years. And Meta can sell the new shares to fund AI without diluting anyone’s vote. It needs it: record revenue, but the first negative free cash flow since its IPO, and about $825 billion of future commitments, mostly AI compute and data centers. My take, from ARD #190: Mark is battening down the hatches, both defensively and proactively, shielding his stake and his super-majority control, and with it Meta’s AI ambitions. A bespoke move for Meta’s realities. A year ago I wrote that Meta’s AI financing deals would be a roadmap for peers, in RTZ #891. Next chapter: raise the money, keep the vote. More here and here.
5. Elon merging SpaceX/xAI/AI/SI & Tesla: As long expected, Elon Musk is planning to merge his two public entities SpaceX and Tesla. Both have formed special committees of independent directors to discuss it, with Musk recusing himself, though he called a merger ‘a good idea’ on SpaceX’s first earnings call, per CNBC. Combined, they would be worth about $3.6 trillion. Musk controls about 84% of SpaceX’s votes but about 20% of Tesla’s, and the deal needs a majority of all Tesla shares. The quarter behind it: SpaceX revenue up 161% to $14.6 billion, and its first profit as a public company, with its AI business, now called Super Intelligence, up ninefold to $7.6 billion, per The Information. My take, from ARD #189: surprise, surprise indeed. When Elon merged xAI and X, I said Tesla was the next likely stop, in RTZ #675, and again when xAI went under SpaceX in RTZ #987. A bigger combined company would also draw more of the passive index money, as I laid out in AI-RTZ #1115. Having run Goldman’s internet research in the nineties, the big question for me is how analysts value a conglomerate of rockets, Starlink, AI, cars and robots. When Elon is laser focused, things get executed. But here he has a lot of things demanding his ‘laser focus’, both in tech and beyond. More here and here.
Additional Items.
1. Outlining the $10 Trillion+ AI Build Banker plans. More here. 2. Understanding the importance of AI Loops Math. More here and 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 1240 ‘MY TAKES’ on key AI events and issues turbulently flowing by. AI Ramblings Daily is typically a 20 minute afternoon podcast on 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 with ARD Episodes 187: ‘The Art of the AI Deal’: Anthropic, OpenAI & Microsoft, 188: ‘One Step Forward, Two Steps Back’: OpenAI, DeepSeek & Idealab, 189: ‘The Fix Is In’ in AI: Musk, Nvidia, OpenAI & AI Agents, 190: ‘Batten Down the Hatches’ in AI: Meta, Anthropic & Banks, and 191: ‘Bouncing Back’ in AI: OpenAI, Anthropic & Mr. Market.)
Up next on Sunday, a fresh ‘Bigger Picture’. Last week’s ‘The Bigger Picture’ is here as you wait. Stay tuned.
(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here)
No posts
About the Author
