Daily Brief Nov 3, 2026
How we checked the items below
For readers who want to dig in. What happened, in brief, is on the Today tab.
This issue draws on material from November 2 to early November 3; main-line item 1 is checked against Oracle's documents of October 28 and February 1, and the From the archive item comes from a SecondSource call of July 14. Last night's collection took in 470 pieces, and this issue uses 12 outside sources with checkable links; the later sections also draw on material from the past week.
1. [Today] Oracle files for US$10 billion of common stock and US$10 billion of bonds at once, matching the funding cap it announced five days earlier
Not proven yet： whether this offering counts toward the cap is still unstated
How we checked
The prospectuses, the investor-day filing and the quarterly report are all statutory filings Oracle made with the SEC; the February plan comes from a press release on Oracle's investor relations site. Every document cited here comes from Oracle itself, with no third-party corroboration. The prospectuses are preliminary, so the share price, the coupon on each tranche and the final amounts are still blank. The cap is one Oracle set for itself and carries no legal force.
2. [Today] Marvell agrees to buy optical-component maker Lumentum in an all-stock deal that values Lumentum at US$115 billion
Not proven yet： the cost savings and market size are the companies' own estimates
How we checked
The deal terms come from the original text of the two companies' joint press release; the exchange ratio, valuation and ownership split are the terms both sides announced. The cost synergies, the market size and the boost to earnings in the first year after closing are forward-looking statements by the companies and have not been checked by any third party.
3. [Today] OpenAI tells a reporter ChatGPT's search volume is about a fifth of Google's; SEO tool maker Ahrefs estimated about 12% in February
Not proven yet： the one-fifth is OpenAI's own figure, and the post doesn't say how it is counted
How we checked
The one-fifth comes from Kantrowitz's post; it is a self-reported figure OpenAI gave him, with no second source. The Ahrefs estimate comes from the original Ahrefs blog post; the 12% is one search-optimization tool maker's estimate, and the topic classification is the author Stox's judgment. Schindler's remarks come from Latent Space's summary.
[This week] AWS's November 2 weekly roundup says OpenAI's GPT-6.1 Terra is generally available on Amazon Bedrock, priced at least 50% below GPT-6.1 Sol.
How we checked
Its quality rests only on OpenAI's own evaluation, which no one else has repeated. GPT-6 Astra is the flagship of the GPT-6 family, and Terra is the model meant for everyday work; the roundup doesn't say where GPT-6.1 Sol fits. The roundup gives no absolute per-token prices. The AWS launch announcement the roundup links to is dated October (AWS launch announcement); the roundup doesn't give the exact day.
[Looking back] (SecondSource call, July 14) On whether synthetic data has a ceiling, SecondSource split fields in two by whether answers can be checked automatically.
How we checked
The old problem is that a new model trained only on an old model's output can't outgrow the old model; SecondSource calls this "the student can't surpass the teacher".
(card)
Sources & accounting
The past 24 hours. 470 new pieces came in from November 2 to November 3: 285 academic papers, 115 social-platform posts, 34 company and personal blog posts, 13 company filings, 11 from a second academic-paper feed (counted separately from the academic papers), 8 subscription newsletters, 3 linked articles and 1 show transcript. Of those 470, 185 were read in full and 127 were screened out as off-topic (1 of them before fetching). Another 4 were logged with a stated reason for not reading closely, and 154 haven't been read yet.