Daily at 8AM KST · Summaries and takeaways from 8 AI articles
We cross-check industry press like TechCrunch and The Decoder, official announcements from OpenAI and Google DeepMind, and community signal from Hacker News.The point isn't what happened but why it matters, tied into one read on the day.
115 days running
1124 stories · 313 terms explained · since 2026-07-23
·what we picked →
2026-11-12
· ~7 min read
· Past edition
8 picked from 85 candidates · ordered by significance
Today's Insight
Nvidia is being tested on three fronts at once: its CEO, chips reaching China and its lead
Nvidia CEO Jensen Huang is caught up in allegations about a relationship with Alice Huang, whom U.S. intelligence reportedly considers a Chinese intelligence asset, and WIRED says Nvidia's stock has fallen about a quarter in a month. Nvidia's board has opened its own investigation. The Singapore cloud provider Megaspeed, where Alice Huang worked, is a major Nvidia customer, and the U.S. has worried that it was diverting chips to Chinese intelligence. Separately, Bloomberg reported, as relayed by AI Times, that the Chinese AI company Moonshot AI appears to have planned to rent export-controlled Nvidia B300 servers inside China. The two cases involve different companies, and both bear on whether Nvidia chips are reaching China.
Continue reading (2)Show less
Anthropic told investors that AI is lowering Nvidia's strongest barrier. Co-founder Tom Brown said Anthropic trains and runs Claude on four chip families, and claimed its own model Mythos cut the work of moving to AMD chips from six to twelve months to three days. That figure is Anthropic's own claim and has not been independently checked.
TSMC will invest up to $1 billion in OpenAI and Anthropic, both big buyers of the chips it makes, and Ars Technica placed the deal within the spreading across the AI industry. OpenAI's third-quarter revenue crossed $10 billion for the first time, but according to The Decoder its adjusted operating loss also grew to $10.9 billion. OpenAI, one of the two companies TSMC is backing, is losing more even as its revenue rises fast.
Signal to watch
Watch for the result of the investigation Nvidia's board has started.
Google launched Gemini Decision 1, a model that scores answer options instead of writing text, and the Gemini Decisions API that runs it. One input can carry up to 1,000 yes/no, multiple-choice and rubric-scoring questions, and the API returns a probability and a 0-to-1 confidence for each answer in 150 to 250 milliseconds. It costs $0.10 per million input tokens with no charge for output tokens, which Google says is nearly 5x cheaper than running Gemini 3.5 Flash-Lite through the Interactions API. It is available now through the Gemini API in Google AI Studio.
Why it matters
Why It Matters
Inside Google, Gemini Notebook uses the API to send each question to the right model tier, cutting latency by up to 40% on everyday queries. According to The Decoder, the startup Jev released a similar model first, and OpenAI, Cloudflare and Microsoft quickly followed.
What to do now
If you build software, you can get an API key in Google AI Studio and try the API on one task with fixed options, such as sorting customer inquiries.
Google Labs launched Tabletop, an experiment where several people work together on one AI-built screen. You add links, files and ideas, then call @tabletop or pick a starter template, and it builds interactive pieces such as trip itineraries, dinner polls, expense splitters and tournament brackets. You drop the invite link in a group chat, and everyone edits at the same time, as in a Google Doc. For now only people 18 or older in the U.S. can sign up, through a waitlist.
Why it matters
Why It Matters
Most AI chatbots answer one person at a time; Tabletop puts the AI in a space the whole group edits. Its templates cover chores the group's organizer usually does by hand, such as schedules and splitting costs.
Nvidia's stock has fallen by about a quarter in the past month, amid allegations that China used a on CEO Jensen Huang. Just before Halloween, The New York Times reported that Huang allegedly had an affair, starting sometime between 2024 and 2025, with Alice Huang (no relation), a former executive at the Singapore cloud provider Megaspeed, and that U.S. intelligence believes she was a Chinese intelligence asset. China's foreign ministry called the story groundless slander, and Megaspeed said U.S. authorities found no evidence of violations. Nvidia's board has opened its own investigation and says it has no reason to believe national security or Nvidia secrets were compromised.
Why it matters
Why It Matters
WIRED says Megaspeed being a major Nvidia customer makes the case more complicated, and reports have circulated for at least a year that the U.S. worried Megaspeed was diverting chips to Chinese intelligence. According to The New York Times, Secretary of State Marco Rubio briefed President Trump in May, right before Huang joined him on Air Force One for a trip to China, and Trump says he did not know.
TSMC plans to invest up to $1 billion in OpenAI and Anthropic combined. According to Reuters, its board approved up to $500 million in OpenAI preferred stock reserved for large investors and $500 million in Anthropic common shares at its expected IPO price. Both companies buy huge quantities of chips made by TSMC. TSMC reported more than $23 billion in profit for the July to September quarter, its sixth straight record.
Why it matters
Why It Matters
The Wall Street Journal reported that analysts warn among investors, suppliers and customers has made real demand for AI hardware hard to judge. Ars Technica notes the amount is small next to Nvidia's $30 billion in OpenAI and $10 billion in Anthropic.
Bloomberg, citing anonymous sources, reported that OpenAI's third-quarter revenue crossed $10 billion for the first time. That is a jump of almost 50% from $6.7 billion in the second quarter, far above the roughly 18% growth between the first and second quarters. Its adjusted operating margin improved from -150% to -109%, and its has reached $50 billion. Anthropic reported $17 billion in revenue for the same quarter.
Why it matters
Why It Matters
According to TechCrunch, OpenAI has benefited from tighter pricing, discounts, better models and more enterprise agent products, but it still trails Anthropic in the enterprise market. According to The Decoder, its adjusted operating loss also grew over the quarter, from $10.1 billion to $10.9 billion. The two companies count annualized revenue differently, since Anthropic includes sales made by its cloud partners, so the run-rate figures should not be compared directly.
Google DeepMind unveiled AlphaGenome 2, an AI model that predicts how DNA changes affect different kinds of cells. Almost every cell in the body carries the same DNA, but heart muscle cells, skin cells and infection-fighting immune cells switch different genes on, so the same can act differently in each. DeepMind says the model, trained on more than 100 million individual cell profiles across 68 tissues, finds nearly twice as many variants that change gene expression as its previous best model, and it can handle new cell types and states without retraining. It is available for non-commercial research through the AlphaGenome API, with a limited private preview for commercial research.
Why it matters
Why It Matters
DeepMind's genomics lead said drug developers target specific proteins and specific cell types, which is why cell-level prediction matters. The model costs more to run, so it could not be used for the September Atlas covering every possible DNA letter change, and it will not replace the original AlphaGenome any time soon. According to SiliconANGLE, it is not approved for clinical or diagnostic use.
Anthropic claimed in an investor presentation that it got Claude running on AMD GPUs in three days. Bringing up a new chip normally takes six to twelve months, but Anthropic's top model Mythos reportedly wrote or ported nearly all of the , the small programs that run on the chip. Co-founder and Chief Compute Officer Tom Brown said Anthropic trains and runs Claude on four chip families: AWS Trainium, Google TPUs, Nvidia GPUs and AMD GPUs. The three-day figure refers to prompt processing on AMD MI355X GPUs.
Why it matters
Why It Matters
is Nvidia's chip-specific software environment, long its strongest advantage because it keeps developers tied to Nvidia hardware. If the claim holds, The Decoder notes, companies could switch chip suppliers on price and availability, which would pressure Nvidia's margin of more than 70%. For now it is a single company's own claim, made to investors.
Bloomberg found financing documents showing the Chinese AI company Moonshot AI was to receive computing services from 32 servers using Nvidia B300 chips, which are under U.S. export controls. The servers are listed in Qingyang in Gansu province and were pledged as collateral in an August loan between their owner, Lingmeixin Technology, and a Beijing bank. They were valued at $64.8 million in total, about $2 million each and roughly 350% above the usual U.S. price. Moonshot denies any involvement, and after Bloomberg asked for comment on November 3, Moonshot's name disappeared from the public financial documents.
Why it matters
Why It Matters
The documents do not prove that Moonshot actually used B300 servers, and how the servers reached China or whether the U.S. granted a separate export license is unknown. Nvidia and Supermicro both distanced themselves, saying the companies involved are not their customers. This is not the first such claim: in July, reports said Moonshot's Kimi K3 was built on a cluster of about 20,000 Nvidia chips.
Read this far? We'll send tomorrow's too
The same briefing you just read, every day. A weekly synthesis on Sundays and a monthly report at the start of each month come with it. Sent at 8AM KST, which is the evening before in the US.
Free · no ads · one-click unsubscribe
We use your email only to send the briefing, stored and sent via services in Japan and the US. Privacy policy
Check your inbox
Find this subject in your inbox and press the link to start your subscription.