Silicon Valley Daily
“I was unfaithful to my wife,” Jensen Huang wrote to Nvidia staff on Sunday, as the company’s board opened an independent probe into whether he was the target of a foreign intelligence effort, following a New York Times report on a classified memo drafted under Marco Rubio. President Trump said Huang should stay on as CEO, while lawmakers in both parties want him to brief Congress on Nov. 18. The compute race kept moving: Anthropic reportedly paid $3.4 billion for 1.7GW of gas generators and Apollo plans $75 billion of AI chip purchases. The AP reported the Commerce Department’s AI safety center has been asked to report informally to the White House, and October data showed professional services adding 70,000 jobs.
Nvidia board opens probe as Jensen Huang admits an affair
Nvidia’s board said on Sunday it is opening an independent investigation, overseen by a committee of independent directors with outside counsel, into reports that Jensen Huang was the target of a foreign intelligence effort; it said it has no reason to believe national security or Nvidia secrets were compromised. Hours earlier Huang told staff in a memo, “I was unfaithful to my wife,” said he had “fallen short” of his responsibility to employees, the board and shareholders, and pledged to cooperate with the inquiry. President Trump said Huang should stay on as CEO, telling reporters, “I haven’t heard that he’s done anything wrong.”
NYT: Rubio-led memo tied Jensen Huang to an alleged Chinese spy
Marco Rubio, as national security adviser, oversaw a highly classified memo this spring assessing that a woman with a close relationship to Jensen Huang worked for China’s Ministry of State Security, the New York Times reported; when the name was unmasked it was Huang, and President Trump did not act. Sen. Chris Coons wrote that “Congress will investigate and get answers,” three House Republican chairs asked Huang for a closed-door briefing on Nov. 18, and Rep. Seth Magaziner wants Commerce Secretary Howard Lutnick to testify alongside him. Nvidia told the Times Huang’s interactions with Alice Huang were consistent with meetings with any customer.
FT: Anthropic pays $3.4B for 1.7GW of gas-fired generators
Anthropic bought commitments for 1.7 gigawatts of gas-fired generators from equipment supplier OnePWR for $3.4 billion, the Financial Times reported, citing sources. The FT called it the largest known purchase of its kind by an AI lab, with Anthropic bypassing traditional suppliers to secure scarce equipment. The deal lands as voters in several states turn against new data center construction. Separately, The Information reported that amid a server shortage Anthropic has been asking where to find 5 to 10 megawatts of power in older buildings, reviving interest in decades-old telecom facilities.
Bloomberg: Ternus presses Federighi to turn Siri into a true AI agent
Apple CEO John Ternus wants more urgency in the company’s AI work and is pushing software chief Craig Federighi beyond Siri tweaks toward an agent that can compete with OpenAI’s Dots and Meta’s Muse, Bloomberg’s Mark Gurman reported. iOS 27.3 is expected to let Siri control Apple’s own apps, such as cropping images in Photos and organizing Safari tabs. Third-party app control, like ordering through Grubhub or booking on OpenTable, is likely to follow in the spring.
The Information: AI companies sold or priced $300B+ of stock in 2026
Excluding Anthropic’s IPO, AI companies have sold or priced more than $300 billion of stock in 2026, with about $100 billion more on tap, The Information reported. Last week’s volatile trading in AI stocks has raised the question of how much more supply the market can absorb.
WSJ: THORChain moves Binance hack proceeds despite FBI pleas
THORChain, which handled about $1 billion of stolen Bybit funds in 2025 and earned more than $12 million in fees, has begun moving proceeds from the Binance hack despite pleas from the FBI, the Wall Street Journal reported. The protocol says it cannot stop transfers. Some crypto developers told the paper they are unable to stop North Korean hackers from laundering funds through their platforms.
All-In: Chamath says a data center freeze makes hyperscalers kingmakers
On All-In’s midterm episode, Chamath Palihapitiya said incoming governors will want measurable economic impact before approving data centers, a slowdown that helps hyperscalers with existing capacity and hurts neocloud renters and model labs. He said the SpaceX–Anthropic compute deal has a mutual 90-day exit, and that SpaceX ending it and repricing at spot rates two to three times higher would be “the kingmaking move.” David Sacks said it is “100%” that Democrats push major AI legislation, predicting a federal AI department that must approve frontier models before release. In a follow-up clip, Palihapitiya said “I got it totally wrong,” arguing the election showed wage earners believe “the economy is broken” while asset owners benefited at both zero and high interest rates.
The Information: Apollo plans $75B of AI chips; Nvidia courts Brightspeed sites
Apollo’s planned $75 billion in AI chip purchases would exceed the chip financing of every neocloud and leave the private equity firm behind only the major hyperscalers, The Information reported. In a separate exclusive, it said Nvidia’s proposal involving telecom operator Brightspeed’s facilities is part of a push to lock up data center capacity and power before rivals. The outlet also reported that OpenAI’s GPT-6.1 Sol performs comparably to GPT-6 Astra at about a fifth of the per-task cost, even as Nvidia hardware gets pricier.
Amsterdam police arrest 150–200 protesters at a Microsoft data center site
Amsterdam police arrested 150 to 200 protesters on suspicion of trespassing at a Microsoft data center site owned by Pure Data Centers, after some entered the construction site by force; police said they seized break-in tools and found allegedly stolen items. The action follows a breach of the site the previous Friday and is the first of several planned by the Stop Dirty Data Centers coalition, with protests also slated for Dublin, Warsaw and Denmark. Dutch broadcaster NOS identified the group as Fossielvrij, which opposes fossil-fuel power for data centers.
Thinking Machines reportedly nears $5B round; Inkling 2 Small swears in its reasoning
Mira Murati’s Thinking Machines Lab is closing in on a $5 billion round at a $40 billion pre-money valuation led by Accel, Business Insider reported, according to a post by Ornn Data. Separately, Vals AI read Inkling 2 Small’s reasoning logs from 2,655 runs across 13 benchmarks and found expletives in 50 runs, 316 in total, 241 of them from two runs that spiraled. The swearing is not a useful failure signal: 96% of failed runs contained none.
AP: US AI safety center asked to report informally to the White House
Officials at CAISSI, the Commerce Department division that assesses new AI systems, were recently asked to informally report to a White House office, two former officials told the Associated Press. The center has a $10 million annual budget, about one-eighth of its UK counterpart’s, has had no permanent director since spring and has stopped publishing its evaluations. Critics worry findings could be muted in deference to industry; the White House pointed to Trump’s new Super Intelligence Force.
SemiAnalysis: Inner Mongolia campus built 120MW in a year at a fifth of US cost
SemiAnalysis visited Galaxy, a campus in Ulanqab, Inner Mongolia planned for up to 800MW of IT capacity, and reported its first 120MW phase was delivered within a year at a fifth of US costs. The campus is owned by Envision, which has installed more than 100GW of wind turbines and builds its own wind farms and power line for the site; all four phases are built for two Chinese internet giants. A local report cited nearly 30,000 construction workers across 22 data center projects in the surrounding zone.
Politico: how a Signal-chat AI safety network beat a $100M super PAC
Politico traced how a small, wealthy AI safety network that organized in Signal chats grew into a coalition that beat Leading the Future, the $100 million super PAC backed by AI industry figures, in the midterms. Bloomberg separately reported that AI safety advocates made gains after a $110 million midterm fight. Anthropic-backed Public First spent more than $63 million, and 10 of its 17 House candidates won.
FT: Apple saved $2.7B in tax by routing profits through Singapore
Apple saved $2.7 billion in taxes this year by routing up to $50 billion in profits through Singapore, the Financial Times reported from new filings. That exceeds the $2.6 billion it saved through Ireland, making the city-state the iPhone maker’s biggest tax shelter.
WSJ: US professional services added 70,000 jobs in October
US professional services added 70,000 jobs in October, the most since July 2022, with consulting, legal and accounting all growing, the Wall Street Journal reported. Those are among the fields seen as most exposed to AI, and the numbers cut against fears of a white-collar job collapse. The Washington Post reached a similar reading, saying the hiring report suggests AI is not yet taking white-collar jobs.
UK minister says data centers will continue as essential infrastructure
UK growth minister Dan Tomlinson told the Financial Times data center construction will continue as “essential infrastructure,” while projects will need to invest in local communities and create jobs to win government support. He said the UK had learned from the “free-for-all in parts of the US.” The UK Compute Roadmap calls for at least 6GW of data center capacity by 2030.
Elvis and Eminem rights holders sue Suno over AI clones
Rights holders of Elvis Presley’s music and the publisher of some of Eminem’s biggest records have sued Suno, alleging it illegally generates music and lyrics that mimic the artists. The Presley suit invokes Tennessee’s ELVIS Act, which bars AI companies from cloning a singer’s voice. Suno said it does not allow prompts naming specific artists or songs; it already faces suits from Sony, UMG, Jason Isbell and Reservoir Media.
Guardian: leaked documents show Cisco deepened Israeli military deals
Internal documents leaked by Cisco workers opposed to the company’s ties to Israel’s military show how it moved to shore up the relationship as the Gaza war continued, the Guardian reported. A September 2025 draft memo said repeated multi-hour outages on Israeli military intelligence networks had been “deeply impacting customer sentiment.” The company pressed ahead even as it faced internal worker protests over its role in the war.
FT: Sequoia partners were blindsided when Roelof Botha joined SpaceX’s board
The Financial Times says Sequoia’s new leaders Alfred Lin and Pat Grady are unwinding Roelof Botha’s legacy, halving the number of GPs and closing his permanent capital fund and London office. A source said Botha joined SpaceX’s board this summer without asking compliance or getting partnership sign-off, even though the role could complicate trading Sequoia’s $20 billion SpaceX stake. Sequoia said it continues to collaborate closely with Botha.
FT: crypto industry braces for Democratic investigations
The crypto industry is bracing for a reckoning after Democrats won the midterms and vowed to investigate President Trump’s family and its ties to the industry, the Financial Times reported. The shift comes days after Treasury moved to cut exchange MEXC off from the US financial system.
Tencent’s Yao Shunyu warns AI will soon reach dangerous biology skills
Tencent chief AI scientist Yao Shunyu, formerly of OpenAI, said at a Hong Kong science symposium on Saturday that AI will soon be able to master biology. “The cyber abilities of models and soon the bio ability of models will reach a threshold where it’s actually possible to do harm,” he said, the South China Morning Post reported. It was a rare public appearance for Yao.
Karpathy builds HackerTube in a weekend with Claude
Andrej Karpathy used Claude Opus 5.5 to filter 43 million comments from 20 years of Hacker News into 6,454 recommended videos, The Decoder reported, and says the result beats YouTube’s own recommendations. He then built a second site with 47,000 videos, “one-shotted with Claude Code in 3 minutes.” His only manual steps were a Google Cloud API key, a rented server and buying the domain.
Replit’s Masad: a scoped $10M project was done for $250K
In a Y Combinator conversation with Paul Graham, Replit CEO Amjad Masad said large companies are using Replit to replace legacy software they paid millions for, starting with internal tools. One customer told Replit a project it had scoped at $10 million cost $250,000 on the platform, and Masad said “the outsourcing industry is sort of dead.” He said Replit no longer describes itself as vibe coding but as a platform for “self-driving companies.”
Robot training data startup GenRobot AI raises $100M Series A+
Beijing-based GenRobot AI, which builds equipment and software to record human demonstrations for training robots, has raised a $100 million Series A+, TechNode reported. Investors include SAIC Motor, Baidu, Shoucheng Holdings and Galaxy General Robotics, alongside GL Ventures, CITIC Goldstone, CCB International and Huatai Zijin. Its Matrix platform syncs, cleans and labels recordings of household, manufacturing and logistics tasks.
Google DeepMind ships MuJoCo 4.0, its first major release since 2023
Google DeepMind released version 4.0 of MuJoCo, the physics engine widely used for robotics simulation, on Nov. 6. The headline feature is speculative contacts, which stop fast-moving objects from tunneling through thin obstacles within a single timestep; the release also adds elasto-plastic deformation for springs and continuum bodies.
Iris Ohyama breach may expose data of 7.7 million customers
Japanese manufacturer Iris Ohyama, which runs the Iris Plaza shopping platform, disclosed that personal information of about 7.7 million people may have been exposed after unauthorized access to a third-party cloud service. The company found a suspicious file in the external cloud used by the store and suspended Iris Plaza and related systems; it first disclosed the incident on Oct. 29.
Alibaba benchmark: best coding agent fully solves 26.4% of real tasks
Alibaba and several Asian universities built Co-Design Bench from real user change requests that coding agents had already struggled with. GPT-6 Astra, the best of ten models tested, met about 70% of each task’s requirements on average but was fully correct on only 26.4% of tasks. The same models score above 99% on older benchmarks.
Sega’s ALL.Net hack leaves Japanese arcade games dark for a week
ALL.Net, the arcade network Sega launched in 2004, suspended service on Nov. 1 after its servers were hit by unauthorized access, leaving networked Sega and Bandai Namco machines in Japanese arcades offline. The operator said personal and billing data were not on the compromised server. Bandai Namco titles have begun resuming on a trial basis, while Sega games remain offline with an update promised for Nov. 10.
Analog Devices completes its acquisition of Alif Semiconductor
Analog Devices has completed its acquisition of Alif Semiconductor, adding Alif’s AI-native microcontrollers and fusion processors to its sensing and signal-processing portfolio. The deal targets edge computing in industrial automation, robotics, defense, energy and wearables. Analog Devices reported more than $11 billion in revenue in fiscal 2025.
Diamond Foundry shows a 200mm single-crystal diamond wafer for chip cooling
Diamond Foundry announced what it calls the world’s first 200mm single-crystal diamond wafer, the size fabs use for older processes, aimed at cooling hot AI chips. Diamond conducts heat at least five times better than copper while staying electrically insulating. The sample shown is still unpolished, and the company says a 300mm wafer is possible before 2030.
The Information: Brian Chesky is self-funding his AI lab for now
Airbnb CEO Brian Chesky is funding his AI lab himself for now, The Information reported. The high cost of AI talent and chips could eventually push the lab to raise outside money.
Trending on X
What X was arguing about today that the news hasn't covered. Views are the posters' own.
Sequoia’s Maguire: Huang acted “incredibly dumb” but is irreplaceable; Musk agrees
Sequoia partner Shaun Maguire posted a five-point take: Huang is one of the greatest entrepreneurs of his generation, did something “incredibly dumb” that “was bad for America,” likely had “no bad intention,” and is “irreplaceable for America’s AI leadership.” Elon Musk replied “I agree.” Alec Stapp argued Trump will not act on the national security risks, so Congress should use its subpoena power and pass the OVERWATCH Act.
Trump says he is an AI bull because Xi is “all-in” on superintelligence
The White House Rapid Response account posted a clip of Trump saying one reason he is “an AI bull” is that President Xi, after their meeting two weeks ago, is “all-in to SI,” adding “Whoever wins that race wins.” In a second clip he said that if the US does not have superintelligence, China will, which “would put us at an unbelievable disadvantage,” and that “the risk can be managed.”
Victor Taelin’s $8M raise at a $120M cap draws mockery; he steps back
Victor Taelin, whom supporters credit with building the HVM runtime, said his company is raising up to $8 million at a $120 million post-money cap to research a successor to GPT-style architectures, and insisted the deck was not a joke. After critics mocked it, one saying Jeff Dean’s fundraising deck was more humble, he wrote that he would take a break from social media. Some defended him, noting Symbolica raised $31 million for symbolic AI with no product.
Kobeissi Letter: US data center construction spending hit a record $96.5B rate
The Kobeissi Letter posted that US data center construction spending rose 90% year over year in September to a record annualized rate of $96.5 billion, the largest annual increase in 10 years. It said the figure has now risen more than 60% year over year for four consecutive months. The post landed the same week voters in several states turned against new data center projects.