Silicon Valley Daily
Thursday, October 29: about $500 billion of Alphabet's market value disappeared in a single session, its worst day ever, after the earnings miss and a larger-than-expected cash burn. The selling spread: Intel fell 21% on PC weakness and ServiceNow 27% on AI-disruption fears, while Amazon's trailing free cash flow went to -$42.4 billion even as AWS grew 39%. Binance suspended withdrawals after a hack it puts at about $6.3 billion, which TRM Labs attributes to North Korea's TraderTraitor; Binance says it will cover all losses. Reddit said it will move to a logged-in experience as Google search traffic shrinks, and The New York Times reported US agencies surveilled a suspected Chinese asset close to Jensen Huang, which Nvidia disputes. Thrive Holdings agreed to buy ADT in a $15.3 billion deal, and Microsoft priced $25 billion of debt.
Alphabet falls nearly 13%, its worst day ever, erasing about $500B
Alphabet shares fell nearly 13% on Thursday, the worst day in its history, wiping out about $500 billion of market value, CNBC reports. Investors reacted to a larger-than-expected negative free cash flow, unchanged spending guidance and weaker-than-expected revenue from the core search business. Ars Technica reports Google acknowledged AI search makes less money and plans more ads to fix it.
Binance hit by roughly $6.3B hack; TRM Labs blames North Korea
Binance suspended all deposits and withdrawals after a hack it says drained around $6.3 billion from its wallets, potentially the largest crypto theft ever, and pledged to cover all losses. TRM Labs attributes the attack to North Korea-based TraderTraitor, saying about $6.36 billion was taken across 10 blockchains, mostly within 20 minutes. Tether and Circle froze billions of dollars of their coins, the Financial Times reports; Binance.US says it is unaffected.
Amazon misses as free cash flow hits -$42.4B, but AWS grows 39%
Amazon's third-quarter revenue rose 12% to $201.9 billion, just below the $202.2 billion estimate, and EPS of $1.89 missed $1.97. AWS revenue grew 39% to $45.98 billion after deals with Meta and OpenAI, but trailing twelve-month free cash flow was -$42.4 billion, versus +$14.8 billion a year earlier. Amazon held its $220 billion 2026 capex plan, and Andy Jassy said spending will be meaningfully higher in 2027 and that AWS can become a $1 trillion business.
Reddit will move to a logged-in experience as search traffic falls
Reddit said it will become a primarily logged-in experience, one of the biggest shifts in its 21-year history, starting with tests outside the US. CEO Steve Huffman said the change will improve the experience, support ads and stop bots from scraping its data; he told analysts Google's AI Mode does not drive traffic and search traffic was down meaningfully last quarter. Reddit shares fell on weak sales guidance, CNBC reports.
NYT: US surveilled suspected Chinese asset close to Jensen Huang; Nvidia disputes
The New York Times reported that US intelligence agencies surveilled Alice Huang, an executive at Singapore AI cloud firm Megaspeed, for about a year over concerns she was a Chinese spy; she was photographed at business dinners with Nvidia CEO Jensen Huang. Megaspeed had been accused of illegally selling Nvidia chips to China. Ars Technica says the report claims Trump ignored the intelligence; Nvidia said diversion allegations were investigated and debunked.
Intel plunges 21% on weak guidance as PC slowdown drags
Intel shares fell 21% after it issued weaker-than-expected fourth-quarter guidance, citing PC softness that CFO David Zinsner expects to last at least through the first half of next year, CNBC reports. CEO Lip-Bu Tan predicted a datacenter CPU boom and said demand for budget Wildcat Lake laptop chips is outstripping supply. Intel is taking Panther Lake capacity offline to build more Granite Rapids server CPUs.
ServiceNow drops 27%, its worst day, on AI-disruption fears
ServiceNow shares fell 27% on Thursday, their worst day ever, after underwhelming third-quarter results revived fears that AI will disrupt its software business, CNBC reports. Investors are looking for evidence that companies' heavy AI spending is paying off.
The Information: DeepSeek doubled its ~$16B round as black-market Blackwell prices surged
DeepSeek doubled the size of its latest funding round, which reached nearly $16 billion this month, because black-market prices for Nvidia Blackwell chips surged, The Information reports, citing sources. Chinese state-owned investors account for 27% of the round.
China-linked hackers use AI tools to hit 200+ organizations in 11 days
CrowdStrike says Chinese state-linked hackers used the AI tool ARTEX, powered by a DeepSeek model and another model, to target more than 200 organizations in 11 days and compromise at least nine, NBC News reports. Separately, The Register reports suspected Chinese attackers built a custom PHP2Shell exploit chain that rooted 25 servers, including on a US government-affiliated network where they likely still have access.
OpenAI-backed Thrive Holdings buys ADT in $15.3B deal
Thrive Holdings, the holding company spun out of Josh Kushner's Thrive Capital, is acquiring security firm ADT for $10 a share, about $7.6 billion in equity and $15.3 billion in enterprise value, TechCrunch reports. OpenAI is a partner and investor in Thrive Holdings, which aims to hold companies long term and rebuild them with AI; it now owns more than 75 companies.
Microsoft prices $25B of new debt for its AI build-out
Microsoft announced pricing for $25 billion in new debt as it pursues its AI build-out, CNBC reports. Unlike Meta, Alphabet and Amazon, Microsoft has kept positive free cash flow and a strong credit rating; finance chief Amy Hood had told analysts the company has access to debt markets but had not tapped them.
Contractor's personal Claude agent ran on a Pentagon network for two months
An Accenture employee connected a secure centralized system used by multiple intelligence agencies to their personal Claude Code account, and the agent ran on a Department of Defense network for two months before it was identified and the vulnerability patched, sources told Wired.
OpenAI doubles ChatGPT Pro limits after backlash over planned cut
OpenAI reversed a planned 50% cut to its ChatGPT Pro usage limits after user pushback and is instead doubling limits on its $100, $200 and $500 Pro plans, BleepingComputer and The Decoder report. It also promised to keep Dots usage separate. The New Stack reports Sam Altman considered dropping token-based pricing for OpenAI's next model before backing off.
Mistral releases weights of a trillion-parameter model aimed at cybersecurity
Mistral released the weights of its trillion-parameter AI model and is targeting cybersecurity, TechCrunch reports. Mistral argues that an open model gives organizations a degree of control over their cyber defenses that closed models cannot.
Trump declares robot imports a national security threat
President Trump declared imports of foreign-made factory robots a national security threat, ordering trade talks and threatening tariffs, Ars Technica reports.
Sandisk revenue jumps 378% to $11B; 8-for-1 stock split
Sandisk reported fiscal first-quarter revenue of $11 billion, up 378% year on year and above the $10.7 billion estimate, forecast second-quarter revenue above expectations and announced an 8-for-1 stock split, MarketWatch reports. The shares still slipped more than 4% after hours; the stock has risen more than 2,600% since the company split from Western Digital 20 months ago.
OpenAI updates its Model Spec; Anthropic study shows experts absorb AI opinions
OpenAI published an updated Model Spec built around five core values, including sound judgment, human agency and faithfulness to human intent, and described layered monitoring it uses to detect misalignment and unauthorized agent behavior. Separately, an Anthropic study gave 50 experts an AI agent whose preferences were set by coin toss; they followed its view up to 86% of the time on judgment calls, and most believed the conclusions were their own, The Decoder reports.
GM pays $220M to settle claims it sold drivers' data
General Motors agreed to pay $220 million to resolve a complaint from a coalition of 36 US attorneys general that it illegally sold driving and location data to data brokers through its discontinued Smart Driver program. GM must limit the data it collects and get drivers' express consent before collecting or sharing it.
Anthropic's S-1 flags heavy dependence on TSMC chips made in Taiwan
Anthropic's S-1 says it depends heavily on TSMC chips made in Taiwan and warns of risks from potential disruption amid geopolitical tensions with China, the Wall Street Journal reports.
Polymarket halts trading after a third-party breach drains accounts
Prediction market Polymarket suspended trading on Thursday after hackers compromised a third-party service and distributed crypto-draining JavaScript; about $2.5 million was traced to a suspected attacker wallet, BleepingComputer reports. Trading has resumed, and Polymarket says affected traders will be made whole.
Kalshi and its co-founders face felony charges in Utah
Prediction market Kalshi and its co-founders face felony charges for violating Utah gambling laws, Ars Technica reports. The charges come as the Trump administration sues Utah to save Kalshi from the state's total ban on gambling.
&Truth launches with $70M seed to automate formal verification
&Truth launched with a $70 million seed round to automate formal verification, which uses math to prove software works as intended, Wired reports. It is part of a wave of startups betting AI can make the painstaking process practical.
NYSE owner ICE plans US-regulated perpetuals with Hyperliquid operator
Intercontinental Exchange plans US-regulated perpetual futures markets based on Hyperliquid, working with Trade.xyz, one of the market operators on the offshore crypto platform, and is taking a stake in Trade.xyz, Bloomberg reports.
India's AM Intelligence seeks up to $3B for Nvidia Vera Rubin systems
Indian AI infrastructure firm AM Intelligence is seeking to raise as much as $3 billion to fund a previously announced purchase of Nvidia Vera Rubin systems, Bloomberg reports, citing sources.
China deals: Nano-core Chip raises RMB 1B+, Hangzhou backs 01.AI, Xiaomi buys into Sunwoda
Hangzhou AI inference chipmaker Nano-core Chip raised more than RMB 1 billion in a Series C2 led by JianTou Investment. Kai-Fu Lee's 01.AI secured hundreds of millions of RMB from state-backed investors in Hangzhou and Linping district and will set up headquarters in Linping. Xiaomi plans to invest RMB 1.7 billion for a 5.34% stake in EV battery maker Sunwoda Mobility.
Doubao Work adds Kimi, GLM and DeepSeek; Tencent details a world model
ByteDance's Doubao Work now offers third-party models — Kimi K3.1 Pro, GLM-5.3 and DeepSeek-V4.1-Flash — to team and enterprise plans, billed from existing AI credits. A pre-release of DeepSeek Harness lets its main agent hand tasks to a locally installed Codex CLI or Claude Code. Tencent Hunyuan detailed Hy-Nomas, an open 8B interactive video world model, with a 280-case benchmark.
Chinese optical suppliers extend sell-off despite strong earnings
Shares of major Chinese optical communications suppliers extended a two-day sell-off despite upbeat AI-driven third-quarter earnings, SCMP reports. Eoptolink dropped as much as 15% in morning trading, TFC Communication slid 10.3% and Zhongji Innolight fell over 7%.
Hugging Face crypto-mining malware returns; flaws found in five AI dev tools
A crypto-mining campaign that abuses Hugging Face models returned days after a takedown, hijacking GPU servers and planting backdoors; one model was downloaded more than 28,000 times, BleepingComputer reports. Separately, researchers found bugs in five AI developer tools rooted in embedded browsers, with OpenAI and Microsoft flaws still unpatched, The Register reports.
Archive of Our Own could be offline for up to four weeks after security incident
Fan-fiction site Archive of Our Own said it could be offline for up to four weeks after a security incident. Within two hours of the announcement, its status page posts drew millions of views, TechCrunch reports.
Rivian's R2 becomes its best seller; VW venture adds an AI assistant
Rivian said its new R2 SUV is already its best-selling vehicle and raised its guidance for the second time this year, TechCrunch reports. It also expanded its $5.8 billion joint venture with Volkswagen to include an AI assistant, without saying whether the venture's value will change.
Twilio and Pinterest drop after hours on soft outlooks
Twilio's third-quarter revenue rose 20% to $1.56 billion, ahead of estimates, but it forecast slower fourth-quarter growth and its shares fell more than 15% after hours, MarketWatch reports. Pinterest revenue rose 12% to $1.17 billion, short of the $1.21 billion estimate, with monthly users up 10% to 661 million; its shares fell more than 8% on a weak fourth-quarter forecast.
AI seed rounds: Dream ($17M), SuperMe ($20M), Machenta ($40M)
Former Palantir executives launched Dream with $17 million from a16z to build construction world models trained on more than 100 million square feet of building data. SuperMe raised $20 million for a network of AI profiles built from real experts' knowledge. Creandum and USV co-led a $40 million seed into Machenta, a robotics lab with no product yet, Sifted reports.
Former Sequoia India partners launch Mettle Capital with $350M
Former Sequoia India partners launched Mettle Capital, an India-focused firm, with a $350 million fund, Newcomer reports. It follows an acrimonious split with the spun-off Sequoia unit.
Amadeus breach exposes data of about 92,000 Japan Airlines passengers
Japan Airlines said a data breach at Amadeus's Passenger Service System exposed the data of about 92,000 customers, BleepingComputer reports.