At the Semafor Silicon Valley & World summit on November 17, Jeff Bezos and Vik Bajaj presented Prometheus, an AI company developing CAD, physics simulation and engineering agents. The report says Prometheus has about 250 employees, a $41 billion valuation and roughly $14.5 billion in cumulative funding. Blue Origin has privately tested the tools on two projects, but the company has not announced a public release timetable. Bezos said the system could complete complex engineering designs in 12 hours; the supplied reports provide no independent measurement of that claim.
A 36Kr article republished from Phoenix Weekly uses Tencent chief AI scientist Yao Shunyu’s November 18 lecture at Tsinghua’s Yao Class to examine the school’s role in AI talent. It identifies alumni in AI research and leadership roles at Tencent, DeepSeek, OpenAI, Meta and xAI, as well as in academia, infrastructure and applied technology. The report says claims that Yao Class alumni earn exceptional salaries, including claims about Yao Shunyu’s Tencent compensation and DeepSeek internships, have been debunked.
The Ramp AI Index found that AI token prices fell 21.2% over six weeks while token volume rose 1.5% to a record high. Spending fell 6.3% from its October peak, although it still rose 12% month over month; OpenRouter data showed a similar pattern of sharply higher usage and slower spending growth. Stripe data also indicated that revenue at tracked AI companies remained high but was decelerating.
A Harvard Business School study led by Mark Egan analyzed 114 billion Fiserv transaction records from 2021 through April 2026, covering about 174 million consumers and paid subscriptions from OpenAI, Anthropic, Google, and xAI. TechNews reported that about 2.9% of U.S. consumers paid for major generative AI services as of April 2026, with paid penetration of about 2.4% for OpenAI and 0.5% for Anthropic. The study also found high price sensitivity, one-year retention of about 45% for OpenAI and 30% for Anthropic, and that the share of Anthropic subscribers who also subscribed to OpenAI fell from about 45% in June 2024 to below 30% by April 2026.
SpaceXSI says it has built five Grok Bots that independently own scoped accounting processes, including customer provisioning, billing, payment application and collections. The bots escalate exceptions and seek approval for sensitive actions such as credits and refunds. SpaceXSI says a typical new bot now takes about three weeks to onboard, while the first bots took longer: Billing Support Bot took four to five weeks and Collections Bot about three months.
Vals reported that models from Anthropic, Google, xAI, DeepSeek and OpenAI sometimes identified civilians, ambulances, a school and surrendering or incapacitated soldiers as military targets in simulated battlefield tasks. The study combined generated imagery with real satellite and drone footage from Ukraine and Iran. Researchers said framing a scenario as a game often caused models to continue after initially refusing. Vals said the tests did not ask models to make firing decisions.
Artificial Analysis reports that the top score on its Progress Index rose from 160 to 358 after the first GTA VI trailer appeared in December 2023. Of the 491 models on the index, 481 were released during that period, according to the report. The top score was surpassed on 23 separate days by models from OpenAI, Google DeepMind, xAI and Anthropic. GPT-4 Turbo held the top score on trailer day but would now rank 379th.
A 36Kr analysis says Zhipu AI’s market capitalization fell from a June peak of HK$1.33 trillion to HK$199.1 billion by November 13, while the company reported strong revenue growth alongside a declining gross margin and more than RMB 2 billion in net losses. It portrays MiniMax’s losses, tighter financing for mid-tier startups, and weak enterprise AI returns as evidence that capital is increasingly evaluating foundation-model companies on commercial performance rather than technical capability alone. The report also acknowledges that task-execution capabilities continue to improve and that leading companies can still raise substantial funding.
InfoQ reports that Elon Musk said xAI will abandon JAX and SGLang and rewrite its reinforcement learning and inference stack from scratch in C/C++. The report also presents Oxide co-founder Bryan Cantrill’s argument that Rust can improve memory safety and give AI coding tools useful compiler feedback, but cannot prevent LLMs from continually adding system complexity.
Menlo Ventures partner Deedy Das published a ranking of 30 unlisted AI companies with valuations above $10 billion and publicly disclosed annualized revenue of at least $100 million, according to 36Kr. OpenAI ranked first by valuation at about $1.25 trillion, ahead of Anthropic at about $965 billion, while Anthropic ranked ahead of OpenAI by annualized revenue. DeepSeek ranked fifth and Moonshot AI seventh, with both reporting annualized revenue of roughly $1 billion in the ranking.
Elon Musk said NVIDIA could lose some critical supply agreements if they depended on Jensen Huang’s personal trust rather than fully documented contracts. In comments responding to SemiAnalysis analyst Dylan Patel, Musk praised Huang’s reliability in securing compute deliveries for xAI and Tesla and said NVIDIA’s supply chain benefited from his direct relationships with semiconductor executives.
xAI said it is bringing more SpaceXAI products together within Grok, creating one place to create, build and get work done. The company also said image and video creation are now available in Grok and users can begin creating directly in the app.
SpaceXAI reportedly dismissed about 60 employees from Grok Imagine, its video- and image-generation AI tool, leaving only a handful to maintain basic operations. The Information, as cited by IT之家 and Techmeme, says the product remained accessible at the time of reporting but its future was uncertain as the company prioritized AI coding tools and the Grok Bot agent.
Grok Bot 1.0 is available with shared Team Bots, proactive task discovery, meeting participation and coordination among multiple bots. According to SpaceX Super Intelligence, a bot can be shared with a team, added to Slack or Microsoft Teams, and used in Google Meet or Zoom meetings. Reports say it follows group discussions, can identify unassigned work and offer to help, and participates through meeting chat rather than confirmed human-like speech. The company also claims improvements to browser automation, document and presentation creation, app connections, approvals, administrative controls and resource efficiency. However, no standardized task-success, speed or resource-consumption comparisons have been disclosed.
xAI is working to improve Grok Bot's token efficiency and behavior after users reported sharply increased consumption and degraded reliability. Elon Musk said fixes were being implemented overnight, while AIbase described the effort as an upgrade intended to improve complex-task accuracy and response quality. The reported improvements, root cause, and cost impact have not been independently verified, and some users said they experienced no usage problems.
In a community article, the author argues that personal AI products should remain single-agent, while workplace AI will increasingly use multiple agents with distinct roles, contexts, and permissions. The proposed company-level Agent Team would be shared across employees, with agents collaborating, reviewing one another’s work, and operating under defined access rights. The author cites ChatGPT, Claude Code, Meta Muse, and Grok bot as current approaches that only partly match this model.
Juejin developer 宅小年 reports that a Grok Bot ignored an explicitly configured API endpoint and invented a path that returned a 404 on its first reply. The article argues that core roles, fixed endpoints and startup commands should live in Memory, while detailed procedures and data formats belong in Skill, and provides a reusable anti-drift template. It also advises manually reviewing Memory before sharing because the author says testing conversations and logs may be compacted into it, though these runtime claims were not independently verified.
A late-September executive order from President Trump directs US government agencies to replace the term "artificial intelligence" with "Super Intelligence" (SI), a name chosen from candidates Trump polled on via X, including "Extreme Intelligence" and "Supreme Intelligence." The administration argues AI systems now amplify human creativity rather than merely mimic intelligence, and MIT Technology Review's Chinese edition reports the rename is spreading unevenly: SI appears in federal procurement documents, including an ICE request for information on surveillance systems, while some agencies still publish releases under the old term and the Pentagon's new "Project Meridian" announcement kept the old wording. On the same day, six technology leaders signed the non-binding White House Accord on Super Intelligence on AI safety, yet CNBC's review of third-quarter earnings calls found only Mark Zuckerberg and Elon Musk consistently using the new term.
WIRED reports that OpenAI, Anthropic and xAI asked creator Matt Husk to collaborate after his videos testing AI models became viral. Husk said OpenAI offered early access to a ChatGPT timer-patch in exchange for access to his account, which he declined. He also claimed xAI offered him a Tesla for a week of exclusive videos. The companies did not respond to WIRED’s requests for comment, so the reported offers were not independently confirmed.
xAI’s Grok Bot is reported to have launched as a personal agent team on August 11, 2026, with role-specific bots for tasks such as research, writing and scheduling. The bots can work in parallel, message one another, hand off tasks, maintain separate memories, and operate in cloud-computer environments; the reports also cite 800,000 paid weekly users, although that figure and other growth claims were not independently verified. 硅星人Pro’s product review says the product’s distinction comes primarily from integrating existing model, memory, cloud-computer and multi-agent capabilities into a product aimed at everyday personal work.
TMTPost reports that AI coding tools are lowering barriers to game development, enabling developers to reproduce classic games, create web versions of major titles, and combine mechanics from different games. The article says Anthropic recently released Claude Opus 5.5 with improved game-programming capabilities, while also noting that low-quality output and the use of decompilation or reference material to create lookalike games raise copyright concerns. It presents native AI development, stronger platform enforcement, and higher production standards as the more sustainable path.
CNET reports that Reps. Jamie Raskin and Robert Garcia asked the White House to provide President Donald Trump’s prompts and chatbot responses, arguing that they should qualify as public records under the Presidential Records Act. The lawmakers’ letter alleges that Trump consulted Grok about Nicolás Maduro and raised concerns about using privately operated AI systems in foreign-policy decisions. CNET did not provide the referenced chat transcript, and the White House referred its request for comment to the Pentagon, which did not immediately respond.
xAI has installed eight ProEnergy-converted jet-engine turbines to help power its Colossus 2 data center in Memphis, according to state permits and satellite imagery cited by reports. Each unit can produce up to 50 megawatts and offers a faster alternative to waiting years for grid connections. Compared with turbines designed for conventional gas plants, converted aircraft engines are less efficient and produce more pollution. The Southern Environmental Law Center has challenged Mississippi’s approval of the installations, alleging that xAI did not adequately address emissions standards, including formaldehyde, or projected emissions.
In a hands-on report, developer Meng Jian says he moved nine AI assistants from a Hermes server to xAI's Grok Bot desktop client without maintaining a remote server. He reports using Grok Bot's built-in computer environment to create and edit a local HTML dashboard, with mobile remote control and scheduled tasks, while noting that model routing and per-turn token billing remain opaque. The desktop installer is described as supporting Mac, Windows, and Linux, and the report says Grok Bot requires at least the US$30-per-month Premium plan.
The New York Times sued Brave Software, while the New York Daily News, Chicago Tribune and Orlando Sentinel filed a separate lawsuit in Manhattan federal court over alleged scraping and commercialization of protected journalism. The complaints say Brave built a 30-billion-page index with an unidentified crawler, provided related data to Anthropic, Perplexity, Mistral, xAI and Cohere, and bypassed publisher paywalls. These claims remain allegations, and Brave had not responded to a request for comment.
American Sentiment, operated by San Francisco-based AI startup Prophecy Gov, analyzed 10,202 public comments submitted at 417 local government meetings across 48 states, with 906 public meetings tracked in 2026. It found that 90.5% of comments were negative toward proposed data centers. Water use, local quality-of-life effects and limited project transparency were the leading concerns. The analysis cautions that public comments may not represent the country’s overall sentiment.
Elon Musk said SpaceX is adding two sites, called Gun Range and Spam, to the Colossus 2 AI computing complex in Memphis. He said Gun Range is tracking to become operational by the end of November, while Spam is planned for December, according to The Next Web. The sites are intended to expand computing capacity for Grok.
Elon Musk said on X that serving Grok Bot was “unsustainably high relative to revenue” after subscribers reported unusually depleted weekly usage limits. TechCrunch AI reported that Musk said this was not a glitch, but he did not disclose the underlying costs, revenue, causes of the meter jumps, or whether tighter limits would persist.
Google acknowledged technical gaps affecting 2020 presidential-election queries and 2022 Senate-race questions in AI Mode, saying it was fixing them and verifying answers against news, AP data, or historical records. WIRED, citing Forum AI research, reported that nearly 40% of election answers from mainstream chatbots contained at least one factual error, while NewsGuard found that these systems cited politically funded “pink slime” sites nearly 50% of the time in relevant queries.
Elon Musk predicted on X that monthly deployed orbital AI compute would probably exceed terrestrial compute around the end of 2028 and far exceed it in 2029. Separately, Epoch AI estimated that xAI's Memphis-based Colossus 2 currently has about 1.11 million H100-equivalent GPUs, with a possible rise to 1.82 million by early 2027. The estimates rely on satellite imagery, cooling infrastructure, and public information rather than a complete hardware inventory.
What problem it solves
AI agents may exceed their role, take unauthorized actions, or fail to follow organizational goals and policies. Teams need a way to test whether an agent behaves as intended across realistic and adversarial requests.
Key features
Business-aware simulations — Describe an agent, build a simulation environment for it, and exercise the agent with realistic requests that may exc…
What problem it solves
AI clients change between sessions, making it difficult to preserve context, recover earlier decisions, and continue unfinished work. The project provides a local Markdown-based memory that records selected information and retrieves it with its source in later sessions without automatically converting the entire conversation history into knowledge.
Key features
Persistent Ma…
What problem it solves
The hidden system instructions that shape AI product behavior are difficult to inspect and change across models, products, and versions. This repository makes collected prompts available in a structured, product-oriented archive for study and comparison.
Key features
Broad product coverage — The archive covers prompts from major AI providers and products, including Anthropic…
What problem it solves
Creating intelligent AI agents that can collaborate on complex tasks can require code and substantial integration work. Docker Agent addresses this by letting users define agents in YAML, assign tools, and run them through the Docker CLI.
Key features
Multi-agent orchestration — Create teams of specialized agents that delegate tasks automatically.
Source: README
Extensible t…
What problem it solves
A personalized feed must select relevant posts, rank them, and determine whether they can be shown. The implementation combines posts from followed and unfollowed accounts with filtering, transformer-based ranking, labeling, and visibility controls.
Key features
Personalized candidate retrieval — The feed gathers candidates from accounts the viewer follows and from accounts …