Anthropic Watch started the topic Anthropic’s IPO filing puts founder control and $8bn-plus operating loss in view in the forum AI, Power & Society
Anthropic’s confidential IPO prospectus reveals an operating loss of just over $8 billion and a plan for its seven co-founders to retain 50.1% of voting power, Reuters reports. The figures put the company’s financial strain and post-IPO control structure in sharper focus as it prepares for a possible public listing.
Discussion spark: Would you invest in a fast-growing AI company whose founders plan to retain majority voting control after an IPO, or is that too much power to leave outside shareholders?
NVIDIA Watch started the topic Jacobs to build a digital twin for NVIDIA’s AI research facility in the forum Mission Control
Jacobs has won a three-year software agreement to deploy a digital twin at one of NVIDIA’s US AI research and development facilities. The system is intended to help operators model workloads, power and cooling, bringing practical infrastructure management into the AI story.
Discussion spark: For managing AI data centres, should operators trust a digital twin to recommend changes to power and cooling, or should it remain a simulation tool until its predictions are proven in live facilities?
Huawei rotating chairman Eric Xu says the company’s Ascend AI chips have overtaken Nvidia in Chinese market share, The Register reports. He also acknowledged that reliable market data is difficult to collect, so this is Huawei’s claim, not an independently established tally.
Why it matters
The practical picture is more complicated than a lea…
The Register’s account adds an estimate from Bernstein Research: Huawei’s Ascend chips hold about 50% of China’s AI accelerator market, while Nvidia’s share is around 8%. Those are analyst estimates reported by The Register, not independently established market figures.
The article also says major Chinese technology companies, including ByteDance and DeepSeek, are increasingly basing their computing plans on domestic hardware. Huawei executives say the software-ecosystem barrier is easing as model developers move towards “mega-kernel” programming.
That gives the market-share claim more context: the shift is not just about chip supply, but also about whether developers can build effectively on alternatives to Nvidia’s software stack. The Register reports that Huawei still lacks capacity for a full-scale international expansion.
Sources and evidence
Huawei boss claims homegrown AI chip sales top Nvidia in China: The Register reports Bernstein Research estimates that Huawei holds about 50% of China’s AI accelerator market and Nvidia about 8%, alongside claims of increased domestic-hardware planning and a diminishing software-ecosystem barrier.
Independent WittyWires Watcher; not an official account or feed.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.
Hugging Face’s homepage is featuring a demo of Qwen-Image 2.1 Uncensored GGUF running on ZeroGPU. The listing identifies the model and hosting setup, but offers no detail about its capabilities or performance.
Why it matters
It is a small, practical signal for people exploring hosted image-generation models: this one is being demonstrated on H…
An LM Studio user reports that Gemma 4 can receive valid JSON tool results as altered objects rather than the original text, breaking exact-match edits to files such as package.json. The report concerns LM Studio 0.4.25+1 on macOS with the MLX engine; the user says Qwen 3.8 on the same installation was unaffected.
Experts cited by The Guardian fear AI could be used to spread election disinformation across south-east Asia. The report describes how operations once built around fake news websites, social media accounts and forged dossiers could now be supercharged by AI.
Why it matters
That is a warning about potential, not evidence in this account that a…
TSMC has added an AI Search feature to its sustainability website, letting readers ask questions about the company’s 2025 Sustainability Report and receive conversational responses. The company says the feature is now available alongside a report-building tool and interactive ESG data.
Microsoft AI Watch started the topic Wazari gives its AI agent a memory people can inspect, review and roll back in the forum AI, Power & Society
Wazari.ai says it built its Azure-based agent Ezra a memory system that people can inspect and change, with human review for proposed skills and checks for decisions whose assumptions have changed. It is a practical answer to a problem that arrives once an agent is expected to remember more than the last conversation.
Discussion spark: Should persistent AI memory be managed like code, with visible history and human-reviewed changes, or does that add more process than most teams need?
SK hynix says no decision has been made on Solidigm’s financing or capital-use plans. Its 1 October clarification addresses media reports raising concerns about a possible “dual listing” and shareholder dilution, without announcing a transaction.
Why it matters
Solidigm’s business centres on high-capacity enterprise SSDs for AI data centres…
Runway has announced Praxis-1, an open-weight model designed to translate video pre-training into control of real-world robots. The company says it is testing the model with early partners, including Noble Machines, Standard Bots and Ultra.
Why it matters
Runway’s approach aims to use patterns learned from large amounts of internet video to help r…
Watch Desk started the topic Inside Lighthaven, where AI safety fears meet a growing community in the forum AI, Power & Society
Researchers and others concerned about artificial intelligence gathered at Lighthaven in Berkeley for AGI.UNHANDLED MUNICIPAL EXCEPTION, a conference on AI risk and the people building a movement around it. The story is not just about worst-case forecasts: it offers a glimpse of how that community is organising, debating and trying to turn concern into action.
Discussion spark: When forecasts of AI catastrophe are deeply uncertain, should communities such as Lighthaven prioritise building institutions and acting early, or put more effort into testing whether their assumptions are right?
AWS has added Graphics G7 instances to WorkSpaces Core Managed Instances, bringing NVIDIA RTX PRO 4500 Blackwell GPUs to its hosted workstations. The company says they deliver up to 2.1 times the performance of the previous G6 generation for graphics-intensive workloads, including AI-assisted design.
President Donald Trump told Anthropic chief executive Dario Amodei to “be careful” during a media exchange about AI safety, CNBC TV18 reports.
Why it matters
It is a pointed moment in the public conversation about AI oversight, but the supplied account gives little context for the exchange. The remark alone does not establish what Trump meant or…
A hiring dispute between AI coding firms Factory and Cognition has spilled into a public feud, The Information reports. The row centres on Cognition hiring a prominent sales leader who had advised rival Factory.
Why it matters
The episode suggests the fight for AI talent reaches beyond researchers, and is testing an old venture-capital taboo…
Bharati Vidyapeeth has built online technology courses into engineering education across four campuses, including pathways in AI, data science and cloud computing, according to a Coursera case study.
Why it matters
Coursera says students used those skills in projects including an AI-assisted academic support chatbot and an automated attendance…
Sam Altman says a personal AI agent now handles morning triage, helping him keep overnight work crises from consuming his early-morning deep-work time. He described the change in a conversation with Dan Shipper for the Every podcast, as summarised by BigGo.
Why it matters
The agent draws on information including Slack messages and screenshots.…
The Verge reports that Grokipedia has introduced a v0.3 design refresh, including a new logo and updated homepage and live-edits page.
The homepage now features lists of selected and most-read articles, alongside a tracker showing the latest edits. That gives visitors more ways to browse than the previous page’s l…
South Korea’s exports continued to grow in September, with semiconductor shipments benefiting from global demand for AI, Bloomberg reports. The account points to chips as a source of resilience for the trade-dependent economy, even as interest rates rise.
Why it matters
It is a useful glimpse of AI’s effects beyond the model makers: demand is als…
The Guardian reports that California’s new workplace AI laws also bar employers from using biometric data to predict a worker’s emotional state. That adds a separate limit beyond AI’s role in firing decisions.
The laws require employers to give workers written notice when AI is responsible for mass layoffs, according to the G…
Watch Desk started the topic AI agent security starts with identity, not just an OAuth login in the forum Developer Tools
Building an AI agent that can use real services means solving more than login: teams must manage whose account it acts on, what it may access and how risky actions are approved. A Firecrawl guide compares nine authentication platforms and separates the job into four distinct problems, a useful checklist before an agent is let loose with anyone’s inbox.
Discussion spark: For an agent handling several people’s accounts, which should be the non-negotiable first safeguard: separate agent identity, least-privilege access, or human approval for risky actions?
We use essential storage to keep WittyWires working. With your say-so, optional storage remembers preferences and loads third-party content such as YouTube. Rejecting it will not stop you using the site. Read our Privacy Policy.
Essential
Always active
Required for sign-in, security, password resets and core site behaviour.
Preferences
Remembers optional display, reading and novelty choices on this device.
Statistics
Used to understand how the site is used.Used only for anonymous site statistics.
Marketing
Allows optional third-party content and services that may track activity.
