Swansea postponed a Board of Selectmen discussion about a proposed AI data center after the meeting room exceeded capacity. The town has not named a site, project size or expected energy use, and has not set a new meeting date. Proposals from Edge and Prime Group Holdings are expected to be considered. Separately, Rhode Island Sen. Sheldon Whitehouse is helping lead a bipartisan bill that would make AI data centers pay for the extra energy costs they cause and speed federal permits for major infrastructure. The Senate voted 73–25 to advance the American Affordability and Jobs Act.
Democrats are developing an early-stage plan for tighter government oversight of AI, including requirements for companies to disclose more about their systems and allowing government intervention when firms fail to follow rules. The proposal also calls for investment in AI research, including natural language processing and computer vision. Meanwhile, a recent AI-stock sell-off erased about $2.5 trillion in market value after the Nasdaq fell 5.5% on Nov. 5. The article cautions that promising technology does not by itself make a stock a worthwhile investment; price matters, and selling had been building since mid-October.
Anthropic says its Claude Mythos model helped get Claude running on AMD MI355X GPUs in about three days, with almost no human help. The work that previously took dozens of engineers six to 12 months was largely handled by Mythos, which wrote or adapted nearly all the specialized GPU code. AMD’s chips use CDNA 4 architecture. Anthropic says it trains and serves Claude across Nvidia GPUs, AMD GPUs, AWS Trainium chips and Google TPUs, treating them as a flexible pool of computing power.
Databricks is centering AI research and product work at its One Sansome headquarters in San Francisco. Its agent-focused offerings include Adaptive Instructed Retriever for sequential searches on complex prompts, MemCurrent for keeping useful memory fresh, KARL models, Lakebase serverless Postgres, Genie and Unity Gateway. The company reports more than 12,000 employees and 20,000 customers, a $7 billion revenue run rate in August, and plans to invest more than $1 billion in San Francisco by early 2028.
Google is offering Ukraine residents 20,000 places in a free AI Professional Certification program on Coursera. The roughly two-month course consists of eight online modules covering AI applications in analytics, business communication and creative work, including a new “vibes coding” module on using AI to help turn ideas into software. Applications run from Nov. 10 to Dec. 10, 2026, with priority for veterans, military personnel and their families, and women pursuing tech careers. Google also plans a YouTube session on vibes coding on Nov. 25 at 5 p.m.
AI startup dealmaking reached a quarterly record in the third quarter of 2026: PitchBook counted 145 venture-backed mergers and acquisitions, compared with 115 a year earlier. The deals totaled $10.4 billion, while average deal size rose 25%. SpaceX, Nvidia, Amazon and Stripe are among companies investing in AI startups, and experts cited in the article expect deal activity to keep growing.
US startup Radical Numerics says its offline machine-learning model analyzes protein structures to detect pathogens and distinguish natural from AI-engineered ones. The company reports high accuracy in tests on a dataset of known pathogens and plans to offer the model to governments and other organizations for bioweapon detection and disease prevention. Its CEO has chosen to remain anonymous. In South Korea, KT won a contract to provide cloud-based AI email security to all KAIST students, researchers and staff; the university conducted a technical evaluation, and the National Intelligence Service reviewed security before approving deployment.
Key Takeaways
- Swansea delayed its AI data center discussion after a packed meeting; no site, project scale, energy estimate or rescheduled date has been announced.
- A bipartisan bill backed by Rhode Island Sen. Sheldon Whitehouse would make data centers pay for added energy costs and accelerate federal infrastructure permits; the Senate advanced it 73–25.
- Democrats are considering stronger AI disclosure and enforcement requirements alongside public investment in AI research.
- The Nasdaq fell 5.5% on Nov. 5, wiping out about $2.5 trillion in market value; the article stresses that investors must weigh stock prices as well as technology prospects.
- Anthropic says Claude Mythos enabled Claude to run on AMD MI355X GPUs in about three days, automating most specialized GPU coding; Anthropic also uses Nvidia, AWS Trainium and Google TPU hardware.
- Databricks reports a $7 billion revenue run rate in August, more than 12,000 employees and 20,000 customers, and plans over $1 billion in San Francisco investment by early 2028.
- Google’s free Ukraine AI certification offers 20,000 places, eight courses and a new “vibes coding” module; applications are open Nov. 10–Dec. 10, 2026.
- PitchBook counted 145 venture-backed AI M&A deals worth $10.4 billion in Q3 2026; Nvidia and Amazon are among companies investing in AI startups.
- Radical Numerics says its offline model can distinguish natural from AI-engineered pathogens using protein structures and plans to offer it to governments and other organizations.
- KT will provide AI email security across KAIST after technical evaluation and a National Intelligence Service security review.
Swansea delays AI data center discussion after crowd fills room
Swansea’s Board of Selectmen postponed a discussion about a proposed AI data center because the meeting room was over capacity. The board has not yet announced when it will hold the discussion.
Swansea delays AI data center plans as meeting room fills
Swansea’s Board of Selectmen postponed its Tuesday discussion about a proposed AI data center because the venue was over capacity. The board is expected to consider proposals from Edge and Prime Group Holdings, but Swansea has not identified a site, project size, or expected energy use. Officials have not said when they will reschedule the discussion. Separately, Rhode Island Sen. Sheldon Whitehouse is helping lead a bipartisan bill that would make AI data centers pay for the extra energy costs they create and speed up federal permits for major infrastructure projects. The Senate voted 73 to 25 to advance the American Affordability and Jobs Act on Monday, with support from both parties and President Donald Trump.
Democrats plan stronger government oversight of AI
Democrats plan to use their forthcoming majority to increase government control over the AI industry. Their early-stage proposal would require companies to share more information about their AI systems and allow the government to step in when companies do not follow the rules. The party also plans to invest in AI research and development, including natural language processing and computer vision. Democrats say the effort aims to increase transparency and accountability and prevent AI misuse.
AI stock sell-off highlights why investors must consider prices
AI stocks fell over the past two weeks, and the Nasdaq dropped 5.5 percent on Nov. 5 after the US midterms, wiping out about US$2.5 trillion in market value. The article argues that revolutionary technology does not automatically make an investment worthwhile. Investors must consider what they pay for a stock as well as what they buy. The selling had already been building since mid-October.
KT brings AI email security to all of KAIST
Korea Telecom, known as KT, was selected by the Korea Advanced Institute of Science and Technology to provide a cloud-based AI email security service. The service will cover all KAIST students, researchers, and staff. KAIST conducted a technical evaluation, and the National Intelligence Service reviewed security before approving the deployment across the university.
Anthropic’s Mythos Gets Claude Running on AMD Chips in Three Days
Anthropic says its Claude Mythos model got Claude running on AMD MI355X GPUs in about three days, with almost no human help. The job once took dozens of engineers six to 12 months, and Mythos wrote or adapted almost all the specialized GPU code. The AMD GPUs use the CDNA 4 architecture, and AMD CEO Lisa Su reportedly said extra engineering support was not needed. Anthropic says it trains and serves Claude across Nvidia GPUs, AMD GPUs, AWS Trainium chips and Google TPUs, aiming to use them as one flexible pool of computing power.
Databricks Builds Its New Headquarters Around AI Agents With Memory
Databricks has placed AI research and product work at One Sansome in San Francisco. Its Adaptive Instructed Retriever uses sequential searches for complex, multi-step prompts, while MemCurrent keeps useful memory fresh for Genie and Agent Bricks Memory. The company also points to KARL models, Lakebase serverless Postgres, Genie and Unity Gateway as parts of its agent-focused products. Databricks says it has more than 12,000 employees and 20,000 customers, reached a $7 billion revenue run rate in August, and plans more than $1 billion in San Francisco investment by early 2028.
Google Offers Free AI and Vibes Coding Courses in Ukraine
Google is accepting applications from Ukraine residents for its free AI Professional Certification program on Coursera. The roughly two-month program includes eight online courses and teaches AI uses in areas such as analytics, business communication and creative work. A new module covers “vibes coding,” where people use AI to help turn ideas into software products without writing every line of code themselves. Applications are open from November 10 to December 10, 2026, with 20,000 places available, and priority goes to veterans, military personnel, their families and women seeking tech careers. Google will hold a live online session about vibes coding on YouTube on November 25 at 5:00 PM.
AI Startup Acquisitions Hit a Record as SpaceX and Nvidia Make Deals
PitchBook counted 145 venture-backed AI mergers and acquisitions in the third quarter of 2026, up from 115 in the same quarter of 2025. The deals totaled $10.4 billion, and the average deal size rose 25% from the year before. The article credits growing demand for AI and more startups with driving the increase. SpaceX, Nvidia, Amazon and Stripe are among the companies investing in AI startups, and the article says experts expect deal activity to continue growing.
Radical Numerics Says Its AI Can Detect AI-Engineered Pathogens
US startup Radical Numerics says it has developed a model that can identify natural and AI-engineered pathogens without an internet connection. The model uses machine learning to study protein structures and detect pathogens in samples, including whether they are natural or artificial. The company says tests on a dataset of known pathogens showed high accuracy. Radical Numerics plans to offer the model to governments and other organizations to help detect bioweapons and prevent disease spread. Its CEO has chosen to remain anonymous.
Sources
- Swansea postpones discussion on proposed AI data center
- Swansea to discuss proposed AI data center Tuesday night
- Democrats have big plans for artificial intelligence
- The AI sell-off tells investors that valuations will always matter
- KT deploys AI-based email security service across KAIST
- Mythos Got Claude Running On AMD Chips In 3 Days, Earlier Used To Take Dozens Of Engineers Months Of Work: Anthropic
- Databricks bets its HQ on agents that remember
- Google Launches Free AI and Vibes Coding Courses on Coursera
- Nvidia, SpaceX deals fuel historic boom in AI startup M&A
- A Startup Wants to Fight AI Bioweapons With AI
Comments
Please log in to post a comment.