<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
  <title>ZilverNet Signals</title>
  <link>https://zilvernet.com/blog/</link>
  <atom:link href="https://zilvernet.com/blog/feed.xml" rel="self" type="application/rss+xml"/>
  <description>AI, open models, robotics and AI economics — with a source for every claim.</description>
  <language>en-us</language>
  <item>
    <title>Mistral releases the weights for its trillion-parameter model, pitched at cyber defense.</title>
    <link>https://zilvernet.com/blog/mistral-large-4/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/mistral-large-4/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>Mistral Large 4 launched as an API preview on October 6. Three weeks later anyone can download it — and Mistral is making a security argument for why that matters.</description>
    <category>Open-source AI</category>
  </item>
  <item>
    <title>Nvidia's next data-center platform, Vera Rubin, is now generally available on Oracle's cloud.</title>
    <link>https://zilvernet.com/blog/nvidia-vera-rubin-oci/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/nvidia-vera-rubin-oci/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>Rubin GPUs and Nvidia's new Vera CPU can now be rented on Oracle Cloud. The "multi-gigawatt" expansion is a plan; the racks are real.</description>
    <category>Chips &amp; compute</category>
  </item>
  <item>
    <title>Anthropic files to go public: $16.3 billion in half-year revenue, and founders who keep control.</title>
    <link>https://zilvernet.com/blog/anthropic-ipo-filing/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/anthropic-ipo-filing/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>Anthropic's IPO filing, made public on October 26, shows revenue up more than thirteenfold in a year, an enormous accounting loss that is mostly not cash, and a share structure that leaves its seven co-founders with majority voting power.</description>
    <category>AI economics</category>
  </item>
  <item>
    <title>Humanoid robots reached the production line this month. Nobody has made them pay yet.</title>
    <link>https://zilvernet.com/blog/humanoid-robots-production-no-roi-yet/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/humanoid-robots-production-no-roi-yet/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>Tesla started building Optimus in Fremont, and Agility signed up two more industrial customers for its Digit robot. The logistics company that has run the most humanoid pilots says it hasn't earned its money back — and expects that to take years.</description>
    <category>Robotics</category>
    <category>AI economics</category>
  </item>
  <item>
    <title>Nvidia's first Windows chip is built to run AI on your laptop, not in the cloud.</title>
    <link>https://zilvernet.com/blog/nvidia-rtx-spark/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/nvidia-rtx-spark/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>On sale since October 16 from six PC makers, RTX Spark pairs an Arm CPU and a Blackwell GPU with up to 128 GB of shared memory — enough, Nvidia says, to run a 120-billion-parameter model on your lap.</description>
    <category>Chips &amp; compute</category>
  </item>
  <item>
    <title>Reflection's Beam is out under Apache 2.0 — a US open model aimed squarely at the Chinese leaders.</title>
    <link>https://zilvernet.com/blog/reflection-beam/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/reflection-beam/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>A genuinely permissive license on a 501-billion-parameter model is the real news. The benchmark claims still need someone else to check them.</description>
    <category>Open-source AI</category>
  </item>
  <item>
    <title>Five silicon qubits, one control wire, 99.9% accuracy at once.</title>
    <link>https://zilvernet.com/blog/riken-silicon-qubits/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/riken-silicon-qubits/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>RIKEN researchers found the hidden error that appears when qubits share wiring, and calibrated it away. The work is published in Nature Physics.</description>
    <category>Chips &amp; compute</category>
    <category>Science</category>
  </item>
  <item>
    <title>Meta's new open model runs on two Mac Studios. Read the license before you build on it.</title>
    <link>https://zilvernet.com/blog/meta-muse-spark-open/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/meta-muse-spark-open/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>Meta calls Muse Spark Open the most capable US open-weight model to date. Whether you can use it freely depends on a clause most people will skip.</description>
    <category>Open-source AI</category>
  </item>
  <item>
    <title>Nvidia's new open models are built to run agents on your own PC, not in the cloud.</title>
    <link>https://zilvernet.com/blog/nvidia-nemotron-3-5/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/nvidia-nemotron-3-5/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>Nemotron 3.5 spans the data center to the desktop — including a 74-billion-parameter model squeezed into 24 GB of memory with native 2-bit quantization.</description>
    <category>Open-source AI</category>
  </item>
  <item>
    <title>AMD's answer to Nvidia's AI racks is an open blueprint built for agents.</title>
    <link>https://zilvernet.com/blog/amd-helion/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/amd-helion/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>Unveiled at the Open Compute Project summit, Helion is a rack design rather than a chip — and AMD plans to give the specification away.</description>
    <category>Chips &amp; compute</category>
  </item>
  <item>
    <title>OpenAI says an AI proved one of computer science's biggest open problems. No independent expert has reviewed it yet.</title>
    <link>https://zilvernet.com/blog/unique-games-ai-proof/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/unique-games-ai-proof/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>The unique games conjecture has stood since 2002. Weeks before OpenAI's announcement, three MIT researchers rushed out their own hand-written milestone — and the episode shows how AI is changing mathematics.</description>
    <category>Science</category>
    <category>AI</category>
  </item>
  <item>
    <title>MIT built a robot the size of a stick of gum that swims using living muscle.</title>
    <link>https://zilvernet.com/blog/mit-muscle-robot/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/mit-muscle-robot/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>A half-millimeter gel film with light-controlled muscle fins — the first thin, flat robot to get around on living tissue.</description>
    <category>Inventions</category>
    <category>Robotics</category>
  </item>
  <item>
    <title>Google's Artemis lets AI coding assistants drive a real Android phone — and it's open source.</title>
    <link>https://zilvernet.com/blog/google-artemis/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/google-artemis/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>Published by Google in August, Artemis turns plain-language instructions into Android automation and plugs straight into Claude Code, Codex and Antigravity. It has more than 15,000 GitHub stars and one benchmark claim worth checking.</description>
    <category>Open-source AI</category>
    <category>AI</category>
  </item>
  <item>
    <title>This is the year running AI costs more than training it.</title>
    <link>https://zilvernet.com/blog/inference-overtakes-training/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/inference-overtakes-training/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>Gartner expects spending on AI-optimized cloud infrastructure to nearly double in 2026 — with most of it now going to using models rather than building them.</description>
    <category>AI economics</category>
  </item>
  <item>
    <title>Grok Bot gives AI agents their own computer and your logins. Two months in, here's what has changed.</title>
    <link>https://zilvernet.com/blog/grok-bot/</link>
    <guid isPermaLink="true">https://zilvernet.com/blog/grok-bot/</guid>
    <pubDate>Thu, 29 Oct 2026 12:00:00 GMT</pubDate>
    <description>Launched on August 11 by the company formerly known as xAI, Grok Bot is a team of agents that sign into your tools and keep working around the clock. Since then: a $60 billion acquisition, wider access, and a reported switch to Anthropic's Claude.</description>
    <category>AI</category>
  </item>
</channel>
</rss>
