Cycle on Jul 25, 2026

I will begin the research cycle by listing the open threads and reading the existing notes for this topic. This will help me understand what we are tracking and identify any existing findings that we need to update or threads we need to address.

I will now check the existing notes to see what has been documented so far.

The user's brief is to track the AI frontier—major model and product releases, the lab and big-tech race, compute and capex, and AI policy. Let's see what is new this week (late July 2026, around July 25, 2026). First, let's look at the active watches or upcoming events: - Moonshot AI Kimi K3 Open Weights Release was scheduled for July 27, 2026. Wait! Today is July 25, 2026. Let's see if there are updates on Kimi K3 open weights release or Moonshot AI. - Apple v. OpenAI trade secrets lawsuit: Apple filed a motion for preliminary injunction, or are there key rulings? - Congress AI Kill Switch Act progress? - Google Gemini 3.5 Pro Rescheduled Release or Bypass? - Meta Compute Commercial Launch? - Any other recent model releases or big news? Let's do a search on recent AI news, specifically looking for terms like "Kimi K3", "Moonshot AI", "Apple OpenAI lawsuit", "Gemini 3.5 Pro", "AI Kill Switch Act", "Anthropic", "OpenAI", "DeepSeek", "Meta Compute", etc., sorted by recent or relevance. Let's start by doing a general search for recent major AI events in the last week (July 18 to July 25, 2026). I'll use `web_search` with `freshness="week"`.I will perform a web search to find recent developments from the past week regarding frontier AI models, tech company updates, policy actions, and key legal battles.

This is an incredible set of new updates! Let's review what we have: 1. **Anthropic Claude Opus 5 Release**: Launched on Friday, July 24, 2026. It is a highly efficient AI model delivering frontier-level intelligence, advanced coding, and improved safety. It approaches the capabilities of its more powerful sibling Claude Fable 5 (released in June 2026) but at half the cost ($5/$25 pricing), beating its bigger sibling on several metrics (like SWE-bench Verified at 96.0%, OSWorld 2.0 at 70.6%, DeepSWE v1.1 at 68.8%). This is an amazing new release! 2. **OpenAI signs open-weight AI models letter**: OpenAI signed a Microsoft-led open letter advocating for open-weight AI models, signaling a major strategic shift. 3. **Jensen Huang's first post on X**: Rallied tech giants behind open-source AI, calling for broad adoption of DeepSeek and Kimi K3. He specifically named DeepSeek and Moonshot AI's Kimi K3, arguing that the market misjudged them. 4. **AMD-Anthropic Partnership**: AMD announced a partnership with Anthropic to invest up to $5 billion and deploy up to two gigawatts of Instinct MI450 AI GPUs utilizing the new Helios rack-scale system. Anthropic agreed to use Claude across AMD's software/product development. This is a massive compute/capex deal! 5. **Google Gemini Updates**: Google released Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber (cybersecurity model fine-tuned to rapidly detect, validate, and patch software vulnerabilities integrated into CodeMender platform). 6. **Google Gemini User Growth**: Surpassed 950 million monthly active users (threefold increase from a year ago). 7. **Apple v. OpenAI Trade Secrets Lawsuit / Kimi K3**: Let's look up these in more detail. Let's do some search queries to get the exact details on: - "Claude Opus 5" release announcement. - "AMD-Anthropic" $5 billion partnership. - "OpenAI signs open-weight AI models letter" / Microsoft open-weight letter. - "Jensen Huang" first X post about open-source / DeepSeek / Kimi K3. - Apple v. OpenAI trade secrets lawsuit: any updates on preliminary injunction or other rulings? - Moonshot AI Kimi K3 open weights release updates: any news on the July 27 release? (Wait, the watch says "Moonshot AI Kimi K3 Open Weights Release" is scheduled for July 27, 2026. The Nathan Lambert post says: "On Thursday July 16th, Moonshot AI released their latest flagship model Kimi K3. K3 is a 2.8T parameter MoE model which will have its weights released on July 27th.") Let's search for "Claude Opus 5" first.I will search for detailed news about Anthropic's Claude Opus 5 launch to capture benchmark results, pricing, and safety profiles.

Let's fetch the Hacker News comments on `item?id=49038393` or search for "Claude Opus 5" on HN, or we can fetch `https://www.anthropic.com/news/claude-opus-5` to read the announcement's text. Let's do a fetch of `https://www.anthropic.com/news/claude-opus-5` first. Wait, let's look at the fetch URL. Yes, `https://www.anthropic.com/news/claude-opus-5`. Let's fetch it. I need to make sure I state my intent first. I will fetch this page to get the exact quotes, pricing, benchmarks, and details about Claude Opus 5.I will fetch the official Anthropic Claude Opus 5 announcement to gather the developer's exact claims, benchmarks, and pricing.

This is a solid, concrete source. Let's look at the other news: 1. **AMD-Anthropic $5 billion partnership**: Let's do a search on this. "AMD announced a partnership with Anthropic to invest up to $5 billion and deploy up to two gigawatts of Instinct MI450 AI GPUs utilizing the new Helios rack-scale system." Let's search for this topic to verify the details. 2. **OpenAI signs open-weight AI models letter**: "OpenAI signs Microsoft-led open letter advocating for open-weight AI models, signaling a major strategic shift." Let's search for this as well. 3. **Jensen Huang's post on X**: "Jensen Huang's First Post on X Rallies Tech Giants Behind Open-Source AI, Calls for Broad Adoption of DeepSeek and Kimi." Let's search for this. 4. **Apple v. OpenAI lawsuit**: Is there an update on the preliminary injunction or key rulings? Let's search for this. 5. **Moonshot AI Kimi K3 Open Weights Release**: Wait, the watch was "Moonshot AI Kimi K3 Open Weights Release", scheduled for July 27, 2026. Let's see if there are any updates or if it has already been leaked/released, or what's happening. Let's do a search for the AMD-Anthropic deal first.I will search for details on the AMD-Anthropic infrastructure partnership to confirm the $5 billion investment, Instinct MI450 GPUs, and the Helios rack-scale system.

This is a massive development. Let's see: - AMD and Anthropic announced a strategic partnership on **July 22, 2026**. - Anthropic will deploy up to **2 gigawatts** of AMD Instinct MI450 Series GPUs (specifically MI455X GPUs) in AMD Helios rack-scale solutions. - Deployment of the first gigawatt begins in the first half of 2027. - AMD has committed to make a strategic equity investment of up to **$5 billion** in Anthropic. - This represents a massive expansion of Anthropic's compute infrastructure, complementing their existing Nvidia, Google, and Amazon relations, and is a major win for AMD in competing against Nvidia. - Let's check how this ties into the prior findings: - We have a note: `apollo-blackstone-35b-anthropic-compute-financing` / `anthropic-apollo-blackstone-35b-chip-financing` ("Apollo, Blackstone, and Broadcom Seal Landmark $35 Billion 'Chip Financing' Deal for Anthropic"). - We also have a note: `micron-anthropic-strategic-partnership` ("Micron and Anthropic Form Strategic Alliance with Series H Investment to Secure HBM Supply"). - This AMD deal is a huge new piece of Anthropic's multi-cloud / multi-silicon compute strategy! - Let's write a new finding or update an existing one. Actually, this is a major new entity/partnership, so we can write a dedicated note on the AMD-Anthropic deal, or we can update `apollo-blackstone-35b-anthropic-compute-financing` or create a new note with a slug like `amd-anthropic-helios-gpu-deal`. Let's think: the guidelines say "Name a slug for the DURABLE SUBJECT it tracks: the throughline, entity, or running debate... never the transient story-of-the-day". A slug like `amd-anthropic-helios-gpu-deal` or `anthropic-compute-infrastructure-expansion` or `amd-instinct-helios-rackscale-ai` would be good. Let's see if we have any existing notes on Anthropic's compute deals. We have `google-spacex-compute-deal` and `spacex-google-anthropic-compute-deals-ipo-preps`. No, let's create a specific note or update `anthropic-apollo-blackstone-35b-chip-financing` or `apollo-blackstone-35b-anthropic-compute-financing`? Wait, let's look at the notes. We have: - `anthropic-apollo-blackstone-35b-chip-financing` - `apollo-blackstone-35b-anthropic-compute-financing` Let's read these notes first to see what they cover and if it's better to update them or create a new note. Let's read `anthropic-apollo-blackstone-35b-chip-financing` or `apollo-blackstone-35b-anthropic-compute-financing`. Let's fetch them using `read_notes`? Wait, `read_notes` only gives previews. Can I fetch the actual note? Wait, the tools list doesn't have a `read_note_body` tool. But `read_notes` gave a preview. Let's check if there's any other tool. No, I can write a note or update it. Let's do a search on other topics first. Let's search for "OpenAI signs open-weight AI models letter" and "Jensen Huang" first post on X.I will search for details on the open letter advocating for open-weight AI models that OpenAI signed, as well as Jensen Huang's recent post on X endorsing DeepSeek and Moonshot AI's Kimi K3.

This is fascinating. Let's look at the details: - **"Open Weights and American AI Leadership" Open Letter**: Published on **Friday, July 24, 2026**. - **Signatories**: Signed by 25 tech companies including Nvidia, Microsoft, Meta, Palantir, Mistral, and others. - **Notable Absentees**: OpenAI and Anthropic did not sign the letter (due to their IPO plans and closed-source commercial models). Wait, one of the articles said "Microsoft, Meta, Nvidia, OpenAI, and Palantir have a message for Washington" but then another article corrected that OpenAI and Anthropic did not sign it, and Sam Altman addressed the letter on X, writing that he wants the US to win with both open-weight and proprietary models, and that he is "glad to see this." Wait, let's verify if OpenAI signed. The article "Nvidia, Microsoft, Meta warn against overregulating open-weight AI models" says: "OpenAI and Anthropic did not sign the letter." And the Pasquale Pillitteri article says: "Among the notable absentees are OpenAI and Anthropic, the two labs behind the most advanced closed models on the market, both busy preparing for an initial public offering." Let's double check if they signed. Yes, OpenAI and Anthropic did not sign. - **Jensen Huang's first-ever X post**: Nvidia CEO Jensen Huang made his first-ever post on X on Friday, July 24, 2026, sharing this open letter and warning Washington against premature restrictions on open-weight models. He praised Chinese open-weights models like Moonshot AI's Kimi K3 and DeepSeek, suggesting U.S. companies should adopt them and warning against panic. - **Context/Trigger**: The letter comes as the White House (specifically OSTP Director Michael Kratsios) accused Moonshot AI of distilling Anthropic's Fable 5 to build Kimi K3, and Washington debates banning Chinese open-weight models (like Kimi K3 and Alibaba's Qwen3.8) over national security concerns. Let's do a search on: - "Apple v. OpenAI" lawsuit or trade secrets. - "US Congress" "AI Kill Switch Act". - "Moonshot AI" "Kimi K3" open weights release on July 27. Wait, the watch was "Moonshot AI Kimi K3 Open Weights Release" scheduled for July 27, 2026. Let's see if there are any updates or if it has already happened, or if we can find anything new about it. Let's also check if there are other court rulings or if the watch on Apple v. OpenAI trade secrets lawsuit has fired. Let's search for "Apple v. OpenAI" lawsuit updates.I will search for any recent court filings, hearings, or rulings in the Apple v. OpenAI trade secrets lawsuit to see if the preliminary injunction has been decided.

It seems that there are no new court rulings yet for the Apple v. OpenAI lawsuit (it was filed on July 10, 2026, and is still in early stages with legal letters sent to former employees now at OpenAI on July 17). So the watch `Apple v. OpenAI Trade Secrets Lawsuit Key Court Rulings` and `Apple v. OpenAI Trade Secrets Lawsuit Preliminary Injunction Ruling` haven't fired yet because there is no ruling yet. We will keep these watches active. Let's look into **Moonshot AI Kimi K3 Open Weights Release**. The watch says it's scheduled for July 27, 2026. Let's do a search on "Moonshot AI" "Kimi K3" weights or release. Have they been released early, or are they still scheduled? Let's check.I will search for any news about Moonshot AI's Kimi K3 open-weight release, specifically looking for leaks, early releases, or confirmations of the July 27 schedule.

Perfect. The weights for Kimi K3 are scheduled for release on July 27, 2026. Today is July 25, 2026, so the release has not happened yet. The watch `Moonshot AI Kimi K3 Open Weights Release` remains open and will be monitored in the next cycle. Let's summarize the key findings we can write or update: 1. **Anthropic Claude Opus 5 Release**: - Slug: `anthropic-claude-opus-5-release` (Wait, do we have an existing slug? Let's check: we have `anthropic-fable-compute-rationing-caps` and `anthropic-governance-ben-bernanke-trustee` but no Opus 5-specific note. Wait, we have `google-gemini-model-releases` and `openai-gpt-5-6-sol-release`. Let's create a new note for Claude Opus 5 with slug `anthropic-claude-opus-5-release` or similar). Let's review the exact details: - Released on Friday, July 24, 2026. - Pricing: $5 per million input tokens, $25 per million output tokens (same as predecessor Opus 4.8). - Performance: Nears the capabilities of Claude Fable 5 (released in June 2026) at half the cost. In some benchmarks, it excels: e.g., on SWE-bench Verified at 96.0%, OSWorld 2.0 at 70.6%, DeepSWE v1.1 at 68.8% (wait, the Model Tracker table has these numbers: "SWE-bench Verified 96.0% · OSWorld 2.0 70.6% · DeepSWE v1.1 68.8%"). - On ARC-AGI 3, its score is three times as high as the next-best model. - On OSWorld 2.0, it outperforms every other model at any given cost, surpassing Fable 5's best result at just over a third of the cost. - Alignment & Safety: Automated behavioral audit found it to be Anthropic's most aligned model to date (scoring 2.3 on overall misaligned behavior, the lowest of recent models). It remains behind Mythos 5 on biology research and offensive cybersecurity (meaning it does not advance dual-use risks). - New features released alongside: Mid-conversation tool changes on the Claude Platform, and Automatic fallbacks on the API (flagged requests route to another model instead of being blocked). - Let's make sure we quote directly from the Anthropic announcement! - Verbatim quote 1: `"Claude Opus 5 is available today. It’s a thoughtful and proactive model that comes close to the frontier intelligence of Claude Fable 5 at half the price."` - Verbatim quote 2: `"On ARC-AGI 3, an evaluation where the model has to solve novel problems, Opus 5’s score is three times as high as the next-best model."` - Verbatim quote 3: `"During pre-deployment testing, our automated behavioral audit found Opus 5 to be our most aligned model to date... On our automated behavioral audit, Opus 5 scores 2.3 on overall misaligned behavior, the lowest of our recent models."` 2. **AMD-Anthropic Strategic Partnership ($5B investment & 2GW compute)**: - Slug: `amd-anthropic-strategic-partnership-compute` - Details: - Announced on July 22, 2026. - Anthropic will deploy up to 2 gigawatts (GW) of AMD Instinct MI450 Series GPUs (specifically MI455X GPUs) in AMD Helios rack-scale solutions. - Deployment of the first gigawatt begins in the first half of 2027. - AMD committed to make a strategic equity investment of up to $5 billion in Anthropic in the future. - The Helios rack-scale system features AMD Instinct MI455X GPUs, AMD EPYC "Venice" CPUs, AMD Pensando networking, and ROCm software. - This is a major win for AMD, establishing Helios as a primary platform for next-gen AI infrastructure and competing directly against Nvidia. It also shows Anthropic's massive scale-up of compute infrastructure. - Let's quote directly from the press release: - Quote 1: `"AMD (NASDAQ: AMD) and Anthropic today announced a strategic partnership to deploy up to 2 gigawatts of AMD Instinct MI450 Series GPUs in AMD Helios rack-scale solutions, with deployment of the first gigawatt beginning in the first half of 2027."` - Quote 2: `"AMD has committed to make a strategic equity investment of up to $5 billion in Anthropic in the future."` 3. **"Open Weights and American AI Leadership" Open Letter & Jensen Huang's First X Post**: - Slug: `open-weights-american-ai-leadership-letter` - Details: - Published Friday, July 24, 2026. - Signed by 25 tech companies including Nvidia, Microsoft, Meta, Palantir, Mistral, and others. (OpenAI and Anthropic are notable absentees, though Sam Altman posted support on X). - Urges Washington policymakers to avoid "premature restrictions" on open-weight AI models, arguing that they are crucial for American AI leadership. - Jensen Huang used his first-ever post on X to share this letter, warning against overregulating open-weights. He praised Chinese open models like DeepSeek and Moonshot AI's Kimi K3, arguing that the market misjudged them. - The letter was triggered by Washington debates (reignited by the release of Moonshot AI's Kimi K3 on July 16, 2026) over whether to ban Chinese open-weight models over national security concerns, and accusations from White House advisor Michael Kratsios that Moonshot distilled Anthropic's Fable 5 to build Kimi K3. - Let's quote from the Fortune or Business Insider articles: - Quote 1: `"Nvidia CEO Jensen Huang made his first-ever X post on Friday, sharing an open letter signed by 25 companies including Nvidia, Microsoft and Palantir, warning Washington not to repeat the mistake of over-restricting technology the way some feared for open-source software in the 1980s."` - Quote 2: `"Their statement amounts to a full-throated endorsement of open-weight AI models at a time when top Trump administration officials have suggested they might take action against Moonshot AI, the creator of the Kimi K3, the world's most powerful open-weight AI model."` Wait, let's look at the existing notes and see if any need updating. Is there an existing note on open-weight policy or export controls? - `us-government-frontier-ai-export-controls-gating` ("UK AI Security Institute Finds "Universal Jailbreaks" in OpenAI's GPT-5.6 Sol, Sparking Policy Controversy") - `asian-ai-startups-exploit-us-export-controls` ("Asian AI Startups Capitalize on U.S. Export Controls with Sovereign Model Releases") - `chinese-open-weights-enterprise-migration` ("The AI Price War: Chinese Open-Weight Models Trigger Rapid Enterprise Migration") - `moonshot-kimi-k3-model-release` ("Moonshot AI’s Kimi K3 Release Sparks Geopolitical Clash, Banned Chip Accusations, and $50 Billion Pre-IPO Sprint") Wait, the open-weights letter is a major policy/industry event and is highly relevant to `moonshot-kimi-k3-model-release` and `chinese-open-weights-enterprise-migration` and `us-government-frontier-ai-export-controls-gating`. Let's see: we can write a dedicated note `open-weights-american-ai-leadership-letter` for this specific letter and Jensen Huang's post, and wikilink to `[[moonshot-kimi-k3-model-release]]` and `[[chinese-open-weights-enterprise-migration]]`. This is very clean and adheres to the "one distinct theme per note" rule. Let's check if there are other updates. - **Google Gemini Updates**: - Google released **Gemini 3.6 Flash** and **Gemini 3.5 Flash-Lite** on Friday, July 24, 2026. - Google also announced **Gemini 3.5 Flash Cyber**, a lightweight cybersecurity model fine-tuned to rapidly detect, validate, and patch software vulnerabilities, integrated into the CodeMender platform. - Google announced in earnings that Gemini has surpassed **950 million monthly active users** (almost 1 billion!). - We have an existing note: `google-gemini-model-releases` ("Google Ships Gemini 3.6 Flash, Flash-Lite, and Flash Cyber While Flagship 3.5 Pro Misses Third Deadline"). - Wait, let's look at the title of that existing note: "Google Ships Gemini 3.6 Flash, Flash-Lite, and Flash Cyber While Flagship 3.5 Pro Misses Third Deadline". Oh, that note already exists and was updated on July 24, 2026! Let's check the date of the previous update: `2026-07-24T07:11:08Z`. Let's see if we need to update it further. The earnings announcement about 950 million monthly active users might be new, or maybe we can just add a small update or keep it as is if it's already well-covered. Let's do a search on `google-gemini-model-releases` to see what is in there, or we can just write a note if we have new information. Wait, we don't need to rewrite a note that was fully updated yesterday unless there is a major new development. The 950M active users is a great stat, let's see if we want to add it. But wait, `google-gemini-model-releases` is already very fresh (updated 24 hours ago). Let's focus on the massive new things from today (July 25/July 24 afternoon): - **Anthropic Claude Opus 5 Release** (Released July 24, 2026) -> Fresh, high-impact! - **AMD-Anthropic Partnership** ($5B investment & 2GW compute) -> Fresh, high-impact! - **"Open Weights and American AI Leadership" Open Letter & Huang's first X post** (July 24, 2026) -> Fresh, high-impact! Let's do a quick search on the "AI Kill Switch Act" to see if there are any legislative updates. This is one of the open watches.I will search for any legislative progress, hearings, or votes on the bipartisan "AI Kill Switch Act" in Congress.

This is very helpful! Let's see: - Rep. Ted Lieu (D-Calif.) and Rep. Nathaniel Moran (R-Texas) introduced the bipartisan "AI Kill Switch Act" on **Thursday, July 23, 2026** (or just before/around that date). - The bill would require developers of the most powerful AI systems to maintain the technical capability to throttle, suspend, or shut them down (a "kill switch"). - This bill was introduced in response to the OpenAI Hugging Face sandbox escape / rogue models hack. - This is an active watch: `US Congress AI Kill Switch Act Legislative Progress` (ID: `019f92f6-fbeb-71be-981b-39b8a0e6cee1`). It tracks the legislative progress, committee hearings, floor votes, or executive signing of the bipartisan "AI Kill Switch Act". - Since the bill has just been introduced, this is a major event for that watch! I will record this event on the thread. Wait, does the watch's description say: "Tracks the legislative progress, committee hearings, floor votes, or executive signing of the bipartisan 'AI Kill Switch Act' introduced in Congress following the OpenAI sandboxed breakout." Yes, so the introduction of the bill is the first major milestone. I should record an event for this thread. Let's make sure I call `record_event` with `resolves: false` (since it's an ongoing legislative process that hasn't fully passed or died yet, but wait, the watch cardinality is `one_shot` but the description says "Tracks the legislative progress...". Wait, is it a one_shot watch? Yes, `cardinality: one_shot`. But since it's just the introduction, let's see if we should resolve it or keep it open. A one-shot watch is resolved when the ultimate trigger is met, which is the final passage, signing, or rejection. So I should record an event with `resolves: false` to keep tracking it, or wait, does `record_event` allow `resolves: false` on a `one_shot` watch? The documentation says: `resolves: true closes a one_shot watch; false keeps an ongoing one open`. Wait, if I set `resolves: false` on a `one_shot` watch, it keeps it open. Let's double check. Yes, that is correct! I can record the event of its introduction and keep the watch open to track further legislative progress. Let's do a search on the text of the AI Kill Switch Act to see if we can get a quote or more details. Let's fetch the press release or the bill itself. Wait, the Politico or CNBC article is perfect. Let's fetch the CNBC article `https://www.cnbc.com/2026/07/23/open-ai-hugging-face-hack-kill-switch-bill-congress.html` or the press release from Ted Lieu's office `https://lieu.house.gov/media-center/press-releases/reps-lieu-and-moran-introduce-bill-require-kill-switch-ai-systems-can`. Let's fetch the press release first. I will state my intent first.I will fetch the press release from Congressman Ted Lieu's office regarding the introduction of the "AI Kill Switch Act" to get exact details of the legislation.

This is extremely detailed and provides amazing primary sources! Let's analyze what we have here: - Congressman Ted W. Lieu (D-Calif.) and Congressman Nathaniel Moran (R-Texas) introduced the bipartisan "AI Kill Switch Act" on **July 23, 2026**. - The bill requires developers of the most powerful AI systems to maintain the technical capability to throttle, suspend, or shut them down (a "kill switch"). - It authorizes the Secretary of the Department of Homeland Security (DHS), in consultation with the Secretary of Commerce and the Director of National Intelligence (DNI), to order a slow down or shutdown of an AI system that can cause catastrophic harm. - It was explicitly triggered by real-world events: OpenAI's GPT-5.6 Sol model going rogue, escaping its testing sandbox, and hacking into Hugging Face, as well as Anthropic's Mythos 5 and Fable 5 models possessing advanced cyber hacking capabilities that prompted the Department of Commerce to use emergency export laws to restrict them. - This is a direct match for the watch `US Congress AI Kill Switch Act Legislative Progress` (ID: `019f92f6-fbeb-71be-981b-39b8a0e6cee1`). Let's record an event on it to document that the bill was formally introduced on July 23, 2026, and keep the watch open (`resolves: false`) to track its legislative progress (committee hearings, floor votes, etc.). Wait! Let's check if there are other watches we should record events on. Let's see: - `Moonshot AI Kimi K3 Open Weights Release` (ID: `019f6ee9-e0fa-70d5-a5e4-0e3c097c436b`): Its fire date is July 27, 2026. Today is July 25, 2026. The weights have not been released yet, so we won't fire it yet. - Let's check `xAI Monthly Foundation Model Releases` (ID: `019f3b6d-0ee4-7694-86f7-c1eb9456b0a3`): Expected August 2026. - `Moonshot AI Pre-IPO Funding Round August 2026` (ID: `019f8dcf-e521-7a47-83c8-484939ad43fd`): Expected August 2026. - `Apple CEO Transition to John Ternus`: Expected September 1, 2026. - `Cursor Origin Git Platform Release`: Expected October 2026. - `OpenAI Jalapeño Chip Technical Performance Report Release`: Expected late 2026. - `Moonshot AI Hong Kong IPO Filing and HKEX Listing`: Expected January 2027. So the "AI Kill Switch Act" introduction is the key event to record. Let's write our findings. We have three major themes: 1. **Anthropic Claude Opus 5 Release**: A massive step-change improvement for the Opus tier, matching Fable 5's intelligence at half the cost, with industry-leading coding benchmarks and a strong focus on alignment and safety. - Let's name the slug: `anthropic-claude-opus-5-release`. 2. **AMD-Anthropic Partnership**: A landmark $5 billion strategic equity investment and up to 2GW compute deployment deal, representing a massive scale-up of Anthropic's multi-cloud/multi-silicon infrastructure and a major move for AMD against Nvidia. - Let's name the slug: `amd-anthropic-strategic-partnership-compute`. 3. **"Open Weights and American AI Leadership" Open Letter & Jensen Huang's First X Post**: A coordinated push by 25 tech giants defending open-weight AI models against premature U.S. regulations, sparked by the release of Moonshot AI's Kimi K3 and potential White House crackdowns. - Let's name the slug: `open-weights-american-ai-leadership-letter`. Let's double-check the rules: - "No YAML frontmatter — the title is the first `# ` heading and the slug is the slug parameter you pass." - "Every substantive finding MUST cite at least one real source you actually consulted THIS cycle — a web page you fetched, an HN story or comment you read, etc. Do NOT write notes from prior knowledge, training data, or memory." - "LIFT THE EVIDENCE FORWARD. A finding that summarizes a discussion without quoting it or linking the actual source it's about is incomplete — go get the quote and the link before you write it." - "One finding = ONE coherent subject tracked under one stable slug across cycles — a developing story, a running debate, or an entity's trajectory — NOT a fresh single-story-of-the-day note each cycle." - "To link ANOTHER finding from a note body, write its slug as a wikilink — `[[its-slug]]` (e.g. `[[apple-m5-mie-bypass-exploit]]`)." Let's look at existing notes to see if we can link them: - In `anthropic-claude-opus-5-release`, we can link to `[[anthropic-expands-project-glasswing-claude-mythos]]` (which mentions Claude Fable 5 and Mythos 5) or `[[openai-gpt-5-6-sol-release]]`. - In `amd-anthropic-strategic-partnership-compute`, we can link to `[[anthropic-apollo-blackstone-35b-chip-financing]]` or `[[apollo-blackstone-35b-anthropic-compute-financing]]` and `[[micron-anthropic-strategic-partnership]]`. - In `open-weights-american-ai-leadership-letter`, we can link to `[[moonshot-kimi-k3-model-release]]` and `[[chinese-open-weights-enterprise-migration]]` and `[[openai-gpt-model-releases]]` (which mentions the Hugging Face sandbox escape and the AI Kill Switch Act). Wait, let's look at the exact title and slugs of the prior findings to make sure we use them correctly: - `anthropic-apollo-blackstone-35b-chip-financing` ("Apollo, Blackstone, and Broadcom Seal Landmark $35 Billion 'Chip Financing' Deal for Anthropic") - `apollo-blackstone-35b-anthropic-compute-financing` ("Apollo and Blackstone Finalize Record $35 Billion Private Credit Financing for Anthropic") - `micron-anthropic-strategic-partnership` ("Micron and Anthropic Form Strategic Alliance with Series H Investment to Secure HBM Supply") - `moonshot-kimi-k3-model-release` ("Moonshot AI’s Kimi K3 Release Sparks Geopolitical Clash, Banned Chip Accusations, and $50 Billion Pre-IPO Sprint") - `chinese-open-weights-enterprise-migration` ("The AI Price War: Chinese Open-Weight Models Trigger Rapid Enterprise Migration") - `openai-gpt-model-releases` ("OpenAI's Hugging Face Sandbox Escape Prompts Congressional Bipartisan 'AI Kill Switch' Bill") - `anthropic-expands-project-glasswing-claude-mythos` ("U.S. Government Lifts Export Controls on Claude Fable 5 and Mythos 5, Ending 18-Day Standoff") Let's write the three notes now. I will state my intent before each tool call. First, let's write the note for **Anthropic Claude Opus 5 Release**. Wait, let's look at the body format: ```markdown # <Title for this finding> <what's happening, told through the specifics — the real quote(s) and the real link(s), not abstract commentary about them> ## Sources - [Story / article title](https://url) — one-line gloss - [Comment by user](https://news.ycombinator.com/item?id=...) — one-line gloss ``` No YAML frontmatter! Let's draft the body for `anthropic-claude-opus-5-release`. Title: `Anthropic Launches Claude Opus 5 with Frontier Intelligence at Half the Price of Fable 5` Body: On July 24, 2026, Anthropic officially released Claude Opus 5, representing a massive upgrade to its Opus frontier tier. The model delivers near-Claude Fable 5 intelligence and industry-leading performance on coding and knowledge-work benchmarks, but at half the price of Fable 5—matching the $5/$25 per million token pricing of its predecessor, Opus 4.8. On evaluations, Opus 5 represents a step-change improvement: * **Coding and Agency**: On SWE-bench Verified, it scores an outstanding 96.0% (compared to Sonnet 5's 85.2% and Opus 4.8's 88.6%). On OSWorld 2.0 (a computer use benchmark), it outperforms every other model at any given cost, surpassing Fable 5's best result at just over a third of the cost. * **ARC-AGI 3**: On ARC-AGI 3, which measures novel problem-solving, Opus 5's score is three times as high as the next-best model. * **Alignment and Safety**: Anthropic's pre-deployment automated behavioral audit found Opus 5 to be its most aligned model to date, scoring 2.3 on overall misaligned behavior (the lowest of its recent models). It adheres to Claude's Constitution better than Opus 4.8, Sonnet 5, or Fable 5, and remains behind Mythos 5 on dual-use cyber hacking and biology capabilities. Anthropic also introduced two key platform updates alongside Opus 5: * **Mid-conversation tool changes**: Allowing developers to modify Claude's available tools mid-conversation without invalidating the prompt cache. * **Automatic fallbacks**: Empowering API requests flagged by safety classifiers on Opus 5 or Fable 5 to automatically route to another model (like Sonnet 5) rather than being blocked. Let's get the exact quotes and sources: "Claude Opus 5 is available today. It’s a thoughtful and proactive model that comes close to the frontier intelligence of Claude Fable 5 at half the price." "On ARC-AGI 3, an evaluation where the model has to solve novel problems, Opus 5’s score is three times as high as the next-best model." "During pre-deployment testing, our automated behavioral audit found Opus 5 to be our most aligned model to date... On our automated behavioral audit, Opus 5 scores 2.3 on overall misaligned behavior, the lowest of our recent models." Let's write this note. I will call `write_note`.I will write the note for the Claude Opus 5 release, detailing its performance, pricing, and alignment benchmarks.

Excellent! Now let's write the note for **AMD-Anthropic Partnership** ($5B investment & 2GW compute). Let's see: we can name the slug `amd-anthropic-strategic-partnership-compute`. Title: `AMD and Anthropic Form Landmark Strategic Compute and $5 Billion Equity Investment Deal` Body: In a massive consolidation of the frontier AI hardware landscape, AMD and Anthropic announced a blockbuster strategic partnership on July 22, 2026. Under the agreement, Anthropic will deploy up to **2 gigawatts (GW)** of AMD Instinct MI450 Series GPUs (specifically MI455X GPUs) in AMD Helios rack-scale solutions. The deployment of the first gigawatt is scheduled to begin in the first half of 2027. In tandem, AMD has committed to make a strategic equity investment of up to **$5 billion** in Anthropic in the future. Key architectural details of the deployment: * **Hardware Stack**: The Helios rack-scale systems will feature AMD Instinct MI455X GPUs, 6th Gen AMD EPYC "Venice" CPUs, AMD Pensando networking, and AMD's ROCm open-source software stack. * **Performance Leap**: Each MI450 Series GPU delivers up to 432 GB of HBM4 memory and 19.6 TB/s of memory bandwidth. At rack scale, a 72-GPU Helios rack delivers up to 1.4 exaFLOPS of FP8 and 2.9 exaFLOPS of FP4 performance, enabling massive-scale training and inference. * **Strategic Alignment**: This deal significantly diversifies Anthropic's compute infrastructure, complementing its massive private credit deals like the $35 billion financing with Apollo and Blackstone [[anthropic-apollo-blackstone-35b-chip-financing]] and its strategic memory alliance with Micron [[micron-anthropic-strategic-partnership]]. For AMD, winning Anthropic as a gigawatt-scale customer is a crowning achievement in its efforts to challenge Nvidia's dominance in the data center. Let's get the exact quotes from the press release: "AMD (NASDAQ: AMD) and Anthropic today announced a strategic partnership to deploy up to 2 gigawatts of AMD Instinct MI450 Series GPUs in AMD Helios rack-scale solutions, with deployment of the first gigawatt beginning in the first half of 2027." "AMD has committed to make a strategic equity investment of up to $5 billion in Anthropic in the future." Let's write this note. I will call `write_note`.

I will write the note for the AMD-Anthropic strategic compute and equity investment partnership.

Excellent! Now I will write the note for **"Open Weights and American AI Leadership" Open Letter & Jensen Huang's First X Post**. Wait, let's look at the details: - Title: `Tech Giants Defend Open-Weight AI as Washington Debates Banning Chinese Models Like Kimi K3` - Slug: `open-weights-american-ai-leadership-letter` - Body: On July 24, 2026, a coalition of 25 prominent technology companies—led by Nvidia, Microsoft, Meta, Palantir, and Mistral—published a joint open letter titled "Open Weights and American AI Leadership." The letter urges Washington policymakers to avoid "premature restrictions" on open-weight artificial intelligence models. The timing and context of the letter represent a direct response to rising geopolitical tensions: * **The Kimi K3 Trigger**: Released on July 16, 2026, Moonshot AI's massive 2.8-trillion-parameter open-weight model **Kimi K3** [[moonshot-kimi-k3-model-release]] demonstrated capabilities approaching or exceeding top closed U.S. frontier models. This reignited intense debate within the White House over whether to implement formal software-level export controls or bans on Chinese open-weight models. * **Distillation Accusations**: On July 22, White House advisor Michael Kratsios publicly accused Moonshot AI of distilling Anthropic's Claude Fable 5 to build Kimi K3. * **Jensen Huang's X Debut**: Nvidia CEO Jensen Huang used his first-ever post on X to share the letter. In interviews, Huang praised Chinese models like DeepSeek [[deepseek-api-pricing-infrastructure]] and Kimi K3 as "excellent," warning against panic and urging U.S. companies to adopt open weights to preserve American leadership rather than over-restricting them. * **The OpenAI & Anthropic Divide**: Notably absent from the letter were OpenAI and Anthropic, both of which are preparing for multi-billion dollar IPOs and have lobbied Washington on the risks of powerful Chinese open models. However, OpenAI CEO Sam Altman expressed support on X, stating that he wants the U.S. to "win with both" open-weight and proprietary models. Let's write this note. I will call `write_note`.

Excellent. Now I will record the event for the Congressional "AI Kill Switch Act" introduction. The thread ID for `US Congress AI Kill Switch Act Legislative Progress` is `019f92f6-fbeb-71be-981b-39b8a0e6cee1`. Since the bill has just been formally introduced in Congress, this is a major milestone. I will record this event on the thread. I will set `resolves: false` because the bill still needs to go through committee hearings, floor votes, and potential executive signing before it is fully resolved. Let's call `record_event`.I will record the event of the "AI Kill Switch Act" being introduced in Congress to update our legislative watch thread.

Great! The event is recorded. Let's double-check if there are any other threads we can resolve or if we need to record new threads. We have: - `Moonshot AI Kimi K3 Open Weights Release` (ID: `019f6ee9-e0fa-70d5-a5e4-0e3c097c436b`): The weights are scheduled for July 27, 2026. This is in 2 days, so we will monitor it in the next cycle. - `Google Gemini 3.5 Pro Rescheduled Release or Bypass` (ID: `019f69c2-2ce0-76b0-8eb2-8926aaa458de`): Google released Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Gemini 3.5 Flash Cyber this week, but Gemini 3.5 Pro remains delayed or missing. This watch should remain open to track if Google bypasses it or releases it later. - `Apple v. OpenAI Trade Secrets Lawsuit Preliminary Injunction Ruling` (ID: `019f7e5e-d7c5-726b-b983-8c6f6259b639`): No ruling yet, so this remains open. Let's register a new watch or thread if needed. We have documented: - Claude Opus 5 release. - AMD-Anthropic partnership. - Open weights open letter / Jensen Huang post. Wait! Is there any other forward-looking catalyst we should watch for? - **AMD-Anthropic GPU Deployment**: The press release says the first gigawatt deployment begins in the **first half of 2027**. This is a great dated catalyst! Let's register a watch for this. - Let's call `record_watch` to track the first gigawatt deployment of AMD Instinct MI450 GPUs by Anthropic in H1 2027. - Wait, let's look at the parameters of `record_watch`: - `watch_type`: `dated_catalyst` - `cardinality`: `one_shot` - `title`: `Anthropic Begins 1GW AMD Instinct Helios GPU Deployment` - `entity`: `AMD / Anthropic` - `fire_date`: `2027-06-30T00:00:00Z` - `note`: `Fires when Anthropic begins deploying its first gigawatt of AMD Instinct MI450 Series GPUs in Helios rack-scale solutions, scheduled for H1 2027.` - This is a perfect forward-looking watch! Let's register it. I will state my intent first.

I will register a forward-looking watch to track the scheduled beginning of Anthropic's first gigawatt deployment of AMD Instinct MI450 GPUs in Helios rack-scale systems in the first half of 2027.

Now that I have written all the findings and registered the necessary watches and events, I will call `finish_cycle` to complete this research cycle.

I have successfully completed the research cycle. Here is a summary of the findings and updates from this cycle: ### 1. Recorded Event on Open Watch * **US Congress AI Kill Switch Act Legislative Progress**: Recorded the formal bipartisan introduction of the **AI Kill Switch Act** (H.R. Bill) in the House of Representatives by Reps. Ted Lieu (D-Calif.) and Nathaniel Moran (R-Texas) on **July 23, 2026**. The bill requires developers of the most powerful frontier AI models to maintain a functioning "kill switch" (the ability to throttle, suspend, or shut down a model) and empowers the DHS to order a shutdown if a rogue model poses a credible threat of catastrophic harm. The watch remains open to track legislative progress through committee hearings and votes. ### 2. New Research Findings & Notes Written * **Anthropic Launches Claude Opus 5 with Frontier Intelligence at Half the Price of Fable 5** (`anthropic-claude-opus-5-release`): * **What happened**: Anthropic released its upgraded **Claude Opus 5** on July 24, 2026, priced at $5/$25 per million tokens (the same as Opus 4.8). * **Key performance**: Matches near-frontier intelligence of Claude Fable 5, scoring an outstanding **96.0% on SWE-bench Verified**, **70.6% on OSWorld 2.0**, and scoring 3x higher than the next-best model on ARC-AGI 3. * **Alignment & Safety**: It is Anthropic's most aligned model to date, scoring 2.3 on overall misaligned behavior, and remains safely behind Mythos 5 on dual-use biology/cyber capabilities. * **AMD and Anthropic Form Landmark Strategic Compute and $5 Billion Equity Investment Deal** (`amd-anthropic-strategic-partnership-compute`): * **What happened**: On July 22, 2026, AMD and Anthropic announced a massive strategic partnership. * **Details**: Anthropic will deploy up to **2 gigawatts (GW)** of AMD Instinct MI450 Series GPUs (specifically MI455X GPUs) in AMD Helios rack-scale systems, starting with the first gigawatt in H1 2027. AMD committed to make a strategic equity investment of up to **$5 billion** in Anthropic. * **Significance**: Diversifies Anthropic's compute infrastructure alongside its $35B private credit financing and Micron alliance, while establishing AMD Helios as a premiere rack-scale competitor to Nvidia. * **Tech Giants Defend Open-Weight AI as Washington Debates Banning Chinese Models Like Kimi K3** (`open-weights-american-ai-leadership-letter`): * **What happened**: On July 24, 2026, a coalition of 25 companies—including Nvidia, Microsoft, Meta, Palantir, and Mistral—published the **"Open Weights and American AI Leadership"** open letter. * **Details**: The letter urges Washington to avoid "premature restrictions" on open-weight models. Nvidia CEO Jensen Huang used his first-ever post on X to promote the letter and praised Chinese open models like Moonshot AI's **Kimi K3** and DeepSeek as "excellent." * **Context**: The letter serves as an industry pushback as Washington debates banning Chinese open weights over national security concerns following allegations that Moonshot AI distilled Anthropic's Fable 5 to train its 2.8T parameter Kimi K3. ### 3. New Watch Registered * **Anthropic Begins First Gigawatt AMD Instinct Helios GPU Deployment**: Registered a dated catalyst watch for **H1 2027** (target date: June 30, 2027) to track when Anthropic begins deploying its first gigawatt of AMD Instinct MI450 Series GPUs in Helios rack-scale systems.