It's Monday, November 2nd: Mistral put the weights of its trillion-parameter Mistral Large 4 on Hugging Face for coding agents and cyber defenders, and Alphabet closed its worst month since its 2004 IPO after a search miss and AI spending pushed free cash flow negative.
👋 Farewell AIC!
By Noah Frank, Head of Marketing @ AIC
This is my last column for The AI Collective. I’m leaving AIC, and before I go, I wanted to reflect on some of the questions I’ve explored here, and invite you to keep thinking about them with me.
Helping grow this newsletter with the team, while writing these columns each week, has been a substantial part of my work on AI. I’ve appreciated having somewhere to work through ideas while the things we’re writing about are still unfolding. There’s plenty I’m leaving with that I haven’t figured out yet.
In “Reality 2 and the Adoption Gap,” I wrote about the distance between what AI makes technically possible and what actually changes in people’s lives. Making something easier to produce doesn’t settle what’s worth producing, who needs it, or how it finds its way into an organization. As knowledge work gets cheaper, the judgment involved in those decisions deserves more attention.
I remain optimistic about what we can do with this technology. I also think getting to a better future will be much more complicated than the capabilities alone suggest. Employers will make decisions about how work gets reorganized. Governments will have to consider how people can support themselves through the transition, often before the evidence feels sufficient. Those choices will help determine who benefits and who gets left trying to catch up.
For Gen Z, the stakes are particularly immediate. We’re making decisions about our education and careers while trying to understand how much the work itself will change. If companies automate more junior tasks, how will people acquire the experience needed to do senior work? How quickly could that happen, and what would tell us it’s happening? These are questions I want to keep investigating. As I’ve written here before, sometimes good AI policy is just good policy.
That work continues through my research at Centaurian AI and my own writing. On my Substack, Borderline Genius, I’ll take a closer look at the economic situation confronting Gen Z, test claims about AI against its effects on real people, and explore how the structures we’ve built shape what happens next. If you’ve enjoyed these columns, I’d love to have you join me there.
Before I go, thank you to AJ and Catherine for the opportunity, and to Joy, Lindsay, Josh, and Tina for everything you’ve put into the newsletter and the community around it. I’m grateful to have worked with you. The newsletter continues with the team, and I hope you’ll keep reading.
And thank you for making room for my writing in your inbox. I’ve appreciated the chance to think through these questions with you, and I’d love to stay in touch.
— Noah
Your AI Agent Works in a Demo. But Can It Survive Production? 🚀
Building an AI agent in an afternoon is easy. Getting it to operate autonomously, access real-time data, process payments, and deliver reliable results in production? That's where things get interesting.
On November 10 in San Francisco, some of the brightest minds in AI are coming together for Run by Apify, a one-day conference dedicated to taking AI agents from impressive demos to real-world deployment.
Here's who's taking the stage:
OpenAI × WorkOS × x402: Kevin Whinnery (OpenAI), Michael Grinich (WorkOS), and Kevin Leffew (x402 co-author) explore what happens when AI agents become first-class citizens of the internet.
Apify × Stripe: Apify founder Jan Čurn and Stripe's Lauren Dai unpack the shift from humans clicking through interfaces to autonomous AI workers getting things done.
IBM: Tejas Kumar dives into agentic taste - how to tell whether your AI agent is actually doing good work, featuring a live before-and-after demo.
And it's not just talks. The Apify team is flying in for the occasion, giving attendees a rare opportunity to connect directly with the builders shaping the future of autonomous AI.
The day wraps up with a networking party alongside speakers, founders, engineers, and the Apify team. 🥂
📅 When: Tuesday, November 10, 2026 | Doors open at 10 AM
📍 Where: The Pearl, 601 19th Street, San Francisco
🎟️ Exclusive AI Collective Discount: 50% OFF
Use code AICOLLECTIVE to get half off your ticket.
Covering what’s happening on the ground in AI, every Monday.
1️⃣ OPEN DEFENSE: Mistral Gives Defenders a Trillion-Parameter Model to Self-Host
Image from Hugging Face.
Mistral released the open weights of Mistral Large 4, a model with about one trillion parameters aimed at coding agents and cyber defense, so companies and governments can run it on hardware they control.
The Hugging Face model card lists a mixture-of-experts design with 49 billion active parameters per token, a 1M-token context window and 128 routed experts. Mistral says it trained the model from scratch on 3,800 NVIDIA Grace Blackwell GPUs in Europe.
In its release post, Mistral says extra reinforcement learning since the public preview lifted its Terminal-Bench 4.0 score by 13 points to 41%. It also claims a tie with GLM-5.3 on the Artificial Analysis Coding Agent Index while using 47% fewer tokens.
Mistral reports a 96.6% average refusal rate on malicious cyber requests and is previewing a specialized cyber variant with vetted partners such as Orange Cyberdefense. The company also writes that it sees "a potential for uplift to conduct real attacks" and is adjusting its guardrails.
TechCrunch frames the release as a cybersecurity play, and Cloudflare's Stephanie Cohen argues in Mistral's post that a defense teams cannot control is exposure. The weights ship under the Mistral Ecosystem Open-Weight License, an MIT derivative that restricts offering the model as a hosted service.
Security teams that cannot send incident data to an outside API now have a frontier-scale model they can host and audit themselves. Check the license terms and the hardware bill first, since Hugging Face notes that only its 2-bit quants fit on a single 512 GB Apple Silicon Ultra machine.
2️⃣ CASH BURN: Alphabet Sheds $758 Billion In Its Worst Month Ever
Image from Google.
Alphabet lost about $758 billion in market value in October, its worst month since its 2004 IPO, as third-quarter results paired a search revenue miss with $9.1 billion of negative free cash flow.
Yahoo Finance calculates the stock fell nearly 19% in October, its worst month as a public company, and lost $475 billion on Oct. 29 alone. Only Nvidia has had a bigger one-day loss, about $593 billion on the day DeepSeek rattled AI investors in January 2025.
Alphabet's third-quarter release shows Google Cloud revenue up 103% to $30.7 billion, with TPU system sales a meaningful contributor. Search and other revenue grew 12% to $63.4 billion, short of the $64.96 billion estimate CNBC cited, and quarterly capex reached $51.4 billion.
CFO Anat Ashkenazi held 2026 capex guidance at $195 billion to $205 billion and told analysts on the earnings call, "We continue to expect our CapEx to increase significantly in 2027." She also flagged near-term Cloud margin pressure from renting third-party compute in a "supply constrained environment."
In Sundar Pichai's remarks, Google's first-party model APIs now process nearly 32 billion tokens per minute, more than four times a year ago, and the Gemini app has passed one billion users. D.A. Davidson's Gil Luria told CNBC about Search, "This may be the quarter we feared."
Developers on Gemini and Google Cloud are the customers this spending serves, and Pichai and Ashkenazi both say capacity still trails demand. Keep a second provider ready for capacity-bound workloads, and watch the Q4 call for the 2027 capex figure that sets how fast new supply arrives.
📰 Other Headlines
OPENAI MODEL SPEC: OpenAI rewrote its Model Spec around five core values, retired most of its detailed style guidance and now calls the model "GPT" instead of "the assistant."
MICROSOFT'S OPENAI REVENUE: Microsoft booked $10 billion from OpenAI arrangements in the quarter ended September 30, nearly 11% of its $92.3 billion total, even as the two compete for customers.
CLAUDE SPEND REPORTS: Anthropic launched reporting tools that show enterprise managers what each team produced with Claude, what it cost and where workflows broke, plus pooled group budgets.
GEMINI EDITS FILES: Google's Gemini app now creates and edits files inside the chat, with PDFs on web and mobile and Docs, Sheets, slides and code on the web.
NVIDIA OPEN TRANSCRIPTION: Nvidia released Nemotron 3.5 Transcribe, a free 30-billion-parameter open-weight model that transcribes up to 40 languages and separates up to eight speakers in hour-long recordings.
DEEPGRAM BUYS DAILY: Voice AI company Deepgram is acquiring Daily, maker of the popular Pipecat voice agent framework, and both companies say Pipecat will stay open source.
GRANOLA ADDS DICTATION: Meeting notetaker Granola launched a dictation feature for Mac that works across apps with unlimited words on every plan, moving onto rival Wispr's turf.
DOORDASH AGENT ORDERS: DoorDash CEO Tony Xu said third-party AI agents have placed only "several hundred" orders in total, against roughly 3 million deliveries a day.
OPERA LOCAL AI: Opera introduced WebAI, a proposed web standard that lets open-weight models such as Google's Gemma run locally in the browser, now in early preview.
🫵 Want your message in front of 200,000 AI builders?
Our partners and sponsors get exclusive placements across the newsletter and access to AIC's in-person network — demo nights, dinners, hackathons, and forums across 180+ chapters.
For all inquiries, send us a note at [email protected].
The AI Collective is built by volunteers across 180+ chapters in 40 countries.
Thank you to the thousands of volunteers around the world who make this work possible. We truly could not do this without you.
🧑💻 About the Editors
About Noah Frank
Noah is a researcher, innovation strategist, and ex-founder thinking and writing about the future of AI and the workforce. His work and body of research explores the economics of emerging technology and organizational strategy. Outside of AIC, Noah heads research for Centaurian AI.
About Joy Dong
Joy is a news editor, writer, and entrepreneur at the intersection of AI and blockchain. Whether she is demystifying complex systems in her newsletter, TEA, or building streamlined solutions through her automation agency, Ownly, Joy’s mission is to make emerging tech accessible and actionable for everyone.
About Lindsay Gross
Lindsay is an AI engineer, researcher, and writer focused on how AI systems behave in practice and what it takes to make them safe. Her work sits at the intersection of AI safety, governance, and product design, and at AIC she writes about the questions that matter most as these systems scale.
1
