- AI Fire
- Posts
- ⚔️ Nvidia Monopoly Under Attack
In partnership with
For two years, everyone relied on Nvidia. Now Amazon wants to sell its own AI chips to the world. At the same time, frontier models are raising security alarms… and researchers just trained AI to detect heart failure years earlier than current medicine.
What's on FIRE 🔥
IN PARTNERSHIP WITH HUBSPOT
Want to get the most out of ChatGPT?
ChatGPT is a superpower if you know how to use it correctly.
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
AI INSIGHTS
The biggest story in tech for the last 2 years has been "Everyone vs. Nvidia." But in his 2025 Letter to Shareholders, CEO Andy Jassy signaled that Amazon is a semiconductor powerhouse with a $20B annual run rate for its custom AI chips.
The annual letter reveals a company that is no longer content just being a cloud provider. Amazon is pivoting to become a merchant of the hardware itself.
$20 Billion Milestone: Amazon’s internal silicon strategy is scaling faster than almost any other non-Nvidia player.
Direct Chip Sales: In a major strategic shift, Amazon is considering selling its AI chips directly to other companies → a direct competitor with Nvidia and AMD in the open market.
Defending the Capex: Jassy addressed concerns about massive AI spending, describing the current moment as a "once-in-a-generation" shift.
Agentic Focus: The letter emphasized that Amazon's future AI investments are laser-focused on "long-horizon" agentic systems.
Amazon has always used its own chips as a way to lower costs for AWS customers. Now, they are leveraging their massive supply chain and data center footprint to offer a high-performance, lower-cost alternative to the "Nvidia Tax."
The "Silicon War" is no longer just about who can buy the most H100s. It’s about who can build the most efficient brains for the next billion agents.
PRESENTED BY MINTLIFY
Are you tracking agent views on your docs?
AI agents already outnumber human visitors to your docs — now you can track them.
PREMIUM AI SOURCES FROM AI FIRE
1. Top 9 Pre-Made Claude AI Skills to Get Back 25+ Hours a Week (Full Power User Guide). Discover the Claude AI skills most users never unlock. These advanced methods quietly replace hours of manual work every single week. Don't go back to manual workflows.
2. Complete Claude Cowork Tutorial to Run an Entire AI Workforce (Full 2026 Guide). Claude Cowork sub-agents crush hours of research in seconds. Link your memory and skills to build a personal workforce that stays active for you all day.
3. ALL Working Prompts in Only 5 Minutes with NotebookLM (Included 1,500+ Free Prompts). Use NotebookLM to turn a massive open-source prompt database into a private system that finds the right prompt when you need it.
4. Prompting Might Be the Wrong Way to Use AI. Stop! Do This Reverse Way Instead. The top 9% don’t “prompt” AI the same way. They build systems. If your AI outputs feel average, this is why. The reverse method fixes that.
5. 7 Best AI Businesses You Can Absolutely Run Solo with Claude Agents. Some people are building $10K/month businesses without hiring anyone. These AI businesses don’t need employees. Claude agents handle the work.
FIRE RECAP: BIGGEST AI NEWS THIS WEEK
⚠️ Anthropic’s Claude Mythos is raising global security alarms after reports it can autonomously discover software vulnerabilities at scale. Governments are already discussing national security implications.
🤝 OpenAI, Google, and Anthropic are quietly teaming up to protect frontier models from model theft, signaling AI weights are becoming strategic assets like chips or nuclear tech.
⚡ Anthropic just secured ~3.5 gigawatts of compute through a major deal with Google & Broadcom, showing the insane scale of future AI training runs. The real race is now about TPU compute.
🐛 Security researchers say AI is flooding bug bounty programs with vulnerability reports faster than humans can handle, hinting that AI hacking could soon become largely automated.
🧠 Facing growing public distrust, OpenAI and Anthropic are funding policy papers and think tanks to reshape the AI narrative. The battle is shifting toward AI policy influence and regulation debates.
TODAY IN AI
AI HIGHLIGHTS
🔥 A man allegedly threw a Molotov cocktail at Sam Altman, then reportedly threatened to burn OpenAI’s HQ. No injuries, suspect now in custody. The tension around AI keeps escalating.
🏦 U.S. officials quietly warned major banks about Claude Mythos Preview, saying Anthropic’s powerful model could expose hidden security flaws that hackers might exploit.
⚡ DeepMind CEO Demis Hassabis says AGI could feel like ten industrial revolutions compressed into just one decade. He still believes it could arrive within five years.
⚖️ OpenAI is backing a proposed law that could limit lawsuits tied to catastrophic AI damage. The AI Safety Act could reshape who is responsible when frontier models cause harm.
🧠 Anthropic is reportedly exploring building its own AI chips, joining OpenAI, Meta, and others racing to control compute supply as demand explodes.
💰 AI M&A: Canva has acquired Simtheory, an agentic AI workspace for building custom AI assistants, alongside marketing automation platform Ortto. The deal expands Canva’s AI and marketing stack, strengthening its push to become an end-to-end system for ideation, creation, and campaign execution across teams worldwide.
NEW EMPOWERED AI TOOLS
📝 Claude for Word enables native AI drafting and editing inside Microsoft Word, with tracked changes preserved.
☁️ Claude Code ultraplan moves code planning to the cloud, letting you review and approve architecture before execution.
🖱️ Clicky is an AI assistant that sees your screen, guiding you step-by-step across apps on Mac.
🎯 aperture replaces resume screening with AI interviews, ranking candidates by real performance.
AI BREAKTHROUGH
Heart failure is usually caught too late. But Oxford researchers just built an AI that turns routine CT scans into an early-warning system by reading "invisible" changes in heart fat. Here’s how it works:
Fat surrounding the heart (pericardial fat) changes when the muscle underneath is inflamed. The AI reads these subtle textural shifts.
Tested on 72,000 patients, the AI hit 86% accuracy in predicting future heart failure.
Patients in the highest-risk group were 20 times more likely to suffer heart failure than those in the low-risk group.
Oxford is already working with the NHS to roll this out across all chest CT scans, turning existing medical data into a preventative shield.
The real win here is timing. Heart failure is usually diagnosed only after the damage is done. By flagging high-risk patients five years out, doctors can intervene with lifestyle changes or medication while the heart is still healthy.
What makes this even a game-changer is the cost-efficiency. We’re simply adding an "intelligence layer" to the scans already happening. If you get a CT for a cough, the AI can check your heart fat and tell you to change your habits today to avoid a hospital bed in 2031.
We read your emails, comments, and poll replies daily
How would you rate today’s newsletter?Your feedback helps us create the best newsletter possible |
Hit reply and say Hello – we'd love to hear from you!
Like what you're reading? Forward it to friends, and they can sign up here.
Cheers,
The AI Fire Team
Reply