Last Week in AI & Tech: NVIDIA’s $20B Speed Grab, a Copyright War, and the Agent Boom
Hey everyone — last week in AI wasn’t about “who has the smartest chatbot.” It was about who can make AI faster, who controls the data, and how AI becomes a worker (not just a talker).
Here’s the full roundup. 👇
🚀 1) NVIDIA’s reported $20B Groq deal: AI speed is the next power move
NVIDIA reportedly signed a $20$20 billion non-exclusive licensing deal with Groq, the company known for super-fast AI inference chips.
The big twist: Groq still stays independent as a cloud provider, but this is being described like an “acqui-hire” because Groq’s founder Jonathan Ross (often linked to the Google TPU story) and President Sunny Madra are reportedly joining NVIDIA to lead a new inference-scaling division.
Why you should care: speed is what turns AI from “cool demo” into “always-on assistant.”
Faster AI = faster replies, cheaper running costs, better real-time tools ⚡
It’s especially important for agents that take many steps (read → decide → act) 🧠
NVIDIA is also said to be integrating Groq’s SRAM-based LPU style tech into its “AI Factory” architecture 🏭
And yes, the headline claim making rounds is “10×10× faster.” Even if that number is marketing, NVIDIA is clearly trying to own the inference layer—where the real money will sit.
⚖️ 2) A major copyright lawsuit: authors vs AI giants
Investigative journalist John Carreyrou (Bad Blood) and five other authors have filed a copyright lawsuit against:
Google, OpenAI, Meta, Anthropic, xAI, and Perplexity.
What makes it spicy: instead of a broad class action, it’s described as individual claims going for max statutory damages—up to $150,000$150,000 per work.
The core accusation: these companies allegedly trained models on books taken from “shadow libraries” like LibGen and Z-Library—without permission or payment.
Why it matters: the next generation of AI may depend less on “more data” and more on “legal data.”
🧩 3) Anthropic launches “Agent Skills”: a cleaner way to build AI workers
Anthropic published “Agent Skills” as an open standard—basically a way to package agent abilities into modular skills that can be loaded when needed.
Instead of writing one giant prompt that tries to cover everything, agents can use “progressive disclosure.” Think: load the right skill at the right time.
Examples mentioned:
“PDF form filler” skill 📄
“Excel analyst” skill 📊
Why this is important: it makes agents easier to reuse, update, and control—especially in workplaces.
🏪 4) Agent hubs are exploding: Agents.blog + AI Agent Store
Sites like agents.blog and aiagentstore.ai are starting to act like the “Product Hunt for agents.”
They’re becoming places where people discover:
New agents and workflows 🤖
Daily updates and research bites 📰
Hackathon projects and tools 🛠️
A claim in the roundup says over 96% of enterprises are moving beyond chatbots toward agents that actually execute tasks. Whether the exact number is perfect or not, the direction is real: companies want AI that does the work, not just chats about it.
🔥 10 rapid-fire updates you’ll see everywhere
Here are the ten extra headlines you listed, explained in simple terms:
1) 🇨🇳 GLM “beats GPT” on hard math
GLM 4.7 reportedly scored 95.7 on “the hardest math test,” described as beating GPT-5.
2) 🧠 Grok milestone
“Grok 4 Heavy” is described as the first AI to score 50 on “Humanity’s Last Exam,” framed as an “impossible” test.
3) 🎮 YouTube prompt-to-games
YouTube is described as letting creators build games from text prompts using “Gemini 3.”
4) 📚 Qwen “infinite memory”
“Qwenlong” is said to support 4 million tokens of context—basically “uploading an entire library.”
5) 💻 Cheap pro coding
“Minimax M2.1” is positioned as “Claude level coding at 10% of the price,” with the claim that the $20$20/month AI subscription is “officially dead.”
6) 🗂️ Agents replace SaaS stacks
ClickUp’s new AI agents are described as doing real work (schedule, email, complete tasks), hinting that multiple apps could merge into one agent layer.
7) ⚡ NVIDIA’s speed bet (again)
NVIDIA is said to have paid $20B for Groq’s tech to make AI “10× faster,” framed as monopoly-building.
8) 🏙️ Two economies by 2026
Claude’s co-founder is quoted saying AI users vs non-users may live in “completely different economies” by 2026.
9) 🇪🇺 Mistral’s coding push
Mistral’s new coding model is described as 7×7× cheaper than GPT-4 and important for keeping Europe competitive.
🔥 12 AI Tools Dropped Last Week & They’re Insane
Last week have been one of those weeks where AI just refuses to slow down. I counted 12 major tool releases in the last few days, and honestly, I couldn’t pick favorites if I tried.
Here’s the quick rundown of what you need to know.
🧭 The simple takeaway
This week was a preview of what’s coming:
Speed will decide which AI tools feel “magical” (NVIDIA + Groq) ⚡
Copyright will decide what data AI can learn from (authors vs AI giants) ⚖️
Skills + agents will decide who actually gets work done (Anthropic + agent hubs) 🤖
We’re not just chatting with AI anymore. We’re moving toward AI that runs tasks, workflows, and whole business processes.
Your Laptop Screen Just Got... Invisible? The Spacetop G1 is Here to Change Remote Work Forever
Remember when people were obsessed with getting thinner and lighter laptops? Well, Sightful just flipped the script—they didn’t just make the laptop lighter, they literally got rid of the screen.
Meet Spacetop G1, and I’m not going to lie... it’s wild.
Humanoid Robots Mega-Roundup (2025 → 2026) 🤖⚡
This newsletter is the grand finale of our mini-series on robotics—now spotlighting the Top Humanoid Robots of 2025.
Humanoid robots have officially moved from “cool lab videos” to factory pilots, home prototypes, and early mass-production plans—especially as China pours serious money into embodied AI and manufacturing scale.
Let’s Explore.
The Shocking Truth: What AI Tools Are Startups Really Spending Money On?
Where 200,000+ Startups Are Investing—And Why It Matters For You
Get a behind-the-scenes look at which AI apps and platforms startups are actually paying for—and why. If you want to stay ahead of the tech curve (or spend smarter), don’t miss these insights and success stories!
👉 Find out where the smart money is going now.
7 Free AI Tools You’ll Wish You Found Sooner
Productivity Boosters for Research, Writing, and Design
Faster research, cleaner writing, easier design—no experience needed! See how free apps like Perplexity, Opus Clip, Gamma, and more can shrink your task list and help you get work done in record time.
👉 Discover today’s top time-saving AI tools.
📨 Cut through the noise!
Follow us on WhatsApp, LinkedIn and Telegram to deliver clean, curated updates on AI and tech—no distractions, just the news you need. 🚀🧠✨