🚀 Nvidia drops a $1 Trillion bombshell

Good Morning! Jensen Huang just turned a hockey arena into the epicenter of the AI world, dropping a staggering $1 trillion demand outlook and unveiling Nvidia's massive shift from chipmaker to the operating system of the AI economy. Plus, you will learn how to instantly transform long-form YouTube videos into stunning, easy-to-read infographics.

Plus, in today’s AI newsletter:

• Nvidia Drops a $1 Trillion Bombshell at GTC 2026

• Moonshot AI Rethinks the Transformer with Attention Residuals

• How to Create an Infographic From Any YouTube Video Using AI

• Manus Drops "My Computer" for Local Desktop Control

• 4 new AI tools worth trying

AI NEWS

🚀 Nvidia Drops a $1 Trillion Bombshell at GTC 2026

At GTC 2026 in San Jose, Nvidia CEO Jensen Huang made it clear the company is no longer just selling GPUs. By unveiling the Vera Rubin supercomputer and the first hardware from its massive Groq acquisition, Nvidia is building the vertically integrated infrastructure backbone for an entirely new computing era.

• Nvidia doubled its AI demand outlook to $1 trillion through 2027, powered by the new Vera Rubin platform, a massive seven-chip, five-rack supercomputer system that delivers a 10x improvement in performance per watt over Grace Blackwell.

• The $20 billion Groq acquisition is already paying off: the Groq 3 LPU ships in Q3. When paired with Rubin in dedicated LPX racks, it boosts inference throughput by an insane 35x per megawatt, solving the massive latency-vs-throughput bottleneck for large-scale AI.

• Nvidia launched NemoClaw, an enterprise-grade reference stack for OpenClaw, positioning itself as the foundational operating system layer for autonomous AI agents.

Nvidia's GTC 2026 keynote marks the exact moment the company stopped being "just" a chipmaker and fully transitioned into the vertically integrated platform powering the entire AI economy. With a $1 trillion demand pipeline and seven new chips shipping as one unified system, the infrastructure decisions Nvidia makes today will dictate the speed, cost, and capabilities of AI for the next decade.

AI RESEARCH

🧠 Moonshot AI Rethinks the Transformer with Attention Residuals

Moonshot AI researchers have released "Attention Residuals" (AttnRes), a new method that replaces the fixed residual mixing in modern Transformers with depth-wise attention. Instead of blindly adding every previous layer's output together, AttnRes allows each layer to selectively choose which earlier representations to use.

• Standard residual connections force all prior layers to accumulate with fixed weights, causing memory states to grow uncontrollably and diluting the impact of early layers.

• AttnRes fixes this by using softmax attention over depth, letting layers dynamically pull only the information they need from earlier in the network.

• To handle the massive memory costs of full depth-wise attention, Moonshot introduced "Block AttnRes," which groups layers into blocks and reduces the memory and communication overhead from O(Ld) to O(Nd).

• In scaling tests, Block AttnRes matched the performance of baseline models that used 1.25x more compute.

Residual connections are one of the most foundational, unquestioned components of modern AI models like GPT, Claude, and LLaMA. By proving that a lightweight depth-wise attention mechanism can fix information dilution and save massive amounts of compute, Moonshot AI might have just triggered a fundamental shift in how the next generation of frontier models will be built.

AI TOOLS

💻 Manus Drops "My Computer" for Local Desktop Control

Manus has introduced "My Computer," a new core capability within its Manus Desktop application that allows its AI agent to operate directly on your local machine. Available immediately for both macOS and Windows users, this update bridges the massive gap between cloud intelligence and local compute power.

• Manus can now access, read, edit, and organize local files, effectively automating tedious file management tasks right on your desktop.

• The agent can execute command-line instructions and fully control local applications, letting it handle complex app development and remote task execution.

• It taps directly into your machine's hardware, meaning you can finally use your idle local GPUs for machine learning or AI inference tasks.

• Users retain granular control over permissions, requiring human approval for actions to maintain security while the AI works in the background.

Until now, most powerful AI agents have been trapped in cloud sandboxes, entirely separated from the local environments where our actual work happens. By giving Manus direct access to the command line, local files, and onboard GPUs, the platform is transforming from a remote chat assistant into a deeply integrated, local automation engine. It’s a massive leap forward in making AI agents practically useful for daily workflows.

HOW TO AI

🗂️ How to Create an Infographic From Any YouTube Video Using AI

In this tutorial, you will learn how to instantly transform long-form YouTube videos into stunning, easy-to-read infographics using Gemini's video analysis and image generation capabilities, saving you hours of note-taking and design work.

🧰 Who is This For

• Content creators looking to repurpose video content into highly shareable social media posts

• Students and researchers wanting visual summaries of long lectures and complex tutorials

• Marketers needing to quickly digest and present industry insights from webinars

• Anyone who wants to turn dense video information into beautiful, bite-sized graphics without needing design skills

STEP 1: Access Gemini and Enable Thinking Mode

Head over to the Gemini website or open the mobile app on your device. To ensure the AI deeply analyzes the video content rather than just giving a surface-level summary, make sure to select "Thinking Mode" before you start. This feature forces the model to reason through the transcript and thoroughly process the information, giving you a much more accurate foundation for your final graphic.

STEP 2: Extract the Video's Core Concepts

Once your environment is ready, find the YouTube video you want to summarize and copy its URL. Paste the link directly into the Gemini chat along with a prompt asking the AI to break down the information.

For example, you can type: “Break down the key ideas and concepts from this YouTube video: [insert your YouTube URL here].” Hit enter, and Gemini will generate a highly structured text summary of the video's most important points.

STEP 3: Launch the Image Generator

Now that you have the key data neatly organized, it is time to translate it into a visual format. Select the "Create Image 🍌" option within your Gemini interface. This activates the platform's advanced image generation engine, which is specifically designed to take the structured concepts from your chat and render them into cohesive, high-quality visual layouts.

STEP 4: Prompt and Generate Your Infographic

With the image generator active, tell the AI exactly how you want the data presented. Type a prompt such as, “Create an infographic that visually summarizes the main concepts from the video.” You can also add specific aesthetic instructions here, like requesting a dark mode theme, corporate branding, or neon colors. Within seconds, the AI will design and output a fully customized infographic that you can download and share instantly.

Apple unveils the $549 AirPods Max 2, with an H2 chip, up to 1.5x more effective ANC, studio-quality audio recording, and camera remote, shipping in early April.

Mistral releases Small 4, its first model to unify reasoning, multimodal, and coding capabilities of its flagship Magistral, Pixtral, and Devstral models.

In an all-hands meeting, Fidji Simo described plans to shift OpenAI's strategy around coding and business users, urging staff to avoid “side quests”.

Apple acquires MotionVFX, a Polish company that develops plugins, visual effects, and motion graphics tools for Final Cut Pro.

⚡ Nemotron 3 Super: Nvidia’s 120B AI for reasoning with super long memory (1M tokens)

🧠 Claude: Anthropic’s AI that can now show interactive diagrams and charts in chat

📚 NotebookLM: New custom styles for infographics to turn info into visuals

🔒 Codex Security: OpenAI Codex security agent that scans code and fixes bugs

THAT’S IT FOR TODAY

Thanks for making it to the end! I put my heart into every email I send, I hope you are enjoying it. Let me know your thoughts so I can make the next one even better!

See you tomorrow :)

- Dr. Alvaro Cintas