🧠 NVIDIA unveils next-gen AI chips

Also: Gemini receives Canvas editor and AI podcasts

Author

Evolving AI
March 19, 2025

Welcome, AI enthusiasts

NVIDIA officially announced the Blackwell Ultra B300 data center GPU today during CEO Jensen Huang's keynote at GTC 2025 in San Jose. NVIDIA says it's "built for the age of reasoning," referencing more sophisticated AI LLMs like DeepSeek R1 which go beyond simple data recall. Let’s dive in!

In today’s insights:

  • NVIDIA unveils Blackwell Ultra AI chip for 'age of AI reasoning'

  • Google adds Canvas editor and AI podcasts to Gemini chatbot

  • Amazon employees sound alarm on DeepSeek privacy risks

Read time: 4 minutes

LATEST DEVELOPMENTS

Source: Yifan Yu

Evolving AI: NVIDIA announced the Blackwell Ultra chip, boosting AI reasoning power.

Key Points:

  • Blackwell Ultra is 1.5 times faster than its predecessor.

  • Nvidia claims a 50x increase in revenue potential over its Hopper chip.

  • Designed specifically for advanced AI reasoning models.

Details:

Nvidia CEO Jensen Huang introduced the Blackwell Ultra AI chip at the annual GTC event. The chip is tailored for the next phase of AI, known as AI reasoning, where systems think similarly to humans. Alongside the Blackwell Ultra, Nvidia launched the GB300 superchip, combining two Blackwell Ultras with its Grace CPU. These powerful new systems are already drawing interest from big tech firms like Amazon, Google, Microsoft, and Meta.

Why It Matters:

With the Blackwell Ultra chip, Nvidia is positioning itself for the next big shift in AI — reasoning. AI models deliver much better results when they have extra seconds to "think," but that requires significantly more computing power. Nvidia’s latest chip meets this demand, placing the company at the center of the AI reasoning era.

Evolving AI: Building an AI business? Becoming ISO 42001 compliant shows your customers that you are taking the necessary steps to ensure responsible usage and development of AI.

With Vanta, achieving compliance can be done in a fraction of the time. 95% of required document templates are pre-built, accelerating the work and helping you build trust.

Start with this free checklist — it breaks down the compliance process step by step.

Download the checklist now →

Source: Google

Evolving AI: Google's Gemini AI assistant now features Canvas, a real-time workspace, and Audio Overviews for turning text into podcasts.

Key Points:

  • Canvas lets users work alongside AI to create documents and code.

  • Audio Overviews turn written content into conversational podcasts.

  • Both features follow similar additions made by ChatGPT last year.

Details:

Google just launched two significant updates to Gemini: Canvas and Audio Overviews. Canvas provides an interactive workspace where users can create documents or code collaboratively with AI. It shows edits live, allows chat-based discussions, and quickly previews HTML and React code. Users can also export completed work to Google Docs. Audio Overviews transform documents or research into engaging AI-hosted podcast conversations, initially available only in English. Both features are accessible globally for Gemini subscribers.

Why It Matters:

Google has quietly launched update after update in recent weeks, clearly finding its stride again. Rather than waiting, they're now rapidly delivering new features at a pace we haven't seen from them in a while.

Source: Ava Horton/BI

Evolving AI: Amazon employees warn about DeepSeek's data privacy issues while promoting Amazon’s own AI solutions.

Key Points:

  • Amazon rapidly integrated DeepSeek into Bedrock due to high demand.

  • Employees are now highlighting privacy risks of using DeepSeek models.

  • Amazon positions its Nova AI as a safer, faster alternative.

Details:

DeepSeek quickly became popular in January, prompting Amazon to add it to their AI platform, Bedrock. Internal documents reveal Amazon is now cautioning employees against sharing sensitive information with DeepSeek due to privacy risks, including potential data storage in China. Employees are encouraged to promote Amazon's Nova models, which are presented as faster, safer, and capable of handling images and video, unlike DeepSeek’s text-focused models. Amazon continues developing its own reasoning model to directly compete with DeepSeek, showing how quickly tech giants must adapt in the AI race.

Why It Matters:

We've heard about DeepSeek's privacy issues ever since it went viral in January. As more users adopt the model, understanding these risks becomes crucial. Sharing sensitive data with AI models that store information overseas can expose companies and individuals to significant security threats.

 👀 Click on the image you think is real

QUICK HITS

🖼️ US appeals court rejects copyrights for AI-generated art lacking ‘human’ creator.

⚖️ Arizona Supreme Court taps AI avatars to make the judicial system more publicly accessible.

🤖 Japan lacks workers to care for the elderly. This company is using AI to help.

🚀 Mistral AI drops new open-source model that outperforms GPT-4o Mini with fraction of parameters.

🎙️ Amazon’s AI-enhanced Alexa assistant is going to need all your voice recordings, and there’s nothing you can do about it.

🦪 Marin County oyster business using AI to help run company.

📈 Trending AI Tools

  • 📝 OneNode - An AI-powered note-taking app (link)

  • 📧 Breezemail - An AI-powered tool for organizing your email inbox (link)

  • 📄 aiPDF - An AI-powered tool for summarizing documents (link)

  • 🎙️ TranscribeMe - A tool to transcribe audio on WhatsApp for free (link)

  • 📑 Canyon - An all-in-one platform for job-seekers that helps you perfect your resume and track your applications (link)

  • 🌐 NetMind Power - A decentralized ML and AI platform (link)

Reply

or to participate.