Everyday AI
Posts
NVIDIA's keynote: what was announced and what it means

NVIDIA's keynote: what was announced and what it means

NVIDA’s big announcements, Google drops two huge features in Gemini, and Adobe goes agentic.

Everyday AI
March 18, 2025

👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI

Outsmart The Future

Sup y’all! 👋

We’re switching things up a bit today.

As we’re partnering with NVIDIA to bring y’all live coverage of their GTC conference, we had an amazing podcast and livestream with Aliya Nur Babul from NVIDIA.

(Go watch/listen here.)

And while we normally break down actionable takeaways from each day’s podcast at the bottom of our daily newsletter, we’re calling an audible today.

We had front row seats for NVIDIA CEO Jensen Huang’s keynote, so you KNOW we had to break it down for y’all and give you the 411.

So at the bottom of today’s newsletter, we’re breaking down the 5 biggest headlines from his speech.

Read on…. What happened today will reshape how AI and businesses work.

✌️

Jordan

(Let’s connect on LinkedIn.)

Today in Everyday AI
8 minute read

🎙 Daily Podcast Episode: Startups aren’t JUST for young coders. Mid-career domain experts are shaping the startup industry, and we talked to a leader at NVIDIA to find out what it means. Go listen.

🕵️‍♂️ Fresh Finds: Google goes open source for drug discovery, Meta testing AI comments, and scammers get busted using AI on Amazon.. Read on for Fresh Finds.

🗞 Byte Sized Daily AI News: NVIDA’s big announcements, Google drops two huge features in Gemini, and Adobe goes agentic. Read on for Byte Sized News.

🧠 Leverage AI: NVIDIA CEO Jensen Huang delivered a memorable keynote that is sure to shape the business world for years to come. We sat on the GTC floor and bring you the 5 biggest takeaways. (No one’s talking about #5) Keep reading for that!

↩️ Don’t miss out: Did you miss our last newsletter? We talked about Baidu's new AI model is better and cheaper than Deepseek, Google's new AI solves scientific mystery, NASA's AI controversy and more.. Check it here!

The Next Frontier of Startups: Domain Experts in the Age of AI

Startups are changing quickly.

That means Venture Capital is changing just as fast. 🏃‍♂️‍➡️

And with that breakneck pace has emerged a surprising trend: the skyrocketing value of domain expertise.

A decade ago, you might expect a thriving startup to consist of a bunch of recent Computer Science grads.

Today? Doesn’t look quite as standardized.

Mid-career professionals, and the domain expertise they hold, may be the secret sauce in today’s playing field of AI everywhere.

Also on the pod today:

Venture Capital’s next power moves 💪
Why domain experts are important in startups ▶️
How VCs are looking at startups differently 👀

It’ll be worth your 32 minutes:

Listen on our site:

Click to listen

Subscribe and listen on your favorite podcast platform

Listen on:

Spotify | Apple Podcasts |
Google Podcasts | Amazon Music |

Here’s our favorite AI finds from across the web:

New AI Tool Spotlight – Aha is an AI-powered influencer marketing team, Twos is an AI-powered list and note-taking app, and the Shorty app helps you create viral videos with AI.

AI and Safety – The Trump FTC erased 300+ blogs on AI, privacy, and consumer protection—raising compliance concerns and benefitting big tech.

Social Media and AI – Meta is reportedly testing AI generated comments. What could go wrong?

AI in Business – The FTC says scammers on Amazon used AI to steal millions from investors.

GPUs – Amazon is offering its Trainium AI chip at a massive discount—just 25% of Nvidia's H100 price—aiming to shake up the AI chip market

AI Models – Google is going after drug discovery with open AI models

Agentic AI — Inside 10 of Adobe’s new agentic workflows.

1. Nvidia CEO Unveils Bold AI Chip Strategy Amid Market Shifts 📊

At Nvidia's GTC keynote in San Jose, CEO Jensen Huang laid out the company's ambitious plans to dominate the evolving AI landscape, revealing new chips—including the Blackwell Ultra, slated for late 2025—and a powerful DGX Workstation PC.

As AI moves from training models to real-time responses, Huang emphasized Nvidia's unique ability to handle both tasks efficiently, despite investor skepticism and a 3.4% stock dip. Huang also teased the next-gen Rubin and Feynman chips, aiming to maintain Nvidia’s edge against rising competitors.

Scroll down for our GTC takes, as we partnered with NVIDIA to bring you closer to the action.

2. Adobe Unveils Agentic AI Revolution at Summit 2025 🗣️

Adobe has announced new AI advancements at its annual Adobe Summit, introducing "Agentic AI" to redefine marketing and customer experiences. According to Adobe, its new suite of purpose-built AI agents—such as Content Production Agent and Audience Agent—will streamline creative workflows, scale personalization, and optimize customer journey orchestration.

The launch also includes Adobe Brand Concierge, an advanced conversational AI tool designed to deliver tailored experiences for B2B and B2C businesses, leveraging real-time data insights.

3. Elon Musk’s xAI Acquires Hotshot in AI Video Race 🎥

Elon Musk’s xAI has snapped up Hotshot, a New York-based startup specializing in AI-powered video generation tools, marking a major move in the race to dominate AI-driven video creation.

Hotshot, known for its text-to-video models like Hotshot-XL and Hotshot Act One, will cease new video creation by March 14 but promises exciting integrations into xAI’s platform, including a potential "Grok Video" feature. The acquisition signals xAI’s ambition to rival OpenAI, Google, and Adobe in an AI video market projected to hit $2.5 billion by 2032.

4. Google Gemini Unveils Game-Changing Tools ✨

Google Gemini has rolled out two major updates—Canvas and Audio Overviews—that promise to transform how users interact with code and content. Canvas offers a dynamic space for real-time document editing and rapid prototyping, making it a powerful tool for coders and creators alike.

Meanwhile, Audio Overviews brings podcast-style summaries of documents and research reports to the app, boosting accessibility for busy professionals. According to Google’s announcement, these features aim to supercharge productivity and are now available globally for Gemini subscribers.

5. Meta’s Llama Hits 1 Billion Downloads 🦙

Meta’s open-source AI model, Llama, has reached 1 billion downloads, a sharp rise from 650 million just four months ago, signaling its growing dominance in generative AI.

According to reports, major players like Spotify and DoorDash are using Llama despite licensing limits, while Meta faces copyright lawsuits and privacy concerns in Europe. With $80 billion in planned AI investments for 2025, Meta is pushing updates like reasoning capabilities and multimodal features to compete with rivals.

🦾How You Can Leverage:

How would Jensen Huang respond?

Yeah, the eyes of the tech world were lasered in on the NVIDIA CEO as the global juggernaut attempts to crawl its way out of a short, yet measurable, slide.

Like much of Big Tech, NVIDIA’s had a rough go the past month or two.

The U.S. stock market’s been sliding, NVIDIA included.

And the recent spate of tiny State of the Art AI models that don’t need a bajillion of NVIDIA’s GPUs to succeed has thrust the AI kingmaker’s longer-term role into a bit of question.

And — for better or worse — analysts look at NVIDIA as the gauge for the tech industry at large.

And the AI and tech industry is what’s driving the U.S. stock market right now.

So, when the NVIDIA CEO, donning his signature black leather jacket and famouslyn watch-free wrist, took the stage of this keynote, the consequences were much higher than a normal tech conference.

If the announcements are meh, the industry could slide even more.

If Jensen and crew crush, the brick wall crumbles. (At least temporarily.)

We were on the floor, bringing you, dear Everyday AI reader, closer to the action the most.

Here’s the 5 most important announcements from Jensen’s keynote, and what they ACTUALLY mean.

(You won’t see these takes elsewhere.)

Let’s get it.

1. DGX Reborn: From DIGITS to a Next-Gen Beast

The new DGX platform replaces DIGITS with DGX Spark and DGX Station.

What is it?

It’s a legit AI supercomputer for $3,000.

DGX Spark features 20 CPU cores and 128 GB of memory. It delivers 1 petaflop of performance. This update pushes AI computing to new heights.

Keep an eye out on an upcoming podcast with a senior leader at Dell on what DGX might mean for enterprise.

2. GPU Roadmap: A Blueprint for AI Domination

For the dorks!

NVIDIA laid out their GPU roadmap for the next few years, which includes the new Blackwell Ultra NVL72, Vera Rubin NVL144 and Vera Rubin NVL576 Ultra, due in the second half of 2027.

NVIDIA also announced its next-gen Feynman GPU, set to launch in 2028 with advanced HBM memory and upgrades like the Vera CPU, NVSwitch NVL-Next, and Spectrum7, succeeding the Rubin and Blackwell GPU families

In simpler terms?

NVIDIA’s GPUs are getting WAY more powerful and will likely be driving a good chunk of the AI we all use for years to come.

3. GROOT N1: Open-Source and Unstoppable

Nvidia’s Groot N1 is NVIDIA new robotics AI model and they open-sourced that dang thing at the end of Jensen’s keynote.

The new AI model is designed to give humanoid robots both a brain for deep thinking and quick reflexes

By releasing simulation frameworks and synthetic data blueprints, NVIDIA is practically saying, “Here’s the cheat sheet to building your own robot overlord.” It’s a bold move, but hey, if it all goes south, at least humanity has the source code to debug the mess?

4. AI’s Inflection Point: The Compute Explosion

There’s obviously been a lot of hubbub about the future of GPUs.

Models are getting smaller and more capable, which, in theory, means big tech may need fewer NVIDIA GPUs to train their frontier models.

Not so fast, said Jensen.

He used a factor of 100 to talk about the increased demand for GPUs.

He said AI now demands 10X more tokens at 10X faster speeds. This results in a 100X surge in compute demand. He called this the industry’s AI Inflection point.

Our take?

For every few billion NVIDIA might lose from model makers needing less compute, they’ll probably make that AND THEN SOME as edge AI increases demand like crazy for personal AI super computers like DGX Spark or DGX Station.

5. Deepseek’s Silent Nod?

We definitely didn’t see this coming.

And why is no one talking about this?

After the DeepSeek $5.5 million training fiasco sent NVIDIA’s stock in a swirl, we were surprised to see the NVIDIA CEO kinda put their R1 reasoning model in a positive light.

In a demo of how reasoning models gobble up more tokens and compute vs. traditional LLMs, Jensen’s slide pitted a traditional Llama model against Deepseek’s R1 and showed that R1 delivered a better and more nuanced response.

Was it for explanatory purposes?

Sure.

Still…. This one’s a head scratcher.

⌚

Numbers to watch

$ 20 million

Halliday’s $20M funding aims to make AI agents safely operate on blockchain with 'immutable guardrails

Reply

or to participate.