Good morning. Hundreds of actors, writers and musicians have published an open letter in response to Google and OpenAI’s policy recommendations, which called for the degradation of copyright protections.

“We firmly believe that America’s global AI leadership must not come at the expense of our essential creative industries,” the letter reads.

— Ian Krietzberg, Editor-in-Chief, The Deep View

In today’s newsletter:

• 🛰️ AI for Good: Google’s wildfire satellite

• 🌎 Europol warns AI is changing the DNA of organized crime 

• 👁️‍🗨️ All the details from Nvidia’s GTC — enough to make a difference?

🎙️ Podcast: The climate impact of AI

In the latest episode of our weekly podcast, The Deep View: Conversations, I sat down with Shaolei Ren, a researcher at the University of California, Riverside, who has extensively studied the water consumption, energy use and public health cost of the data centers that power generative AI.

These data centers are at the core of the climate impact and imbalance of today’s GenAI.

Check it out:

AI for Good: Google’s wildfire satellite

We’ve talked a lot about the ways machine learning algorithms are being leveraged to advance the early detection of wildfires. But these efforts mature slowly, and were unable to prevent the recent, and devastating, L.A. wildfires from getting out of control. 

Still, researchers and companies are pushing hard to make early wildfire detection a reality. 

What happened: This week, Google launched its first wildfire detection satellite, the first entry in a wildfire-detection constellation masterminded by the Earth Fire Alliance, a nonprofit launched by Muon Space and Google.org. 

• The problem with current wildfire detection is that satellite imagery comes either too late (on 12-hour cycles), or the images themselves are simply too coarse to identify wildfires at controllable stages. 

• This project — christened FireSat — aims to leverage machine learning algorithms to remedy that problem; the satellites come equipped with infrared sensors and are designed to autonomously scan areas roughly 5x5 meters in size for signs of small — and therefore containable — wildfires. Part of this scan involves a historical comparison of that region, to determine if there are any flame-induced changes. 

Why it matters: Once fully operational, the satellite constellation will act both as a global emergency response tool and a tool for research, enabling better models of wildfire behavior, which, in turn, could enable better responses on the ground. 

Tidy up your finances with BILL. Get a Roomba.

This spring, tidy up your finances—and your floors. Score a free Roomba when you check out BILL & Expense, our knockout combo of the BILL Divvy Corporate Card and free expense management software.

With BILL, you can:

• Control messy, unnecessary spend with customizable limits

• Streamline reconciliations and reporting

• Ditch the paper clutter by tracking spend in real time

• Clear a path for growth with scalable credit lines from $1,000-5M¹

Take a 30-minute demo to see how BILL can refresh your finances—then sit back while your Roomba does the same for your floors.²

Europol warns AI is changing the ‘DNA’ of organized crime 

We’ve been hearing about AI-accelerated fraud almost from the moment ChatGPT launched in November of 2022. 

It comes in a lot of forms, ranging from deepfaked audio to deepfaked imagery and text, often designed in some way to trick you or your loved ones into giving money to fraudsters. And according to Europol, the law enforcement agency of the European Union, the tech isn’t just accelerating small-time fraudsters; it’s enabling a vast network of organized crime. 

The details: In a new report, Europol said that artificial intelligence is vastly expanding “the speed, scale and sophistication of organized crime, creating an even more complex and rapidly evolving threat landscape for law enforcement.”

• Using large language models (LLMs) and generative AI — which has been designed to be highly accessible and user-friendly — criminal organizations are able to easily overcome language barriers, target global victims “with precision,” create sophisticated malware and produce child sexual abuse imagery. 

• Armed thusly, they can extort, blackmail, impersonate, discredit and deceive victims. Massive phishing campaigns can now be automated; large-scale cyber attacks can be executed with ease and precision; and the potential rise of even more autonomous systems “could pave the way for entirely AI-controlled criminal networks, marking a new era in organized crime.” 

This is a level of autonomy that just about every developer is working hard to achieve. 

“The same qualities that make AI revolutionary — accessibility, adaptability and sophistication — also make it a powerful tool for criminal networks," according to the report. 

The landscape: In late February, Europol arrested 25 people for the illegal production and distribution of AI-generated child sexual abuse material, one of the first cases of its kind. 

“The very DNA of organized crime is changing,” Catherine De Bolle, Europol’s executive director, said. Tackling it requires a global collaboration that prioritizes victim protection, leveraging those same new technologies being used by criminals to fight the source of the crime at its roots. 

Why CIOs are scaling with business observability

Technology teams aren't just responsible for performance metrics anymore. CIOs are now turning to solutions offering a centralized view of organizations' telemetry enriched with business context, also known as business observability.

In this eBook, you'll learn about business observability best practices through real world use cases from Datadog customers like Wayfair, and get an actionable framework for implementing your own strategy.

Download the eBook now

• Language learning: Start speaking a new language in just 3 weeks with Babbel. With expert-crafted lessons and immersive content, you'll be ready for your next adventure in no time. Get 60% off while their Spring Sale lasts - start today! *

• Monetization: Most are harnessing the power of AI these days but what if you could get AI to work for you?

  • Learn how to monetize AI by registering for a free web class from an award-winning AI company. *

• Alphabet to buy Wiz for $32 billion in its biggest deal to boost cloud security (Reuters).

• AI search is starting to kill Google’s ‘ten blue links’ (The Verge).

• ‘New technology like this is fantastic and it’s important to embrace it’ says 93-year-old who takes a trip in self-driving Nissan (The Irish Times).

• AI scammers on Amazon duped investors out of millions with ‘passive income’ scheme, FTC alleges (CNBC).

• Tesla looks to India at a moment of crisis (Rest of World).

All the details from Nvidia’s GTC — enough to make a difference? 

I said yesterday that Nvidia $NVDA ( ▲ 2.82% ) had a lot to prove at GTC to bring investors firmly back on board. 

And when CEO Jensen Huang took the stage yesterday, he did so with a lot of passion and a lot of excitement. We’ve talked about the importance of optimism in AI, and Huang does a really good job at selling optimism for the strength of the business and the possibilities of the future. 

But — and I’ve been saying this a lot lately, when it comes to Nvidia — it doesn’t seem like it was enough. The stock retreated modestly in the lead-up to Huang’s keynote address, then continued to slide during it, falling as much as 3%. 

Of course, the macroeconomic environment — with Fed Chair Jerome Powell speaking later today — can’t be ignored: all the major indices fell Tuesday, led by declines across the tech sector: $MSFT ( ▲ 0.61% ); $GOOG ( ▲ 3.01% ); $META ( ▲ 1.48% ). But, at least so far, GTC hasn’t sparked a surge in the opposite direction. 

Some key takeaways: Huang talked for more than two hours, without, as he noted, the aid of a teleprompter. This is about on par with the length of his 2024 keynote, but it’s roughly twice as long as his 2023 keynote. He made a bunch of announcements, many of which were anticipated. But before we get into the announcements, he did address demand: and the story he’s painting is that there is a level of demand for his company’s chips that just won’t quit. 

• In its peak year, Nvidia shipped 1.3 million of its older Hopper GPUs to the four major cloud service providers, Amazon, Microsoft, Oracle and Google. Those same four companies have purchased a total of 3.6 million Blackwell chips for the year so far, where each Blackwell chip contains two GPUs. “AI,” according to Huang, “is going through an inflection point.” 

• He then doubled down on what he said during Nvidia’s last earnings call, that the rise of ‘reasoning’ models is a very positive sign for Nvidia’s business. The idea, according to Huang, is that the reasoning process — which leverages Chain of Thought (CoT) technology — causes a massive increase in inference-time compute, meaning more chips are needed even if a model is more efficient to train. 

A demo of DeepSeek’s reasoning-enabled R1 model Huang showed on stage used 150 times as much compute and generated 20 times more tokens compared to Llama 3, a non-reasoning model. For Nvidia, more compute means more revenue. 

Huang stated that Nvidia is “now in full production of Blackwell,” adding that, in the second half of the year, “we’ll easily transition into the upgrade … the Blackwell Ultra.” 

“Customer demand is incredible,” Huang added. 

The roadmap: He also laid out Nvidia’s chip roadmap for the next several years. The first of these is the Vera Rubin chip, a combination of a custom-built CPU (Vera) and an advanced GPU (Rubin) that Huang said will start shipping in the latter half of 2026. 

The chip is named for the late astronomer Vera Rubin. 

And the upgraded version of the Vera Rubin chip, the Vera Rubin Ultra, will begin shipping during the second half of 2027. 

• Amid all this, Huang announced that Nvidia is officially partnering with General Motors to “build their future” self-driving fleet. The partnership will operate on a massive scale, with Nvidia providing the tech to enable AI-assisted manufacturing and AI-assisted work in the enterprise, in addition to enabling GM’s self-driving software. 

• GM said that its Super Cruise driver assistance software will be available on 750,000 miles of North American roads by the end of this year. 

We also saw announcements of a new photonics switch, to enable greater GPU integration within data centers, and an open-source piece of software called Dynamo, which is intended to accelerate and scale reasoning models within the data center. 

And … robotics: It wouldn’t be a GTC keynote without a live demo. To help him unveil a new robotics partnership between Nvidia, Disney Research and Google DeepMind, Huang was joined on stage by a Star Wars-esque robot (should we call this one a droid?) that whistled in a way most reminiscent of its forefather, R2-D2. 

In addition to this, Nvidia open-sourced Groot, its family of models designed specifically for robotics. 

During the talk, shares of Nvidia clawed back some ground, but it was a temporary reprieve. The stock closed the day down nearly 3.5%. 

Huang is very good. He makes it hard not to get excited by progress and potential. 

The trouble (sorry!) is that the future he’s trying to achieve is not one that can be reached on a straight, even line. He said during the keynote that the world has totally misunderstood scaling laws. The reality is that remains to be seen. It remains unclear if scaling current systems up will lead to breakthroughs that will enable robotics and self-driving cars to operate reliably and at scale. 

There’s no evidence that it will. And the challenges there are rooted in physics fundamentals, making them exceedingly difficult to overcome. 

And it’s good to hear that demand remains strong, but all that says is that the AI ecosystem is still being propped up. Nvidia isn’t representative of consumer demand or adoption. Nvidia’s performance is representative of capex spend from Big Tech, a multi-billion-dollar initiative that is pushing GenAI whether consumers want it or not. 

I don’t think Huang said anything explosively new or unexpected. It was a confirmation and reiteration of points he’s been making for years, now. 

And in the midst of all that, Wall Street, for one, was unable to find evidence — at least, in the near term — that Nvidia’s fundamentals, or the fundamentals of the AI revolution, are enough to offset the economic impact of the trade war. 

The additional problem here that I’ve been thinking about is that if things keep going at the same rate that we’ve seen so far, capex from the mega-caps is going to slow down next year, something that will certainly impact Nvidia’s demand numbers. The only way this won’t happen is if Big Tech’s AI investments suddenly start unlocking massive revenue returns, a phenomenon that has yet to occur.

🤔 Your thought process:

Selected Image 1 (Left):

• “The weird white glacier in the lake looks like it dropped in from another planet.”

💭 A poll before you go

Thanks for reading today’s edition of The Deep View!

We’ll see you in the next one.

Here’s your view on GTC:

37% of you don’t think GTC can help … things are too far gone.

But 30% of you think GTC will be a game-changer.

I guess we’ll find out.

Something else:

• “Not in the current world economic climate. The world is marking time essentially until the effect of tariffs become clearer. No company will stick it's neck out for the next 6-8 months.”

If you want to get in front of an audience of 450,000+ developers, business leaders and tech enthusiasts, get in touch with us here.

BILL’s Disclaimer: The BILL Divvy Card issued by Cross River Bank, Member FDIC, and is not a deposit product.

¹Credit lines and the advertised range are not guaranteed and will be determined upon application approval.

²Terms and conditions apply: see offer page for more details.