Monday AI Business Update: Trump's Nvidia Bet, Google's AI Swarm, and a Security Breach

Nov 16, 2026 11:30 AM
Premium
Advertisement
Monday AI Business Update: Trump's Nvidia Bet, Google's AI Swarm, and a Security Breach
AP Photo/Mark Schiefelbein
VIP

The Trump administration reportedly wants to give Nvidia and more than two dozen other American AI companies a leg up overseas. After Nvidia's recent troubles, that's quite a choice. But Washington has its eyes on China.

Some AI rollouts and announcements take longer to understand than the tech they're presenting. So I'm trying something new here on the VIP pages: the Monday AI business update. Each week I'll attempt to pick a few stories and explain them while cutting through the jargon and text speak.

I figured this needed some sort of rating system, too. Why not? We're calling it the Chip-O-Meter. Five chips means we've had a big week. One chip? Maybe it's time to unplug the computer.

This week's Chip-O-Meter: 4 out of 5 chips. Big moves from players across Capitol Hill and beyond. New useful technology, and one security mess that cost the industry a chip.

Nvidia seems like the logical place to start, especially after the month it's had.

The White House is reportedly preparing to name Nvidia, Amazon, Microsoft, OpenAI, Anthropic, among more than two dozen other companies as American AI "national champions." It's part of President Trump's effort to sell more American technology overseas, where China has been competing for customers and influence.

There's a lot of government assistance money at stake with this new title. Companies could get fast-track status for quicker regulatory approvals to export their computers, chips, and other technology. This could also help them win foreign contracts and get access to federal financing.

Imagine a customer overseas looking for everything needed to run AI. The U.S. wants American companies supplying it, and wants to make buying the whole package easier.

The program has been around for about a year, with Commerce Secretary Howard Lutnick expected to announce the selections. The official list of companies hasn't been released.

Nvidia has had one heck of a month. Jensen Huang is out as CEO after several damaging allegations involving a suspected Chinese intelligence connection and multiple affairs. And plenty of questions remain about where the company's chips have been going. I've covered some of those questions, including where restricted Nvidia technology may have ended up.

Trump has already offered his own suggestion for Huang's replacement. Now his administration reportedly wants Nvidia helping lead America's overseas AI push.

Google had something new to show off Tuesday. It's called Gemini 4 Carbon, and the company says it's the smartest model it's made so far.

Carbon makes diamonds, and Google’s new model Carbon has a pretty tough job. Its goal is to solve tough software problems and provide research where a simple answer just won't cut it. Google says its Argon model is still best suited for simpler, more routine work.

Google also has a little experiment called "swarm." The name sounds like something you'd call an exterminator about, but here's what the company says it does: "To put Carbon’s capabilities to work on larger and complex projects, we’re previewing /swarm in Google Antigravity."

Give the system one instruction, and it can coordinate several AI agents on separate parts of a project. One might write code, another test it, and another look for mistakes. Imagine assigning a website's design, programming, and quality checks to that team.

Google is still testing swarm, but I'd love to see what happens when a client changes their mind halfway through bundling or refactoring a website’s backend. That's when the extra help will count.

Anthropic's newest little tool is a bit closer to what the average office worker needs and would find useful. Claude now has a seat in Google Meet, and it's very powerful. It can follow the conversation, answer questions in the chat, and start putting together plans before the meeting ends. Score one for the meeting that really can now be an email.

You can see where that comes in handy. Somebody asks for a schedule, and Claude works on one while everyone else is still arguing about the dates. We have all been there before, but with this tool, you might even finish the meeting with fewer things left to do and maybe even end the meeting on time!

Right now, only Anthropic’s Team and Enterprise-tier customers can test it out; support for other meeting platforms is coming later. I'd still love to see who gets the notes before inviting Claude into a confidential conversation.

Finally, a gentle reminder that AI companies aren't safe from very old-fashioned security problems.

Cursor, which makes an AI coding tool, says hackers got a key connected to Fireworks AI and used information they accessed to get into other systems. Eventually, malicious versions of Anthropic's Claude Code extension appeared in Cursor's marketplace.

Cursor explained the potential reach of the attack: “At most 5,000 users who requested specific cached versions of the extension on October 29 potentially received the malicious build.”

The bad software sat there for roughly 11 hours. Anyone who received it may have been targeted with an attempt to install a backdoor on their computer. Cursor says the attack has been contained, and Anthropic wasn't responsible for putting the malicious extension in the marketplace.

That leaves us at four chips this Monday. The government wants to help American AI companies win more business, Google's got a new model to test, and Anthropic may have found a way to shorten your to-do list. Cursor's security mess is a good reminder to be careful about what we're plugging into our computers. National Cybersecurity Awareness Month was last month, but you can never be too secure!

That fifth chip can stay in the drawer for now.

Recommended

Trending on RedState Videos