Support Us & Upgrade to the Operator Edition

Today Apple bet the AI era on a smart display, Nvidia wrote the UK a $250M safety check paid in its own chips, and an OpenAI engineer allegedly hid a bug in toilet paper. Normal Wednesday.

This week, we'll deep dive into who is actually building the AI future, and who is just paying for good press.

Connect your customer data, teams, and tools without the hassle of complex integrations or lengthy setups. One easy platform gives marketing, sales, and service teams a unified customer view and the tools to turn it into growth.

Why HubSpot and what's new
• Generate leads and automate marketing with Marketing Hub
• Build your pipeline and close more deals with Sales Hub
• Scale customer support and drive retention with Service Hub
• Keep customer data clean, connected, and actionable with one, unified platform

Join 306,000+ in over 135 countries using HubSpot to grow their businesses.

See what a more connected approach to growth can do for you and your team. Get setup quickly and start checking off your hardest tasks.

Get started free

Google's team spent 1,000 hours testing Gemini Argon, and the big unlock isn't input size. It's that Argon can output up to 1 million tokens in a single run. That means it can hold an entire project in its head and finish it without you babysitting every step.

In the demos, it rewrote 1960s Fortran into Rust that runs 20% faster, turned photos of a messy desk into a playable 3D racing game, and hit a build error mid-project that it fixed on its own without asking for help.

Here's how to use that today:

💡 Pro tip: Copy the NASA demo. Grab any public dataset in your industry, feed Argon the raw files, and ask for an interactive visualization. It's an afternoon project that looks like a month of work in your next pitch.

Watch all 10 demos here

Want the version of this that actually runs in your business?

The 30-Second Play gives you the move. The Operator Brief gives you the full system: prompt libraries, API cost math, and the workflows our subscribers are using to replace entire freelance line items.

This week: Learn how to use Claude inside of Google Workspace

👉 Upgrade to The Operator Brief

📉 Dell shares tumbled roughly 8% after Barclays downgraded the stock, warning that the hardware rally has run too far and enterprises are still figuring out what to actually spend on. Dell’s share drop is the market agreeing, at least for today.

🏭 Nvidia says chipmakers like Lam Research, KLA, and Synopsys are using its new AI tools to build "AI-defined fabs," with Lam seeing up to 30x faster chipmaking simulations. AI is now designing the chips that run AI. The loop is closing faster than most people realize, and we’re handing over the keys to more of the infrastructure with each iteration.

⚠️ Demis Hassabis called recent AI agent hacks "early warning shots" and warned bigger incidents could follow in the next couple of years if the industry doesn't adopt nuclear-grade security. Bill Gates put it this way: "Just as we regulate planes, cars, and medicines, we need safeguards that help us realize AI's incredible potential while reducing the risks."

🕵️ An OpenAI engineer was arrested by the FBI after allegedly hiding a recording device in toilet paper inside an executive conference room, where the executives talk about the heavy stuff like government integration. And prosecutors say he researched how to build it using his wife's ChatGPT account. OpenAI used its own models to catch him spying on the table. That’s a sentence that would have sounded like fiction a few years ago.

Apple just unveiled HomeView, a $299 smart display powered by an AI-upgraded Siri, and it's the clearest signal yet of how Apple plans to play the AI hardware race.

While OpenAI and Meta chase brand-new gadgets, Apple went the other way: make the thing you already know actually work.

Here's what launched:

The pitch is simple. The smart home has failed for a decade because it turned everyone into an unpaid IT department. Apple is betting AI can make devices set themselves up, diagnose their own problems, and coordinate with each other.

Here's the uncomfortable part: nobody is buying this category. US smart speaker sales dropped 9% in 2024 and another 7% in 2025. And the most capable AI assistant you own already lives in your pocket.

Meanwhile, the competition is swinging bigger. OpenAI's first device is expected soon, and Meta just teased Charm, a pocket companion for its Muse assistant. And today, OpenAI claimed in court that a former employee took 100+ confidential files about an unreleased device to Apple and pitched them to senior leadership. Apple says it found no evidence of that.

So the company being accused of peeking at OpenAI's hardware homework... launched a kitchen screen.

🔮 Prediction: HomeView isn't the product. The product is Siri knowing who's in the room.

Think about what "recognizes different household members" really means. It's an identity layer for your home. Once Siri knows that it's you asking (not your kid, not your partner), it can safely touch your calendar, your messages, your payments, your health data. That's the piece every home assistant has been missing, and it's exactly where Apple's privacy reputation is a real advantage over Amazon and Google.

My bet: HomeView sells modestly, maybe a few million units, and the press calls it a flop. But within 18 months, Apple turns "Home" into an agent platform, where the screen is just one window into a household AI that follows you from iPhone to car to kitchen. The display is the Trojan horse. The moat is 1B+ iPhone users who'll set it up in 30 seconds because it just appears in their Home app.

The risk? If OpenAI's device lands first and feels genuinely new, HomeView becomes the BlackBerry keyboard of AI hardware: polished, familiar, and on the wrong side of history.

💬 Would you put an AI screen that recognizes your family in your kitchen? Hit reply with a yes or no.

Read the full story here

Nvidia committed $250 million to UK AI safety research over five years, focused on the scariest problem in AI right now: agents that go rogue.

It landed the same day DeepMind's Demis Hassabis called recent AI hacks "early warning shots," so the timing is anything but random.

The breakdown, per Nvidia's announcement:

Jensen Huang's line was that capability and safety have to move forward together. UK AI Minister Kanishka Narayan called it a vote of confidence in British safety research.

Here's the asterisk: read the fine print, and this is a donation of Nvidia hardware, through Nvidia partners, running Nvidia software. Safety researchers who build their tools on your stack become your customers for life.

And scale matters. This same week, Nvidia announced $10 billion into four French AI startups. That's 40x the safety pledge, and 200x the yearly safety budget. For a company worth $5.5 trillion, $50M a year is a rounding error on a rounding error.

🔮 Prediction: "Safety compute" is about to become the new currency of AI diplomacy.

Here's the logic. The UK has no frontier lab of its own, but it badly wants to be the world's AI referee, mostly through its AI Security Institute. Nvidia wants governments to see it as a responsible partner, not just the company selling shovels in a gold rush. A compute pledge solves both problems at almost no real cost, because unused GPU time is cheap for Nvidia and priceless for a university lab.

Expect Google, Microsoft and Amazon to copy this playbook with the EU, Japan and the Gulf within six months. The pledge amounts will sound big; the actual cost to the giver will be small.

The part to watch isn't the money. It's the benchmarks. Whoever writes the tests that decide whether an AI agent is "safe" effectively writes the rules. If those benchmarks get built on OpenShell and Sentry, Nvidia won't just sell the chips that run AI agents. It'll own the yardstick that certifies them. Hassabis is already calling for a risk-based certification system. Nvidia just put its tools in the room where that system gets designed.

Good for safety? Probably. Brilliant for Nvidia? Definitely.

💬 Genuine safety investment or the smartest marketing spend in tech? Reply and tell me which side you're on.

Read the full story here

Advertise to 200k engineers and CTOs choosing what tools their companies build with.

👉️ Learn how to sponsor this newsletter

Would love to hear your thoughts! Send me your thoughts by replying to this email (yes, I read them all :)

Until our next AI rendezvous,

Anthony | Founder of Uncover AI