Instagram | Sponsor this newsletter

My fellow AI explorers

Big week in the war between Washington and the labs. The same day Microsoft was calling a laptop a "dream machine" and Scorsese was quietly storyboarding with a German image model, Trump signed an order asking the frontier labs to let the government peek under the hood before they ship. Voluntarily. Wink.

In today’s edition:

• 🏛️ Trump's new AI order: hand us your models (please)

• 🎬 Scorsese embraces AI, and film Twitter loses it

• 💻 Microsoft's Build keynote and Uber’s $1,500 Claude Code leash

Sponsored by BTQ Technologies, Inc

Washington Just Bet $2 Billion on Q-Day. The Trillion-Dollar AI Labs Don't Have a Plan.

The U.S. Commerce Department just signed nine letters of intent worth $2.013 billion with quantum-computing companies — and took equity in every one. That's the loudest signal yet that "Q-Day," the day a quantum computer cracks today's encryption, is a real deadline — not a thesis. That encryption? It’s what secures the internet, your bank account, your phone, and government communications.

• 💵 IBM leads with $1B of the Commerce Dept investment, plus $1B of its own, for a domestic quantum foundry.

So, if the government is betting billions on Q-Day’s arrival, shouldn’t some of the biggest tech companies — the AI labs — be thinking about this threat?

• 🍎 Apple shipped post-quantum encryption to iMessage in 2024. Cloudflare protects 43% of all web traffic with it. The frontier AI labs? Still zero.

• 🎯 OpenAI, Anthropic, and xAI are worth over a trillion combined. Published post-quantum roadmaps? Zero.

Every model weight and user prompt sits behind encryption Q-Day puts on a clock. AI built the target. Almost nobody's building the lock. That's where BTQ Technologies (NASDAQ: BTQ) comes in, building the defense layer while the trillion-dollar labs ship none. BTQ is already developing what protects critical infrastructure: a post-quantum chip (QCIM), self-destructing signatures (OSS), and the first quantum-safe Bitcoin testnet — already past 200K blocks.

🔮 If quantum is the artillery, BTQ is the armor. Washington just told the market the artillery is coming.

🏛️ THE MAIN EVENT

🎬 Trump Wants a 30-Day Look at Your Frontier Model. Before You Ship It.

President Trump signed an executive order on Tuesday asking AI companies to hand their most advanced models to the federal government for national security vetting before public release. The catch that makes this one interesting: participation is voluntary.

Back on May 21, Trump postponed a White House ceremony and refused to sign a bill, reportedly worried it would dull America's edge in AI, only to quietly sign one now that’s not quite as tough on the AI industry.

Here's what this order actually does:

• 🔍 Creates a framework to vet the national security risks of the most advanced AI systems for up to a month before release.

• 🧪 Tasks DHS and other agencies with building a benchmark to measure the "advanced cyber capabilities" of AI models.

• 🤝 Asks frontier labs to "voluntarily" give the government early access to whatever gets flagged as frontier.

The backstory is the whole story.

The original draft reportedly asked companies for a 90-day pre-release testing window. Then Elon Musk and Mark Zuckerberg allegedly got on the phone to lobby against it. What shipped is a softer, voluntary, 30-day version. The labs pushed; the order bent.

And the timing is not subtle. Anthropic just confidentially filed for an IPO, OpenAI is gearing up for its own, and SpaceX (which owns xAI) could hit the public markets within days at a trillion-plus valuation. The most powerful labs in the world are about to face public-market scrutiny, and the government just asked, oh so nicely, to see the goods first.

🔮 Prediction: "Voluntary" is the tell. This is the federal government finding out it has almost no hard leverage over the frontier labs, and just how submissive it has become. The government is testing how much it can get by asking politely while everyone's pre-IPO and wants to look like a good citizen. The moment these IPOs close, the incentive to play nice dissipates. After that, watch for every lab to frame government access as a material business risk.

AI Movies

🎬 Scorsese Just Went AI. Film Twitter Is Not Okay.

Martin Scorsese, the man who said Marvel movies aren't cinema, signed on as partner and advisor to AI image startup Black Forest Labs, the team behind the FLUX model. He's been quietly advising for about a year. Now it's public, and the backlash is loud.

• 🖼️ He used FLUX to storyboard his next film, What Happens at Night, starring DiCaprio and Jennifer Lawrence.

• 🎨 His pitch: after 70 years drawing his own storyboards, this is just a faster way to show his crew the pictures in his head.

• 🌲 Black Forest Labs is a 70-person shop based not in San Francisco but in Freiburg, Germany.

The reaction was immediate and brutal: fans feeling let down that one of the last great holdouts joined the party. But Scorsese's been telegraphing this for a while; back in 2023, he said he hoped new tech would push cinema into a new form.

But here’s the nuance everyone's skipping: he's using it for pre-production concept art, not to generate the actual film. He’s found a faster way to show his crew the pictures in his head.

🔮 Prediction: When the director who declared what is and isn't cinema starts storyboarding with AI, the "real artists don't touch this" line stops holding. Expect the next 12 months of Hollywood AI adoption to hide in pre-production, storyboards, concept art, previz, where nobody can see it in the final cut.

30-Second AI Play

💻 Microsoft Built a "Dream Machine." Uber Built a $1,500 Leash. Here's How to Avoid Uber's Mistake.

Two stories collided this week. At Build 2026, Nadella unveiled the Surface RTX Spark Dev Box, an Nvidia-chip-loaded machine he called a "dream machine" (he's on the waitlist too), built to run agentic tools natively on Windows. Meanwhile, Uber capped employees at $1,500/month per AI coding tool after burning its entire 2026 AI budget in four months.

The lesson is the same: agentic AI scales spend faster than you think. Here's how to stay on the dream-machine side:

1. Pick one recurring task first. Not "use AI everywhere." One repeatable workflow (code review, support triage, and research summaries).

2. Add a human review step. Uber's COO admitted he can't yet draw a line between rising Claude Code usage and shipped features. Measure output, not token burn.

3. Track saved hours vs. cost, per person. Uber engineers were quietly generating $500 to $2,000/month each. A dashboard would've caught it in week two, not month four.

4. Set a per-tool cap before you scale, not after. Caps feel restrictive; surprise bills feel worse. Set the ceiling while it's still cheap to change.

5. Kill the leaderboard. Uber gamified usage with internal rankings. You get exactly what you incentivize, and they incentivized spend.

💡 Pro tip: The companies winning with AI in 2026 aren't the ones spending the most. They're the ones who treat it like a supervised co-worker, not a slot machine.

Other Relevant AI News!

🤖 Microsoft is racing to tame OpenClaw, the open-source agent software that's wildly popular in China and helped Apple sell Macs, but is still too risky for most businesses to run.

🧠 Microsoft also teased a new in-house reasoning model, MAI-Thinking-1, plus a unified Copilot "super app," its clearest move yet to stop depending on OpenAI.

💸 A Gartner forecast says AI agent software spending will hit nearly $207 billion in 2026, up 139% in a single year, which is the macro behind Uber's budget blowout.

🚀 SpaceX, which owns xAI, could beat both OpenAI and Anthropic to the public markets, with an IPO that may value it well over $1 trillion. The three giants have been neck and neck, racing to go public first with the belief that getting out ahead will lead to the best market performance.

Golden Nuggets

• 🏛️ Washington asked the frontier labs to hand over their models. "Voluntarily" is doing a lot of work in that sentence.

• 🎬 Scorsese using AI for storyboards just moved the goalposts for what "real artists" will and won't touch.

• 💸 Uber torched a year of AI budget in four months. Cap and measure before you scale, or you're next.

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