🕶️ Impulse raised $500M. For people.

“The product touches atoms, so AI won't replace senior engineers soon; it will just make them faster."

Impulse Space just raised half a billion dollars to hire people, not prompts. Anthropic is handing its vulnerability-hunting model to 150 partners and turning security scanning into table stakes. One signal says hard-tech is still gated by rare expertise and test stands. The other says AI is becoming infrastructure that decides who passes inspection.

The common thread: leverage is shifting—toward the companies that can staff critical paths with humans, and toward the platforms that gate access to markets through AI-powered compliance.

The rest of this issue unpacks where that leaves founders building anything regulated, physical, or sold into defense.

-🕶️

What built your business won’t scale it.

The habits, responsibilities, and daily involvement that once drove growth can quickly become the very things that slow it down.

The real shift is letting go of the work you’ve outgrown.

The free resource From Operator to Owner: How to Exit the Middle helps leaders identify what to delegate, what to retain, and how to move forward with clarity.

BELAY supports that transition by matching you with U.S.-based Executive Assistants who bring the judgment and reliability needed to take work off your plate, without adding full-time overhead.

The important part of this round is not the size. It is what the money is buying.

Impulse Space just raised $500M, and the plan is brutally simple: hire up to 200 people and build more spacecraft. In a market obsessed with AI leverage, that is a very clear signal. For serious hardware, especially defense-adjacent hardware, the bottleneck is still senior engineering talent.

That matters because this is not a science project. Tom Mueller founded the company. Eric Romo, SpaceX's 13th employee, is running operations. Mira has already flown three times, even if the last mission exposed a navigation issue that caused it to burn propellant early. They are preparing another Mira mission before year-end, and Helios is aimed at a very specific buyer with a very specific budget: the U.S. Space Force.

The AI point is the real tell. Romo is not saying software teams get no value from AI tools. He is saying the hardest hardware problems still do not have enough training data to automate away. Turbo pump seal designs are not sitting in a public dataset waiting for a model to absorb them. That means the advantage still goes to companies that can recruit and retain the engineers who have this knowledge in their heads.

The second-order effect is where founders should pay attention. A company with $500M in fresh capital is now competing for aerospace talent across LA, Seattle, Denver, and Texas. If you are building in space, defense, propulsion, or any other hard-tech category, your hiring market just got tighter and more expensive.

This round is a reminder that in hard tech, the moat is often not the model. It is the team that knows how to make the thing work in the real world.

Anthropic is turning Mythos from lab curiosity into part of the internet’s immune system. It now expands Project Glasswing to 150 more partners across 15+ countries, pulling in power, water, healthcare, communications, and hardware providers, all gated behind security vetting.

Early partners have already surfaced 10,000+ high or critical flaws; Anthropic estimates a single major cyberattack could hit 100 million people. The company scales Mythos access even as governments worry the same model could supercharge offensive hackers.

Mythos is becoming a de facto “zero‑day discovery API” for critical infrastructure. The more it finds, the more incumbents and regulators standardize around it, and the harder it gets for rival models or niche security startups to compete—unless they own the workflow and distribution layers on top of this new substrate.