What If a Full-Body Scan Cost Less Than a Spa Day?

I spent way too many hours last year typing prompts into Midjourney. Making cat detectives in raincoats. Reimagining my room as a Wes Anderson set. The usual stuff.

So when I saw the headline — Midjourney built a full-body medical scanner — my brain did a little skip. The AI image company? The one that makes those sometimes-six-fingered dreamscapes? Yes. That one.

The more I dug in, the less random it felt.

Photo by Accuray on Unsplash

Here’s the part that got me.

Midjourney has zero outside investors. No Andreessen, no Sequoia, no SoftBank knocking on the door asking about quarterly growth. They’re profitable — around $300 million in 2024, closer to $500 million in 2025 — and almost all of it comes from people paying $10 a month to generate art. That’s it. No ads, no enterprise contracts, no data harvesting.

And they used that money to build a whole-body ultrasonic scanner.

The deal with Butterfly Network is real — $15 million upfront plus $10 million a year, confirmed in an SEC filing. The scanner uses 40 Butterfly ultrasound chips that fire ultrasonic waves at an incredibly high rate — the company describes the sensor ring as working “like a dolphin using its echolocation” — and captures 17 gigabytes of raw data every second. Two petaflops of on-site compute turns that into a submillimeter-resolution 3D map of your body.

The current prototype takes about 20 minutes, not the 60 seconds they’re aiming for. And it’s not FDA-cleared for diagnostics — only body composition maps under general wellness rules. The team is nine people led by an ex-Apple Vision Pro engineer who has never built a medical device before.

None of that sounds like a sure thing. It sounds like a moonshot.

But here’s what I can’t stop thinking about.

The whole-body MRI at the fancy clinic near me costs 4,000 bucks and takes over an hour inside a tube loud enough to need hearing protection. Midjourney’s pitch — eventually a few dollars, 60 seconds, done in what looks like a spa — isn’t just cheaper. It’s a different category of thing. A scan is the way you get a massage. A monthly check-in with your insides.

Holz has talked about wanting to prevent 30% of all human deaths through early detection. There’s no evidence for that number yet — radiologists have pushed back, and a lot of what kills us (accidents, Alzheimer’s) won’t show up on any scan, and this is a company still figuring out a 20-minute prototype. But the ambition tells you something about how he thinks.

David Holz studied physics at NASA and neuroscience at Max Planck. At 22, he co-founded Leap Motion, and in enthusiast circles, people still talk about the hand-tracking model he built on a CPU cluster before TensorFlow existed — 600 million parameters, running on hardware that shouldn’t have been able to handle it. He later shipped an open-source AR headset because he thought the industry was moving too slowly.

You can see the pattern. He builds things that seem slightly crazy until they don’t. And he does it without asking permission.

That’s the thread I keep pulling.

Most hardware startups raise hundreds of millions before they ship anything. Medtech is worse — years of regulatory uncertainty, clinical trials, reimbursement puzzles. The standard playbook says you need VC billions and a 10-year roadmap before you even start.

Midjourney’s playbook is: have a bunch of paying customers in a completely unrelated business, use the profits to fund insane R&D, and announce whatever you build when you’re ready. No pitch decks. No board meetings. No investors are telling you to focus on your core competency.

The scanner might fail. The tech has real limitations — ultrasound struggles with bone, air, and deep tissue. The regulatory path for anything beyond wellness mapping is unclear. Holz himself told reporters he imagines “a future where the FDA had a class of devices,” which is founder-speak for “we don’t know how this gets approved yet.”

But the model itself — the way they’re building — is worth paying attention to regardless.

What else could get built this way? What other impossible hardware projects are sitting inside profitable little companies that nobody thinks to look at? What happens when you don’t need to convince a Sand Hill Road partner that your idea makes sense within their fund’s timeline?

I don’t know if Midjourney Medical changes healthcare. I don’t know if the scanner ever gets past 20 minutes or finds its way through the FDA.

But I know I’d rather watch a company try something this weird with its own money than wait for another committee to tell us what’s possible.

If this post resonated with you, buy me a coffee ☕ — it helps me continue sharing stories, ideas, and reflections.