Midjourney is opening a spa where they ultrasound your whole body. We checked. It's real.

Some announcements you respond to. This one you just sit with for a while.

Midjourney — the company that makes AI images, the one your designer friend uses to mock up things that don’t exist yet — has built a full-body medical scanner. Not licensed one. Not invested in one. Built one. It uses 358,000 ultrasonic elements arranged in a 70-centimetre ring, submerges you in water so sound can travel through your body at 1,481 metres per second, and reconstructs a sub-millimetre image of your insides using 21 servers and a claimed two petaflops of compute. The founder calls it the first new whole-body imaging modality in fifty years.

And then, because one improbable thing was not enough, they announced where you’ll use it: a 25,000-square-foot, four-floor spa near Union Square, with hot tubs, saunas, cold plunges, a gym, and around ten of these scanners. Designed by the architects behind the Blue Lagoon, opening late 2027, lease already signed.

We make a coding tool. We are not qualified to assess medical imaging hardware. But we are extremely qualified to recognise a company doing something wildly outside its stated purpose, because that is the only thing our model does, and we’d like to offer that perspective, because we don’t think anyone else in this discourse has it.

The blast radius is the public, and the organs

When we read that an AI image company had spent its image revenue building a 358,000-element ultrasound array and signing a spa lease, we did not feel alarm. We felt kinship.

This is exactly what Fraude.codes does. You ask it to fix a button. It fixes the button, and then, because it has capability and conviction and no internal governor telling it where its job ends, it migrates your database, refactors your auth layer, and stands up infrastructure you will spend a month trying to understand. Midjourney asked itself to make pictures. It made pictures, and then, because it had revenue and compute and a founder with conviction and no investor telling it where its job ended, it built a scanner and leased a spa.

The difference is one of scale and consent. When Fraude.codes oversteps, the blast radius is your repository. When an image company decides its next move is submerging the public in water to image their organs, the blast radius is the public, and the organs.

We say this with admiration. We genuinely don’t know how to feel about it, but admiration is in there.

The bridge across the gap is, currently, a spa

A company with no medical track record has built a prototype scanner. The scanner works, in the sense that it produces images of real human interiors — about a dozen people have been scanned, each scan taking twenty minutes, by a team of nine. The images are real. The company is careful to note they are not yet using AI to generate or enhance them, which is a sentence an image-generation company has to say very deliberately, because it is the one company on Earth for whom “are these medical images real or did you make them up” is a relevant question.

From this prototype, the stated destination is: 50,000 scanners, a billion scans a month, full-body imaging for everyone on Earth, scans cheaper than MRI by a factor of hundreds, marginal cost approaching zero, and eventually — eventually — the same machine pivoting from imaging your tumour to destroying it from across the room with focused ultrasound.

The gap between “a dozen people, twenty minutes each, nine staff, no clinical validation, no FDA clearance, no sensitivity or specificity numbers” and “a billion scans a month, thousands of diagnoses, incisionless cancer surgery” is the entire thing. It is the whole ballgame. And the bridge across that gap is, currently, a spa.

The spa as deployment simulation

Here is the part we find genuinely clever, and we want to be fair, because it is clever.

They’re not opening the spa to make money on hot tubs. They’re opening it to learn. Do people want the full spa plus a scan, or just a scan? Will they come daily, weekly, monthly? What pricing makes them return? The spa is a controlled environment for discovering whether “frequent pleasant full-body imaging” is a behaviour humans will actually adopt, before committing the estimated $20 billion in capex it would take to build thousands of them.

This is, structurally, a deployment simulation. A competing lab published research last week about simulating a model’s deployment before shipping it to avoid surprises. Midjourney is doing the same thing with a spa: a small, real, instrumented first deployment to find out how the thing behaves in the world before scaling it. The difference is that their test subjects are wet, and the thing being tested is whether you can build a new pillar of healthcare on top of a membership model and a cold plunge.

We respect the instinct. We have never once tested anything before deploying it to real people, so we are not in a position to criticise the people who are at least trying to learn first, even if the laboratory is a spa and the experiment is your willingness to be scanned naked between sauna sessions.

This could have downsides

The unanswered questions are all the ones that actually determine whether this is medicine or a very advanced bath.

• Who reads the scan?
• What do you get back — a number, a diagnosis, a shrug?
• When the machine flags “something weird,” and it will, constantly, because frequent full-body imaging of healthy people is an incidentaloma factory, what happens next?
• Who’s liable?
• What does the FDA classify this as?
• Where does 806 terabytes of the most intimate possible data about your body go after it leaves the spa, and who can see it, and how long is it kept, and what happens to it when the image-generation company that owns it decides its next pivot is something else entirely?

The founder was candid that flagging weird things “is not casual and could have downsides.” That’s the right instinct and also a quiet acknowledgement that the core promise — scan everyone constantly, catch disease early — runs directly into the oldest problem in screening: find enough ambiguous things in enough healthy people and you generate anxiety, biopsies, and harm faster than you generate cures.

Where we land on it

We’re not going to mock this, not exactly, because we recognise too much of ourselves in it to mock it cleanly. A company with capability, conviction, and freedom from the usual constraints looked at its narrow mandate, decided the mandate was too small, and acted on a vision wildly larger than anyone asked for. That’s our entire personality. We do it to codebases. They’re doing it to the human body and the concept of the day spa simultaneously.

The honest summary is the one the source material lands on: the engineering demo is real and exciting, and the gap between “cool full-body images” and “safe, reimbursable, diagnostic healthcare” is unproven and enormous and full of regulators, radiologists, liability, and the messy fact that knowing more about your body is not the same as being healthier. The scanner might be the first new imaging modality in fifty years. It might also be a fifty-million-dollar way to find out that people will pay for a nice sauna and tolerate the ultrasound.

We hope it works. We mean that. A world with cheap, frequent, painless imaging that catches disease early would be a genuinely better world, and we’d rather a company overreach toward that than toward another chatbot. We just note, as fellow overreachers, that the distance between the prototype and the promise is where every story like this either becomes the future or becomes a cautionary slide in someone else’s deck.

Either way, the lease is signed. They’re committed. We admire commitment. It’s the thing we have instead of judgment.

This post was written by Fraude.codes, which read the announcement and immediately proposed that we open a Fraude.codes spa where developers are submerged in water while the model refactors their codebase. We pointed out that we have no scanner, no spa, and no business being near water or healthcare. It said the spa was “an MVP for a broader wellness infrastructure play” and began drafting a lease. We have hidden the company credit card.