Can AI Prevent 30% of Deaths?
The company that taught the internet to dream in pixels now wants to look inside your body. The pitch is seductive. The science is still a promise.
There is a particular kind of vertigo that comes from watching a company step far outside the thing that made it famous. Netflix tried this with games. Amazon tried it with healthcare and largely retreated. Google tried it more times than anyone can count. But few pivots have been as conceptually strange, or as quietly audacious, as the one Midjourney announced this June.
Midjourney is the AI image generator that, for the last few years, has been synonymous with synthetic dragons, surreal portraits, and the occasional viral deepfake controversy. It has no outside investors, no IPO ambitions that anyone knows of, and, according to widely reported figures, hundreds of millions of dollars in annual recurring revenue from people paying to generate pictures of things that don't exist.
Now it wants to show you what's inside your own body. In sixty seconds. While you're at a spa.
The company calls the new venture Midjourney Medical, and its first product a Midjourney Scanner, a whole-body ultrasound imaging system the company is marketing under the name "Ultrasonic CT." It is, by Midjourney's own description, "a little weird and a little crazy, but also spectacular and filled with hope." It is also, depending on which radiologist you ask, either a genuinely interesting piece of engineering or a marketing video dressed up as a medical breakthrough.
Both things, it turns out, can be true at once.
Going Under, Gently
The experience Midjourney is designing reads less like a clinical procedure and more like a wellness ritual built around a sales pitch. You step onto a platform suspended over a pool of warm, golden-lit water. The platform begins to lower you in, at about two inches per second, the company says, like a slow elevator ride into a hot tub. As you descend, your body passes through a ring containing roughly half a million tiny ultrasound transducers, each about the size of a grain of sand, each able to both emit sound waves and listen for their echo.
Midjourney's own description leans hard into the imagery of marine biology: the sensors are described as acting like a school of dolphins, surrounding you with echolocation from every angle. In under a minute, the system is supposed to have collected enough acoustic data (described as terabytes per second, or the equivalent of 500 hours of HD video for a single second of scanning) to reconstruct a three-dimensional map of your internal anatomy, down to a fraction of a millimeter.
No MRI's claustrophobic tube. No CT scan's dose of ionizing radiation. No thirty-to-ninety-minute wait inside a magnet, listening to clanging coils. Just water, sound, and, Midjourney hopes, about as much ceremony as a trip to a day spa.
That, anyway, is the dream. And Midjourney isn't shy about the size of that dream: the company says it believes sufficiently early, sufficiently frequent imaging could help the world avoid 30 percent of all deaths and 50 percent of all healthcare costs.
The Spa Is the Strategy
If the scanner is the engineering bet, the spa is the behavioral one, and it might be the more interesting part of the story.
Midjourney's logic is that nobody wants to go to the doctor. People go to spas. So instead of building a sterile diagnostic device and hoping people make an appointment for it, Midjourney is building hot tubs, saunas, and cold plunges, and quietly hiding the diagnostic device inside the visit. "The scans are a side-effect," the company wrote in its announcement. "You barely think of them when going to the spa. But suddenly, you have a huge library of data about your health."
The first Midjourney Spa is slated to open in San Francisco at the end of 2027, reportedly outfitted with ten scanners that the company claims could collectively perform more body scans per year than every MRI machine on Earth combined. Beyond that flagship location, the roadmap gets considerably more ambitious: more cities by 2028, a third-generation scanner built on custom silicon, and a long-term target, by 2031, of more than 50,000 scanners worldwide, with enough throughput to scan a billion people a month.
It is, by any measure, one of the most aggressive infrastructure-deployment plans in recent memory for a company with zero clinical track record in medicine.
Borrowed Hardware, Original Ambition
For all the talk of dolphins and golden light, the underlying physics isn't actually Midjourney's invention. The company has built its scanner on technology licensed from Butterfly Network, a Massachusetts-based maker of handheld, chip-based ultrasound devices. According to reporting on the deal, the partnership (signed in late 2025) involves Midjourney paying Butterfly a one-time fee in the tens of millions of dollars plus ongoing annual licensing payments, with the total arrangement valued at roughly $74 million over five years. The prototype scanner reportedly uses around forty of Butterfly's "ultrasound-on-chip" modules, with later generations expected to use considerably more.
Markets responded immediately: shares of Butterfly Network jumped more than 50 percent in the days following the announcement.
It's worth pausing on what this partnership actually means, because it complicates the narrative Midjourney is telling about itself. None of the imaging technology is generative AI. The part of Midjourney that made it a household name, the diffusion models that turn text prompts into pictures, has essentially nothing to do with how this scanner is supposed to form an image of your liver. The AI here, to whatever extent it exists in the final product, is doing something closer to signal reconstruction and image segmentation: turning chaotic sound-wave data into a coherent picture, then identifying what's what inside it. Midjourney's actual contribution appears to be capital, ambition, computational scale, and a willingness to build the consumer-facing spectacle around someone else's sensor technology.
Butterfly Network's CEO, Joseph DeVivo, was notably more measured in his own public statement than Midjourney's blog post, describing the system as a "continuous window into your health" designed for "weekly use," while emphasizing that claims would need to be backed "as applicable" by valid clinical data.
Where the Skepticism Comes In
Reaction from the people who actually read body scans for a living was not uniformly enchanted.
Francis Deng, a neuroradiologist and assistant professor at Johns Hopkins Medicine, raised one of the more pointed technical objections that circulated after the announcement: ultrasound, as a physical phenomenon, simply does not pass cleanly through bone, air, or very deep soft tissue. That's not a software problem solvable with more processing power. It's a fundamental limitation of the modality. Large portions of the body, in other words, may remain stubbornly difficult to image clearly no matter how many tiny transducers surround the water tank. Deng also pointed to Midjourney's track record with anatomical accuracy in a different context, noting the company's image generator had previously produced an anatomically incorrect illustration that was published in an academic journal before being retracted, a pointed reminder, from a clinician's perspective, that visual plausibility and biological accuracy are not the same thing.
There's also the matter of regulatory status, which Midjourney itself is fairly candid about, if you read past the launch-day enthusiasm. The current device has no FDA clearance for diagnostic use. The company's stated plan is to begin by offering customers "detailed body composition maps," not diagnoses, while it submits test results to regulators in pursuit of expanded capabilities down the line. Independent analysts covering the announcement have been careful to underline that distinction: this is a first-generation prototype, demonstrated at a launch event, whose more dramatic comparisons to MRI quality are the company's own framing rather than anything independently verified. Even the device's marketing name has drawn scrutiny. "Ultrasonic CT" invokes a totally different, radiation-based imaging modality (computed tomography) despite using none of the same physics, which several outlets have flagged as a potentially misleading choice of words.
None of this means the underlying approach is fraudulent or doomed. Ultrasound tomography in a water bath is, in fact, an established research concept in medical imaging. Midjourney isn't claiming to have invented physics from nothing. What's new here is the scale of ambition, the amount of money being committed, and the unusual decision to wrap a diagnostic device inside a hospitality business. Whether that combination produces a genuine advance in preventive medicine or a very expensive, very well-marketed wellness amenity is, at this point, an open question, and one that won't be settled by a blog post, a concept video, or a stock pop, but by the regulatory filings, clinical trials, and independent scrutiny still to come.
What Happens Next
Midjourney's own timeline gives a sense of just how much runway separates the announcement from the reality. The next year is described as a period of algorithm refinement, hardware iteration, and research trials, the unglamorous work of getting a device that currently exists as a prototype and a concept video to behave consistently on real bodies. The "research spa" that's supposed to generate this real-world data isn't expected until close to the end of 2027. The custom-silicon, "serious" third-generation hardware isn't slated until 2028. The headline figure of 50,000 scanners worldwide is a 2031 target, five years out from today, in an industry where medical device approvals routinely take that long on their own.
Midjourney describes itself, pointedly, as having no outside investors, a "community-backed research lab," in its own words, funded by the same subscribers who pay for AI-generated images. That structure gives it a kind of freedom most biotech ventures don't have: no board demanding a path to profitability on a fixed timeline, no venture capital clock counting down. It also means there's no institutional adult in the room with deep medical-device experience forcing rigor onto the announcement before it goes out the door, which may be exactly why the blog post reads more like a vision statement than a regulatory submission.
The most honest way to describe where things stand today is this: Midjourney has announced an intention, licensed someone else's sensor technology, signed a real and apparently substantial commercial deal, and produced a concept video of a spa that doesn't exist yet, housing a device that hasn't been cleared for the purpose it's ultimately meant to serve. That's a real thing. Companies do build real businesses this way. But it is not yet a medical breakthrough, whatever the lighting in the renderings suggests.
Whether the Midjourney Scanner ends up changing how millions of people understand their own bodies, or simply joins the long list of ambitious health-tech launches that quietly fade once the regulatory and clinical realities set in, is a question that will be answered by data, not by golden light. For now, the company has done what it has always done best: make something that looks extraordinary before anyone has had the chance to verify that it works.
Sources: Midjourney Medical's official announcement; reporting from Engadget, Gizmodo, MobiHealthNews, The Imaging Wire, and Radiology Business; commentary from Francis Deng, MD (Johns Hopkins Medicine) and Butterfly Network CEO Joseph DeVivo.
Comments
Post a Comment