Half a Million Tiny Dolphins

Midjourney announced something this week that made me check whether it was April 1st. It was not. The company known for generating images from text prompts is building a full-body medical ultrasound scanner. They are also building spas where you can go to get scanned.[1]

The device is called, straightforwardly, the Midjourney Scanner. You step onto a platform. The platform submerges you in water at a rate of about five centimetres per second. Your body passes through a ring containing half a million squares, each the size of a grain of sand. Each square can emit ultrasonic waves and record the ripples that bounce back. Midjourney compares them to dolphins using echolocation. So a scan is apparently like being surrounded by half a million tiny dolphins from every angle.[2]

The result, the company says, is a 3D map of your body accurate to a fraction of a millimetre. It looks like an MRI but takes less than 60 seconds instead of the 60 to 90 minutes a full-body MRI typically requires. That speed difference is the real point. MRI machines are expensive, slow, and scarce. If you could get a full-body scan in under a minute, the economics of preventive imaging change entirely.[3]

Midjourney is developing the scanner with Butterfly Network, a company that makes handheld ultrasound devices. The partnership gives Midjourney access to Butterfly's ultrasound-on-chip technology, which miniaturises what used to require bulky equipment into something that fits in a pocket. The project lead is someone who previously worked on the Vision Pro at Apple before joining Midjourney in late 2023.[4]

The timeline is ambitious but not absurd. Over the next 12 months, Midjourney will refine the algorithms and the hardware, run research trials, and work on a second-generation design. The first spa location is planned for San Francisco sometime in 2027. FDA approval for diagnostic capabilities will follow. By 2028, they want to launch a third-generation machine with custom silicon for better image quality. By 2031, the goal is 50'000 scanners worldwide.[5]

Fifty thousand scanners. To put that in perspective, there are roughly 40'000 MRI machines in the entire world right now. Midjourney is talking about surpassing that number within five years, with a device that did not exist as a concept 18 months ago.

The company's claim about outcomes is even bolder. They say that with enough early imaging, the world could avoid 30 percent of all deaths and 50 percent of all healthcare costs. That is a staggering assertion. It is the kind of number that sounds like it came from a pitch deck, not a medical journal. But the underlying logic is not crazy. Many serious conditions, cancers especially, are far more treatable when caught early. If full-body scans become cheap and fast enough to do annually, the detection curve shifts earlier. Whether that translates to 30 percent fewer deaths is a question for clinical trials, not blog posts.[6]

What I find interesting here is not the medical claim. It is the pivot. Midjourney built its reputation on an AI image generator. The image generator is still the product most people know them for. But the company is apparently asking itself, as it wrote in its announcement, "How do we want to be different?" and "What do we want to become?" The answer they arrived at is: medical hardware.

That is a jump that very few companies make successfully. It requires different expertise, different regulatory pipelines, different sales channels, and a different relationship with the people who use the product. Going from "type a prompt, get a picture" to "step into a pool of water surrounded by half a million ultrasonic transducers" is not a product extension. It is a reinvention.[7]

The comparison to MRI is deliberate. MRI machines use powerful magnetic fields and radio waves. They are loud, claustrophobic, and require specialised technicians to operate. The Midjourney Scanner uses ultrasound, which is safe, non-invasive, and does not require a shielded room. The tradeoff has always been resolution: ultrasound cannot match MRI for soft tissue detail. Midjourney's bet is that with half a million transducers and AI-assisted image reconstruction, they can close that gap.[8]

Maybe it works. Maybe it does not. The company has not published clinical data yet, and the FDA has not approved anything. But the engineering premise is sound. Ultrasound-on-chip technology has been advancing rapidly. AI image reconstruction has gotten dramatically better. The idea of combining those two things into a fast, full-body scanner is one of those ideas that seems obvious in retrospect but required someone willing to spend years and significant capital to actually build it.[9]

The spa framing is the part that makes people raise an eyebrow. Midjourney says the first locations will feel like spas, not hospitals. That is a deliberate choice. Medical imaging has a branding problem. People associate it with illness, anxiety, and waiting rooms. Rebranding a full-body scan as a wellness activity, something you do regularly and voluntarily, changes the adoption curve. It is the same logic that turned genetic testing from a medical procedure into a consumer product.[10]

Whether 50'000 scanners by 2031 is realistic, I have no idea. But I have learned not to bet against companies that are willing to build something physically ambitious while everyone else is shipping another language model. The image generator pays the bills. The scanner is the dream. I will be watching to see if the dream survives contact with the FDA.