{"slug": "david-holz-is-taking-midjourney-from-image-models-to-a-full-body-ultrasound", "title": "David Holz is taking Midjourney from image models to a full-body ultrasound scanner", "summary": "David Holz's Midjourney announced Midjourney Medical, a full-body ultrasound scanner that uses sound waves to create 3D body maps in under 60 seconds. The self-funded AI lab is moving from image generation into medical hardware, aiming to make body scanning fast and cheap, though the claims lack peer review or regulatory approval.", "body_md": "David Holz's [Midjourney](https://midjourney.com/?ref=runtimewire) has announced [Midjourney Medical](https://www.midjourney.com/medical/blogpost?ref=runtimewire), a move that pushes the self-funded AI image lab into medical hardware with a planned full-body ultrasound scanner and a consumer venue it calls the Midjourney Spa.\n\nThe announcement is deliberately outside Midjourney's familiar lane. Midjourney, best known for image and video AI models, says the medical project is \"not related to anything you've seen from us so far.\" The throughline is Holz: before Midjourney became one of consumer AI's most recognizable creative tools, he co-founded Leap Motion, the gesture-tracking hardware company that tried to change how people use computers. [Midjourney's first hardware](/article/midjourney-first-hardware-reveal-david-holz) reveal, which RuntimeWire [covered earlier](/article/midjourney-first-hardware-reveal-david-holz), already pointed back to those interface roots. Midjourney Medical makes that connection explicit: this is not a new prompt box, but another bet on how humans interact with machines.\n\nMidjourney describes the scanner as an underwater ultrasound computed tomography system, not a conventional CT scanner using X-rays. A person would step into a shallow pool, stand on a platform, and descend through a ring of sensors. The sensors send ultrasonic waves through the body from many angles, then Midjourney reconstructs the changed waveforms into images and AI segmentation maps. Midjourney says its goal is for the scan itself to take no more than 60 seconds.\n\nThat distinction matters because some coverage has called the machine a CT scanner. Midjourney's own technical pages and careers posts use the phrase \"ultrasound CT\" or USCT, while the announcement describes sound waves moving through water and tissue. In plain English, this is an ultrasound reconstruction system wrapped in a full-body consumer experience, not the same modality as a hospital CT scanner.\n\n### The machine is the story, but the business model is the tell\n\nThe hardware pitch is built around volume: more body data, gathered faster and cheaper, repeated often enough that changes over time become visible. Midjourney frames the target metric as \"megabytes per second per dollar\" of information about the body. That line is doing a lot of work. It recasts medical imaging from an episodic clinical procedure into an information product, which is closer to how Midjourney already thinks about compute, data, and consumer access.\n\nThe scanner design is ambitious. Midjourney says the platform lowers a person at about 2 inches, or 5 centimeters, per second, through a ring made of roughly half a million tiny elements. Each element is described as both a speaker and a microphone, emitting ultrasonic waves and recording returns millions of times per second. Midjourney says the system produces terabytes of data each second and would require thousands of computers to split the reconstruction task.\n\nMidjourney claims the output can form a 3D body map \"down to a fraction of a millimeter\" and look similar to MRI at nearly 100 times the speed. Those are company claims, not independently validated findings. The announcement includes reconstructed slices, AI segmentation overlays, body-volume sweeps, side-by-side USCT and MRI images in its [scan gallery](https://www.midjourney.com/medical/scan_gallery?ref=runtimewire), and phantom segmentation examples, but it does not include a peer-reviewed paper, FDA filing, trial readout, or third-party benchmark.\n\nThat is the central tension in the launch. Midjourney is very good at showing images that make a future feel legible. Medical imaging requires the harder proof: repeatability, sensitivity, specificity, safety, calibration, clinical workflow, and regulatory review. Midjourney appears to understand that gap. The first consumer-facing product will be detailed body composition maps, according to the company, not broad diagnostic readings. Midjourney says diagnostic capabilities generally require FDA approval and that it plans to submit regular test results to the FDA for expanded capabilities.\n\n[FDA device regulation](https://www.fda.gov/medical-devices/device-advice-comprehensive-regulatory-assistance/overview-device-regulation?ref=runtimewire) is not a side quest here. The FDA's device framework can require premarket clearance or approval depending on device classification and intended use, and diagnostic claims change the burden. A spa-like venue may make the scanner feel casual, but the second Midjourney starts telling users what a scan means medically, it is operating in a different world from image generation.\n\n### The spa is distribution, not decoration\n\nMidjourney is not proposing to sell a scanner into hospitals first. It is proposing a consumer environment. The [Midjourney Spa](https://www.midjourney.com/medical/spa_gallery?ref=runtimewire) is planned for San Francisco in 2027, with hot tubs, saunas, cold plunges, and scanning rooms. Midjourney says the first spa will open around the end of 2027, after a 12-month period focused on refining algorithms and hardware, running research trials, moving toward second-generation hardware, and building the first research spa.\n\nThat venue is the most Midjourney part of the plan. Midjourney's homepage describes a 60-person, self-funded, community-funded, distributed team, and the medical effort builds on that orientation toward a direct relationship with users. If the image model business gave Midjourney a direct relationship with millions of creative users, the spa would give it a direct physical relationship with people and their health data.\n\nThat is a much bigger leap than moving from still images to video. A recurring scan network depends on local operations, construction, staffing, service reliability, privacy controls, medical-device quality systems, and user trust. Midjourney's [medical careers page](https://www.midjourney.com/medical/careers?ref=runtimewire) shows the company hiring for the less glamorous parts of that bet: quality systems, device history files, verification and validation, HIPAA-aware data pipelines, scanner operations, facilities, and a spa project manager.\n\nThe job postings are a useful counterweight to the announcement's future-facing language. They show Midjourney knows this cannot be solved by model weights alone. It needs hardware integration, ultrasound experimentation, cloud reconstruction, regulatory quality processes, construction management, and on-site operations. That is what makes the project serious. It is also what makes it hard.\n\n### Holz is choosing a harder frontier than creative AI\n\nMidjourney's move is not a random brand extension. Holz has spent his career around interfaces that translate physical reality into computational form. Leap Motion tried to capture the hands. Midjourney captured imagination through image generation. Midjourney Medical is trying to capture the body as a recurring data stream.\n\nThe strategic reason to do it now is clear. Consumer AI products increasingly sit between people and health questions, but they are starved for personal, longitudinal, physical data. Midjourney's post says people increasingly talk to doctors and AIs about health, and that better decisions require more awareness of the body. The scanner is Midjourney's proposed input layer for that future.\n\nThe part Midjourney does not yet prove is whether the images are clinically useful at the breadth, cost, speed, and reliability the company imagines. Full-body preventive imaging is already a contested consumer category because more scans can mean earlier detection, but also incidental findings, anxiety, follow-up costs, and unclear outcomes if not tied to evidence-based care. Midjourney goes even further by packaging scans as a side effect of a spa visit.\n\nThe company's roadmap is expansive. Midjourney says it plans to begin scaling to more cities and upgrade to a third-generation scanner in 2028.\n\nThe more immediate test is narrower: can Midjourney build a scanner that produces reliable body composition maps, operate it safely in San Francisco, and generate enough trust that people return? If it can, Midjourney Medical becomes more than a hardware reveal. It becomes a case study in whether a profitable consumer AI lab can use its independence, compute stack, and community funding to enter a regulated physical market.\n\nHolz is betting that Midjourney's next interface is not a chat window or an image canvas. It is a pool of water, a ring of sensors, and a body moving slowly through sound.", "url": "https://wpnews.pro/news/david-holz-is-taking-midjourney-from-image-models-to-a-full-body-ultrasound", "canonical_source": "https://runtimewire.com/article/midjourney-medical-ultrasound-scanner-david-holz", "published_at": "2026-06-18 02:41:25+00:00", "updated_at": "2026-06-18 02:56:35.185912+00:00", "lang": "en", "topics": ["ai-products", "ai-research", "ai-infrastructure", "computer-vision", "ai-startups"], "entities": ["Midjourney", "David Holz", "Midjourney Medical", "Leap Motion", "RuntimeWire"], "alternates": {"html": "https://wpnews.pro/news/david-holz-is-taking-midjourney-from-image-models-to-a-full-body-ultrasound", "markdown": "https://wpnews.pro/news/david-holz-is-taking-midjourney-from-image-models-to-a-full-body-ultrasound.md", "text": "https://wpnews.pro/news/david-holz-is-taking-midjourney-from-image-models-to-a-full-body-ultrasound.txt", "jsonld": "https://wpnews.pro/news/david-holz-is-taking-midjourney-from-image-models-to-a-full-body-ultrasound.jsonld"}}