Our Take
A hardware pivot from a text-to-image company that has tested a dozen people and makes no claim about diagnostic accuracy or FDA approval, positioned as preventative scanning for biohacking.
Why it matters
This signals Midjourney's intent to diversify beyond generative AI into direct-to-consumer health hardware, but the lack of published benchmarks or independent validation means the imaging claims remain unproven. The spa model also sidesteps medical device regulation by framing scans as body composition mapping rather than diagnosis.
Do this week
Healthcare vendors: audit whether Midjourney Medical's data privacy claims and liability model (still forthcoming) meet your compliance requirements before considering integration.
Midjourney reveals ultrasound hardware and spa ambitions
Midjourney CEO David Holz announced The Midjourney Scanner, a full-body ultrasound device built in partnership with Butterfly Network. The scanner uses a ring of underwater transducers (40 Butterfly Ultrasound-on-Chip modules per system) to send ultrasonic waves through a user's body from multiple angles, then reconstructs 3D images of muscle, fat, bone, and organs in approximately 60 seconds.
The scanning process requires the user to step onto a platform that descends through water containing thousands of ultrasound transducers. The system processes the returning waves using two petaflops of computing power to generate detailed volumetric images. To date, about a dozen people have been scanned (company-reported).
Midjourney plans to open a "spa" location in San Francisco's Union Square before the end of 2027, equipped with 10 scanners alongside gym facilities, saunas, and cold plunges. The company is framing the scanners as tools for body composition tracking rather than diagnostic imaging, allowing it to avoid FDA clearance as a medical device. Users will be able to share their scan library with doctors or AI health tools, though Midjourney says data privacy details will be disclosed closer to launch.
Holz stated that he envisions eventual imaging quality comparable to MRI without radiation or strong magnetic fields, and suggested users could be scanned daily to track changes from diet and exercise. He also indicated interest in an FDA regulatory pathway for exploratory scanning devices that would allow broader data collection without diagnostic claims.
The real story is regulatory positioning, not imaging innovation
Midjourney's hardware move is a business diversification play, not a breakthrough in ultrasound imaging. Butterfly Network's transducer technology is established; Midjourney's contribution appears to be compute and commercial deployment via the spa model.
The critical gap is absence of independent benchmarking. Holz claims image quality "comparable to MRI in many ways" but provides no peer-reviewed validation, side-by-side comparisons, or independent reproducer data. The phantom scan shown validates sensor alignment under controlled conditions, not diagnostic utility or clinical accuracy.
The spa-as-regulatory-moat is the sharper move. By classifying scans as "body composition maps" rather than diagnostic imaging, Midjourney sidesteps FDA 510(k) clearance and the clinical evidence burden. This model works until regulators or insurance companies demand proof of safety or accuracy for scans marketed to health-conscious consumers seeking daily tracking. Holz's comment about a future FDA "class of devices" for exploratory scanning hints at where this heads: a regulatory category that permits collection without clinical validation.
The unanswered question is what Midjourney's generative AI platform contributes to this effort. The company has petaflop-scale compute capacity available after fulfilling image generation demand; repurposing idle hardware for health scanning is economically rational but technologically unremarkable.
What to watch before integrating or relying on these scans
Health tech vendors considering partnerships or data integration should request published imaging benchmarks comparing Midjourney Scanner output to standard ultrasound and MRI on the same anatomies. Without independent third-party validation, the spa scans are closer to consumer wearables than clinical imaging tools.
Check the data privacy policy when it ships. Midjourney will hold detailed volumetric body data for potentially millions of users; the terms governing access, retention, and third-party sharing will determine whether hospitals, insurers, or researchers can ethically use the library. Any integration should depend on explicit consent models and audit trails.
If you are a user considering daily scans for personal health tracking, understand that sharing results with your doctor does not make them clinically validated. Physicians may decline to act on non-diagnostic scans, especially if image quality or anatomy identification is inconsistent.