UPDATED: June 18, 2026 @ 2:45PM PT
PUBLISHED: June 18, 2026 @ 11:59AM PT
The AI image company best known for generating dreamlike art from text prompts has announced its first hardware product — and it is not a camera, a headset, or anything in its image-generation wheelhouse. On Wednesday, Midjourney CEO David Holz took the stage at a San Francisco jazz club to unveil the Midjourney Scanner, a full-body ultrasonic imaging device, along with plans to open a flagship "Midjourney Spa" in Union Square by late 2027. The pivot is so jarring that even Holz acknowledged it: "The first hardware project we're releasing is a little different from the cat pictures we've been making."
What the Scanner Actually Does
The device works by having you step onto a platform that slowly lowers you into a shallow pool of water, passing your body through a ring of ultrasonic sensors at roughly two inches per second. According to Midjourney, the system uses 40 Butterfly Ultrasound-on-Chip imaging modules per scanner and generates a full-body 3D map with no radiation and no powerful magnets. The stated goal is a scan time of 60 seconds — comparable to a full-body MRI that currently takes 60 to 90 minutes. Holz claimed the image quality is "in many ways superior to even MRI machines," a statement that has been met with pointed skepticism from engineers and clinicians who want peer-reviewed data before anyone gets too excited.
One important caveat that did not make the highlight reel: the current Gen 1 prototype takes around 20 minutes per scan, not 60 seconds. That target figure is aspirational. As reported by BigGo Finance, Holz himself described the prototype as a "MacGyver version" built largely from off-the-shelf components. A Gen 2 device is planned for late 2026; a Gen 3 built on custom silicon is further out. About a dozen people have been scanned so far.
The Hardware Partnership
The technical foundation rests on a licensing deal signed in November 2025 with Butterfly Network, a medical ultrasound company with FDA-cleared handheld devices. The five-year co-development agreement involves up to $74 million in expected payments to Butterfly, including a $15 million upfront fee and $10 million in annual license payments. The Midjourney partnership was significant enough to drive Butterfly's stock up roughly 27 percent after its Q4 2025 earnings call and contributed $6.8 million of Butterfly's record quarterly revenue, as reported by Business Wire. Butterfly is a real, credentialed hardware partner — that much is verifiable. Whether its chips, in this configuration, produce images that can do what Holz claimed is an entirely different question.
Here is an irony worth sitting with: as The Next Web pointed out, the scanner itself uses almost no generative AI. "We're not even using any AI in this yet, just really cool hardware and software," Holz told Bloomberg. The AI component is limited to labeling and segmentation of scan results. The technology that made Midjourney famous and profitable turns out to have almost nothing to do with the product it is now staking its future on.
The Union Square Spa Plan
The consumer rollout centers on a roughly 25,000-square-foot flagship space in Union Square, according to The Next Web. The facility will pair 10 scanners with hot tubs, saunas, cold plunges, and a gym — positioning a full-body scan as something you do after a cold plunge, not after getting a referral from your doctor. Midjourney's own announcement describes the spa as a place "you love going, whether it's by yourself or with friends," where "the scans are a side-effect." The Union Square location is slated to open at the end of 2027.
The concept draws on a tech-wellness convergence that has been simmering in San Francisco for years, from the Noe Valley infrared sauna studios to the SoMa longevity clinics. But this is something considerably more ambitious — and more medically loaded — than a hot yoga membership. The framing that a diagnostic-adjacent body scan is simply a pleasant lifestyle amenity is doing enormous work here, and it will be tested hard by regulators, insurers, and ultimately customers who may find their scan results concerning and have nowhere to take them.
The Skeptics Are Out
The announcement landed with a mix of genuine excitement and technical alarm in professional circles. On LinkedIn, Bartek Włodarczyk, a synthetic data CEO with a PhD in AI, called the announcement "hype laundering," writing that the MRI-level claim lacks clinical validation, regulatory clearance, sensitivity and specificity data, or peer-reviewed reproducibility studies. Oleg Karakash, an AI automation engineer who weighed in on the same thread, noted that "ultrasound penetration drops sharply past 15cm and soft tissue contrast is bounded by acoustic impedance, not signal processing. AI can sharpen the picture but it can't make ultrasound see brain tissue or deep contrast the way MRI does. A 60-second full body scan at MRI-level resolution would be a physics breakthrough, not a software product." A nuclear physicist commenting on the post wrote that reconstructing usable images from full-body acoustic waves "is not something to be taken for granted," calling the MRI comparison an overstatement. Another commenter — a cardiovascular surgeon — flagged that the promotional video showed a woman stepping out of the water bath completely dry, suggesting the imaging demonstrations may be renders rather than real scan outputs.
To be fair, some credentialed voices were more open. A former medical imaging industry executive writing on the same thread said that, if the scan quality holds up, the technology "can absolutely disrupt waiting times for patients as well as the cost of scan dramatically," while noting it would not replace modalities like PET-CT or contrast-enhanced MRI. A note published by Latent Space summarized the professional consensus as: the engineering demo is interesting, but "the gap between 'cool full-body images' and 'safe, reimbursable, diagnostic healthcare product' is the whole ballgame." The scanner also appears to be limited to the body from the shoulders down — submerging a person's head creates obvious logistical problems — which means it cannot replace neurological MRI regardless of how the physics questions resolve.
Separately, researchers at USC and Caltech published findings in January 2026 demonstrating a 3D hybrid system combining rotational ultrasound tomography with photoacoustic imaging that achieved whole-body imaging without radiation — suggesting the underlying scientific territory is real and actively developing. What that paper also makes clear is that academic researchers with full institutional backing and peer review are still working on proof-of-concept trials. Midjourney, with nine people and a dozen subjects scanned, is at an earlier stage than the launch event implied.
The Theranos Question
The comparison is uncomfortable but unavoidable. Theranos promised revolutionary blood-testing technology that would transform consumer healthcare, framed the consumer rollout as a wellness product to sidestep clinical validation, attracted credulous investors and media coverage before the science could catch up, and imploded spectacularly. Midjourney is not Theranos — the physics of ultrasound tomography are established, Butterfly Network is a real hardware partner with SEC filings, and David Holz has not claimed the scanner is already diagnosing diseases. The Midjourney Scanner, as described, occupies a narrower initial lane: body composition maps, not diagnoses. But the pattern of grand claims preceding clinical data, the deliberate positioning just below the FDA's diagnostic threshold, and the consumer-spa delivery mechanism all carry echoes that observers have been quick to name. At least one LinkedIn commenter posed the Theranos comparison directly. AI Chat Daily noted in its coverage that "a consumer spa selling near-diagnostic body scans is closer to the Theranos-era wellness category than to medical imaging." The burden is on Midjourney to distinguish itself through published data, not positioning.
The Pivot Economy
It is worth noting that Midjourney's move runs in the opposite direction of another high-profile San Francisco-adjacent pivot happening simultaneously. This month, the company formerly known as Allbirds — the sustainable wool sneaker brand that once had a cult following in these very neighborhoods before its revenue collapsed from $298 million in 2022 to $152 million in 2025 — completed its rebrand to Smartbird and pivoted into AI compute infrastructure, as reported by Gizmodo. A shoe company becoming an AI company; an AI company becoming a hardware medical spa operator. Both pivots were greeted with investor enthusiasm and public bafflement in roughly equal measure. The new CEO of Smartbird reportedly told Business Insider she was "blissfully unaware of all things Allbirds" and that "in a few months, people won't even remember the shoes." Whether Midjourney's customers will eventually forget the cat pictures remains to be seen.
Regulatory and Privacy Questions
For now, Midjourney Medical is sidestepping the FDA's most demanding requirements by positioning the scanner as a body composition tool rather than a diagnostic device. Holz said the company intends to layer on FDA approvals incrementally — starting with things that are easy and building toward diagnostic capabilities over time — but anything that crosses into clinical diagnosis triggers the FDA's full Software as a Medical Device framework and premarket review requirements, as R&D World noted. The path from "body composition map" to "MRI rival" runs directly through that regulatory gauntlet, and the FDA has not been known to expedite it.
Data privacy is an equally serious open question. A spa that routinely collects high-resolution 3D scans of your internal organs is generating an extraordinarily sensitive category of personal health information — more intimate, in some ways, than anything in your medical records today. How those files will be stored, who can access them, whether they can be shared with insurers or employers, and what deletion rights customers will have are questions Midjourney has not yet answered in any meaningful detail. Midjourney's existing privacy policy was written for an image-generation platform. It was not designed with internal organ maps in mind.
The Baggage Midjourney Brings to This
The company arriving at Union Square with a medical scanner is also the company currently fighting consolidated copyright lawsuits from Disney, Universal, and Warner Bros. Discovery in federal court in Los Angeles, with a discovery cutoff set for August 2026 and trial motions expected by late that year, per court records at CourtListener. Warner Bros. called Midjourney "a bottomless pit of plagiarism" in its complaint. Three of Hollywood's five major studios have now sued the company. Midjourney is defending on fair use grounds, but the potential financial exposure, if courts side with the studios, could reach into the billions. Holz has not explained how the scanner program would be funded if that litigation goes badly.
Midjourney is entirely self-funded, having reached an estimated $500 million in annual revenue without taking outside investment, as noted by Time. That financial independence is real and meaningful — and it means there is no VC board demanding a pivot back to core competencies. It also means there is no institutional investor backstop if the scanner program burns cash, the lawsuits produce adverse rulings, and the Union Square retail lease turns out to cost what Union Square retail leases typically cost.
What to Make of It
What Midjourney unveiled Wednesday is a prototype scanned on about a dozen people, built by a nine-person team, with no FDA clearance, no published imaging data, and no disclosed pricing, packaged inside a spa concept slated to open on a street that has been navigating its own revival struggles. The hardware partner is credible, the underlying physics of ultrasound tomography are real and actively being researched at the academic level, and Holz has a track record of building things that seemed implausible. None of that means the scanner will work as claimed at scale, clear the FDA on the described timeline, or that the residents of San Francisco will line up to lower themselves into a pool in Union Square and have their organs mapped. As The Next Web put it plainly: "The most accurate way to describe the Midjourney Scanner is an ambition with a launch event attached." The real verdict arrives when published imaging data and formal FDA submissions appear — not before.
