Midjourney, the AI Image Company, Now Wants to Scan Your Whole Body in 60 Seconds

TL;DR

Midjourney, the profitable text-to-image company, just announced Midjourney Medical and a prototype full-body ultrasound scanner that claims to image your entire body in about 60 seconds with no radiation and no magnets. It runs on Butterfly Network's ultrasound-on-chip technology: 40 imaging modules, roughly half a million sensors, and over two petaflops of on-site compute. The first machines are slated for a San Francisco "spa" by the end of 2027, with a stated target of 50,000 scanners by 2031. The hardware is real and the Butterfly deal is signed. The grand health claims are not, and the FDA has cleared nothing yet.

Wait, the image-generation company?

Yes. Midjourney built its name and its revenue on prompt-to-image generation, famously without outside venture funding. Now it is pointing some of that cash at hardware. The new division, Midjourney Medical, is led by Ahmad Abbas, who worked on Apple's Vision Pro before joining Midjourney's consumer hardware effort in late 2023.

This is a bigger leap than the usual model-company-ships-an-app move. It is software money buying its way into regulated medical hardware, a domain where iteration speed is throttled by physics, manufacturing, and the FDA rather than a training run.

How the scanner actually works

The pitch is tomographic ultrasound, which Midjourney describes as a kind of "Ultrasonic CT." You stand on a platform that lowers you into a shallow ring of water lined with sensors. Ultrasonic waves fire through your body from every angle at once, and the reflections are reconstructed into a 3D map of muscle, fat, bone, and organs down to roughly millimeter precision. Midjourney likens it to dolphin echolocation.

The numbers are where it gets interesting for this audience. Butterfly's statement puts the current prototype at 40 Ultrasound-on-Chip modules and approximately half a million sensors scanning simultaneously, generating enough data that reconstruction takes over two petaflops of processing. The company says it is designed for weekly use, and frames the roughly 60-second scan against the 60 to 90 minutes a traditional MRI can take.

The Butterfly deal is the real engine

None of this works without Butterfly Network, the ultrasound-on-chip company that shrank ultrasound transducers onto silicon. The two signed a co-development agreement whose terms were filed on November 17, 2025, worth up to $74 million to Butterfly over a five-year term. The work runs through Butterfly Embedded, the licensing arm formerly known as Octiv.

"The future of health starts with awareness: understanding your body earlier and tracking how it responds to factors like diet, exercise, and behavior." Joseph DeVivo, Butterfly Network President, CEO, and Chairman.

Butterfly also signaled where this goes next: future generations are "expected to utilize substantially more imaging modules," and Midjourney has talked about a third-generation machine with custom silicon by 2028. Butterfly's stock jumped on the news, which tells you the market reads this as a real licensing pipeline, not a science-fair render.

Why a builder should care

Strip away the spa branding and this is a compute-and-reconstruction problem. Half a million sensors firing simultaneously produce a firehose of raw signal, and turning that into a usable 3D volume in seconds is an inference and signal-processing workload. Two petaflops sitting next to a bathtub is the tell: the hard part is the reconstruction stack, which is squarely in AI-adjacent territory.

It is also a preview of where generative-AI money flows when the core product matures. A cash-generating model company is vertically integrating into sensing hardware, the same way the frontier labs are reaching into chips and data centers. The lesson for indie builders is less "go build a scanner" and more that the durable moats are increasingly physical.

The hype, and the part to be skeptical about

Midjourney's framing leans hard on prevention. It claims the scanner is nearly a hundred times faster than an MRI and pitches a future where early detection could, in its words, avoid 30 percent of all deaths and 50 percent of all healthcare costs. Treat those as marketing, not findings.

Whole-body screening of healthy, asymptomatic people is genuinely contested in medicine. It surfaces incidental findings that trigger anxiety, follow-up scans, and procedures, many of which never would have hurt you. "Designed for weekly use" is exactly the pattern that worries clinicians about overdiagnosis. On top of that, ultrasonic CT at this resolution and whole-body coverage is unproven against MRI for soft tissue, and the device is a prototype: it needs FDA clearance before it can be deployed as anything more than a wellness gadget. The spa positioning conveniently lives in the lightly regulated wellness lane while the clinical claims wait on the regulator.

The timeline

• Now: working prototype announced June 2026, built on 40 Butterfly modules.
• End of 2027: first Midjourney Spa in San Francisco, reportedly housing around ten scanners alongside saunas, cold plunges, and a gym.
• 2028: a third-generation machine with custom silicon for better image quality.
• 2031: a stated target of 50,000 scanners worldwide.
• Gating everything: FDA approval, which the company has not yet obtained.

Key Takeaways

• The hardware is real and the deal is signed. Butterfly confirmed 40 ultrasound-on-chip modules per prototype and a co-development agreement worth up to $74 million over five years.
• It is a compute problem in a wellness costume. Roughly half a million sensors and over two petaflops to reconstruct one scan put this closer to an inference pipeline than a traditional imaging device.
• The health claims are marketing. "Avoid 30 percent of deaths" and "100x faster than MRI" are Midjourney's pitch, not peer-reviewed results, and whole-body screening carries real overdiagnosis risk.
• Nothing ships without the FDA. This is a prototype with a 2027 spa target and a 2031 scale target, all gated on clearance the company does not have.
• AI money is going physical. A profitable image-gen company vertically integrating into medical sensing is a signal about where mature AI cash flows next.

Sources: Butterfly Network investor relations, Engadget, AuntMinnie, PYMNTS