Midjourney’s billion-scan bet: hype, hurdles, and the ultrasound spa revolution

The Midjourney Scanner represents a groundbreaking leap in medical imaging by employing an innovative ultrasonic CT approach that utilizes a ring embedded with approximately 358,000 to half a million ultrasonic sensors. These sensors fire structured sound waves up to 1000 times per second as the patient is lowered through a water-filled chamber, enabling rapid acquisition of several hundred high-definition body slices within about 60 seconds. This method achieves MRI-like 3D imaging quality of muscle, fat, bone, and organs without the use of radiation or magnets, dramatically reducing scan times and costs.

Central to this technological feat is Midjourney's strategic $74 million licensing and co-development partnership with Butterfly Network, which provides exclusive access to Butterfly’s ultrasound-on-chip modules. Each scanner system integrates 40 of these advanced transistor-based ultrasound chips, enabling the compact assembly of hundreds of thousands of sensors into a scalable ring array. This collaboration not only empowers the scanner’s unprecedented resolution—detecting motions at picometer and even femtometer scales with half-millimeter tissue precision—but also signals a potential new business avenue for Butterfly beyond its traditional handheld ultrasound devices.

The Midjourney Scanner’s data processing capabilities are equally impressive, handling a staggering throughput of 17 gigabytes per second and reconstructing up to 800 terabytes of raw data per scan across 21 servers powered by two petaflops of compute. This immense computational infrastructure supports the creation of detailed 3D anatomical maps from hundreds of ultrasonic slices, facilitating not just one-time diagnostics but longitudinal monitoring of patients’ internal health. By fusing AI-driven analysis with cutting-edge hardware, Midjourney is pioneering a transformative shift in how anatomical data is captured and interpreted.

Midjourney has made a strategic pivot from AI image generation to medical hardware by developing a rapid, full-body ultrasound scanner that leverages Butterfly Network’s ultrasound-on-chip technology under a $74 million, five-year licensing and co-development agreement. This partnership grants Midjourney exclusive access to Butterfly’s FDA-cleared sensors, enabling the company to build on proven technology rather than inventing new hardware from scratch, positioning it to disrupt the medical imaging market with a cost-effective alternative to MRI.

Central to Midjourney’s market deployment strategy is the launch of consumer-facing wellness spas, beginning with a flagship location in San Francisco’s Union Square by late 2027, equipped with around 10 scanners. These spas aim to transform the scanning experience into a casual, spa-like visit—complete with amenities like hot tubs and saunas—making whole-body health imaging as approachable as a trip to the sauna, while simultaneously generating a longitudinal health data library to fuel AI development and create a powerful data flywheel.

Midjourney has set extraordinarily ambitious global deployment targets, aiming to scale to 50,000 scanners worldwide by 2031 and achieve a staggering one billion full-body scans per month. This goal implies a high utilization rate, requiring each scanner to perform a scan approximately every 130 seconds continuously, underscoring both the company’s confidence in its technology’s speed—60-second scans with no radiation—and the scale of its vision to democratize whole-body imaging on a planetary scale.

Funded entirely by its existing AI image-generation subscription revenue and free from external investor pressure, Midjourney is leveraging its financial independence to invest boldly in this long-term medical hardware venture. Beyond diagnostics, the company is exploring therapeutic applications of ultrasonic technology, such as non-invasive tumor ablation and targeted peptide delivery, while planning a broader device ecosystem that includes two additional large-scale medical scanners and two consumer wellness devices, signaling a comprehensive market deployment strategy that extends well beyond the initial spa concept.

Midjourney’s ultrasonic scanner represents a significant leap in non-invasive, rapid whole-body imaging by generating high-resolution 3D maps of internal tissues within about 60 seconds, capturing details as small as half a millimeter. Designed for accessibility and patient comfort, the scanner is integrated into spa-like environments—starting with a flagship location in San Francisco—encouraging frequent health monitoring outside traditional clinical settings. This approach leverages AI to interpret the massive data throughput, which reaches 17 gigabytes per second, promising a new paradigm in preventative medicine without the risks of radiation or magnetic exposure.

Despite its groundbreaking design, the Midjourney scanner faces inherent physiological and technical limitations common to ultrasound technology, notably its inability to image through bone or air. This restricts visualization of critical organs such as the brain, lungs, and bowels, limiting its clinical utility in comprehensive diagnostics. Experts like psychiatrist Scott Alexander and cardiologist Eric Topol have expressed skepticism, emphasizing that MRI and CT remain indispensable for imaging these inaccessible regions, underscoring that the scanner is unlikely to replace existing modalities but rather complement them in specific contexts.

While whole-body screening raises concerns about high false-positive rates and uncertain clinical benefits—as noted by radiologists and the American College of Radiology—the Midjourney scanner may carve out niche applications where ultrasound tomography has proven value, such as detecting breast cancer in dense tissue where mammography falls short. Additionally, its rapid scan times and lower operational costs compared to MRI, which averages around $2,000 per scan, position it as a potentially transformative tool for routine health monitoring and early disease detection, especially if scaled effectively.

Midjourney’s data strategy hinges on creating a proprietary data flywheel fueled by frequent, low-cost full-body ultrasonic CT scans, aiming to deploy 50,000 scanners by 2031 to generate up to a billion scans monthly. By licensing FDA-cleared ultrasound chips from Butterfly Network, which already has 150,000 devices in use, Midjourney strategically focuses its innovation on data reconstruction and AI model development rather than hardware invention. This approach mirrors Tesla’s continuous data generation for self-driving AI, positioning Midjourney to build a durable AI company through vast, longitudinal imaging data that can enhance diagnostic accuracy over time.

Despite the ambitious hardware rollout and data accumulation plans, Midjourney currently faces a significant gap in AI diagnostic validation. Founder David Holz candidly acknowledged that, as of mid-2026, the company is not yet using AI to interpret scans, having scanned only about a dozen people without FDA clearance or peer-reviewed validation. This underscores the challenge of translating the massive proprietary data into clinically validated AI diagnostics, a critical hurdle that Midjourney must overcome to realize its vision of accessible, AI-driven whole-body health imaging.

Midjourney's whole-body health imaging technology faces formidable economic and regulatory hurdles that cast doubt on its global scalability. Despite its innovative ambition, the bulky design and logistical impracticality—such as the need to immerse frail patients in a large water-filled tank—contrast sharply with the portability and cost-effectiveness of traditional ultrasound machines, which can be wheeled bedside. Furthermore, regulatory approval remains uncertain given the scanner's limited ability to image organs obscured by bone or air, restricting its clinical utility compared to MRI or CT scans. While the technology could pioneer low-risk mass screening, its economic viability hinges on demonstrating clear clinical benefits and cost reductions relative to established modalities, a threshold not yet crossed.

Midjourney’s ambitious target of deploying 50,000 scanners worldwide by 2031 to achieve a staggering one billion scans per month exposes daunting operational and financial challenges. Each scanner would need to perform a continuous scan every 129.6 seconds—a pace deemed unrealistic given the 60-second scan time per individual. Even doubling the fleet to 100,000 units only relaxes this to one scan every four minutes, still an impractical cadence. This operational bottleneck compounds the immense infrastructure costs, with estimates reaching $100 billion to establish 5,000 spa locations globally, each requiring a $10 million server farm to handle data processing. Such scale raises skepticism about the project’s feasibility, with commentators urging caution to avoid the venture becoming vaporware despite acknowledging Midjourney’s bold attempt to outpace established medical AI players like OpenAI and Anthropic.

By mid-2026, the debate over frequent whole-body ultrasound scanning has crystallized into a clash between tech optimists and cautious medical professionals. Tech advocates argue that "all information is good, and more information can only improve decisions," championing the promise of AI-driven, accessible imaging to revolutionize health outcomes. However, seasoned clinicians counter that decades of oncology research have tempered expectations, revealing that early detection does not universally translate to better prognoses—some cancers remain untreatable regardless of timing, while others respond well even when caught late. This nuanced understanding highlights a disconnect between public enthusiasm and medical reality, underscoring the complexity of translating data quantity into meaningful health benefits.

The clinical value of frequent ultrasound scans is further complicated by the modality’s inherent physical limitations and the high risk of false positives, which can cascade into a costly and stressful cycle of unnecessary follow-ups. Ultrasound’s inability to penetrate bone or air, combined with its failure to distinguish harmful from benign masses, means that even the most advanced imaging cannot reliably pinpoint true health threats. Statistically, a test with 95% accuracy for a rare disease affecting 1 in 1000 people results in 98 out of every 100 positive scans being false alarms, subjecting healthy individuals to invasive procedures and anxiety. This potential for overmedicalization fuels medical professionals’ wariness, as the promise of more data risks becoming a "nightmare" of unintended consequences rather than a straightforward path to improved care.