Midjourney's Pivot from AI into Medical Hardware

Nelson Advisors
1 hour ago
11 min read

The Computational, Clinical and Financial Realities of Midjourney Medical: A Strategic Assessment of Whole-Body Ultrasonic Computed Tomography

The entry of Midjourney Inc. into the medical hardware sector represents a significant shift for a company previously known for subscription-based generative artificial intelligence software. Unveiled by Chief Executive Officer David Holz on June 17th 2026, under the newly formed "Midjourney Medical" division, the "Midjourney Scanner" is proposed as a full-body, high-throughput ultrasonic imaging system designed to capture a comprehensive three-dimensional map of internal human anatomy in under 60 seconds. This sudden expansion into medical hardware is part of a broader corporate diversification program.

The scanner is one of eight active initiatives at Midjourney, divided equally between four software projects and four hardware developments. The hardware pipeline is managed by Ahmad Abbas, who joined the company as Head of Hardware in late 2023 after serving as a Hardware Engineering Manager on Apple’s Vision Pro development team. Among these hardware initiatives is a walk-in spatial environment dubbed the "Orb," first announced in August 2024, highlighting the company's interest in physical, sensor-rich user experiences.

Midjourney operates without venture capital or outside equity, self-identifying as a "community-backed research lab". While this corporate structure protects the company from external investor pressure, a pivot into capital-intensive medical manufacturing presents significant financial risks.

These developments are occurring alongside ongoing copyright litigation brought by major entertainment entities, including Walt Disney Company and Warner Bros. Discovery, which could impact the company's financial reserves and long-term capital allocation strategies. By entering the highly regulated medical device market, Midjourney is transitioning from a high-margin software-as-a-service model to a business with complex supply chains, strict compliance demands, and prolonged clinical validation cycles.

Technical Architecture and Acoustic Design

The physical design of the Midjourney Scanner replaces the dry, enclosed bore of standard diagnostic machinery with a vertical, liquid-immersion scanning tank. To initiate a scan, a user stands on a platform illuminated by a golden light pool, which descends into the water at a steady rate of five centimeters (two inches) per second. Water serves as the physical acoustic coupling medium, replacing the localized gels used in traditional ultrasound to ensure continuous wave transmission across the skin.

As the platform lowers the user's body, it passes through a ring containing approximately 500,000 sub-millimetre sensors. These sensors operate as dual-channel acoustic units, dynamically transmitting and recording high-frequency acoustic waves from hundreds of angles. This multi-angle approach captures backscatter, reflection and transmission data through vertical cross-sections of the body.

This sensor array generates massive real-time data streams, with a single second of scan data producing the information equivalent of roughly 500 hours of high-definition streaming video. Processing these raw acoustic signals into a coherent volumetric reconstruction requires over two petaflops of dedicated on-device computational power. The primary transducer hardware relies on semiconductor-based "Ultrasound-on-Chip" technology developed by Butterfly Network (NYSE: BFLY). Traditional ultrasound systems use fragile, expensive piezoelectric ceramic crystals that are physically tuned to narrow frequency bands.

Butterfly's silicon platform integrates thousands of micro-machined acoustic transducers directly onto a complementary metal-oxide-semiconductor (CMOS) chip. This architecture allows a single sensor to electronically emulate linear, curvilinear, and phased array wave profiles. The first-generation Midjourney Scanner prototype integrates 40 of these Butterfly modules, with subsequent models planned to scale this number.

The core reconstruction pipeline does not rely on generative AI models to construct the anatomy, avoiding the risk of synthetic hallucinations in the raw physical data. The tomographic reconstruction uses physics-based signal processing algorithms. Deep learning is used in the post-reconstruction phase. Once the physical acoustic data is resolved into a 3D volumetric map, neural networks perform anatomical segmentation and labeling. This software layer automatically identifies boundaries for fat tissue, muscle groups, skeletal structures, and visceral organs, allowing users to view a labeled AI overlay alongside the raw tomographic data.

Physical and Computational Limitations of Whole Body USCT

While the concept of an acoustic-based whole-body computed tomography scanner offers clear advantages—specifically the absence of ionizing radiation and strong magnetic fields—it faces fundamental physical and mathematical challenges. Chief among these is acoustic attenuation and impedance mismatching at tissue interfaces. Acoustic waves travel through different human media at varying velocities and attenuation rates. In soft tissues, such as muscle, liver, or fat, acoustic waves propagate with relatively low attenuation (ranging from 0.3 to 1.8 dB/cm at 1 MHz). However, when an acoustic wave encounters a boundary with a highly contrasting acoustic impedance, most notably bone or air, the physical behaviour of the wave changes dramatically.

Because cortical bone attenuates acoustic energy at rates up to 50 times greater than soft tissue, and air interfaces reflect sound waves almost completely, traditional diagnostic ultrasound is physically blocked by the skeletal structure and gaseous pockets within the gastrointestinal tract. This physical limitation makes deep brain imaging through the skull, or detailed imaging of organs obscured by bowel gas or the ribcage, highly difficult.

Furthermore, historical attempts to commercialise Ultrasound Computed Tomography (USCT) have been limited to highly homogeneous, non-osseous regions, such as breast tissue screening. In breast imaging, the absence of bone and major gas interfaces allows sound waves to pass through the tissue relatively unimpeded. Extending USCT to the entire human body requires resolving complex non-linear inverse scattering problems.

Standard tomographic reconstruction algorithms assume that waves travel along straight paths, which is a reasonable approximation for X-rays in computed tomography. However, ultrasound waves undergo refraction, diffraction, multiple scattering, and phase shifts as they move through tissue. To generate an accurate, millimetre scale image, the system must solve a highly non-linear, partial differential equation (PDE)-constrained optimisation problem known as Full Waveform Inversion (FWI).

To overcome these computational bottlenecks, Midjourney's hardware relies on substantial local computing power (over two petaflops). The company is also exploring physical neural operators, such as Physics-Informed Neural Networks (PINNs) and Fourier Neural Operators (FNOs), to accelerate these wave propagation calculations.

However, even with advanced mathematical models, physics dictates that standard acoustic energy cannot easily penetrate dense adult bone or deep, gas-obstructed thoracic structures. Consequently, early iterations of the scanner are expected to show high-resolution details in peripheral areas like limbs, joints, and superficial muscle groups, but will likely struggle to match MRI resolution in the deep abdomen, thorax, and pelvic cavities.

Strategic Alliance and Butterfly Network Financial Dynamics

The commercialisation roadmap for the scanner relies on a licensing and co-development agreement with Butterfly Network. This partnership, established under Butterfly's licensing division (formerly known as Octiv and rebranded as Butterfly Embedded), was first disclosed in an SEC Form 8-K filing on November 17th, 2025.

The agreement outlines up to $74 Million in expected milestone and licensing payments to Butterfly over a five-year term, presenting a significant commercial opportunity for the chipmaker. The financial impact of the partnership was realised in the fourth quarter of 2025, where it contributed $6.8 Million in licensing revenue to Butterfly. This cash injection helped drive Butterfly’s total quarterly revenue to $31.5 Million, marking 41% year-over-year growth and supporting the first positive net cash flow quarter in the company's history.

Financial Metric / Period	Q4 2025 Performance	Q1 2026 Performance	FY 2026 Financial Guidance	Midjourney Contract Terms
Total Revenue	$31.5 Million (41% YoY Growth)	$26.53 Million (25% YoY Growth)	$117 Million to $121 Million (20% to 24% Growth)	Up to $74 Million over a 5-year term
U.S. Revenue	$26.8 Million (55% YoY Growth)	$21.4 Million (25% YoY Growth)	Primary driver of growth via Embedded partnerships	Co-development revenue recognized dynamically
Midjourney Revenue Contribution	$6.8 Million (Recognized in Q4)	Major driver of U.S. revenue growth	Implied sustained milestone payments	Tied to licensing and co-development phases
GAAP Gross Margin	67.3% (Driven by high-margin licensing)	68.9% (Upward trend sustained)	Sustained expansion through software/IP licensing	High gross margin profile of Embedded platform
Adjusted EBITDA Loss	$3.2 Million (Improved from $9.1M)	Not Disclosed in detail	Projecting loss of $21 Million to $25 Million	Helps offset core R&D operating expenses
Cash & Cash Equivalents	$150.5 Million	$138.0 Million	Stable liquidity position for platform scaling	Milestone-based execution risk remains

This co-development model helps de-risk Midjourney’s hardware program by utilising an established semiconductor manufacturing supply chain. Butterfly's third-generation handheld probe, the Butterfly iQ3, serves as a validated foundation for the underlying chip design. This allows Midjourney to focus its resources on software, 3D spatial reconstruction, and the physical design of the submersion tank. However, the milestone-based payment structure introduces execution risks for Butterfly, as future revenue is tied to Midjourney hitting specific development, manufacturing scale, and regulatory targets.

The Commercial Playbook: Spas as a Regulatory Sandbox

To address the long timelines and high costs of obtaining medical device approval from the US Food and Drug Administration (FDA), Midjourney has designed a consumer-facing launch strategy that utilizes regulatory pathways for non-diagnostic wellness devices.

The company is branding its physical locations as "Midjourney Spas" rather than clinical imaging clinics. By marketing the scanner's initial output as a "detailed body composition map" (measuring skeletal structures, body fat distribution, and muscle volumes) rather than a clinical diagnostic tool, the company can commercialise the scanner without immediate FDA diagnostic clearance.

Strategic Phase	Timeline	Operational & Technological Focus	Regulatory Context
Phase I: Optimisation	Mid-2026 to Mid-2027	Algorithm fine-tuning, hardware trials, second-generation prototype design.	Internal research; no public deployment.
Phase II: Spa Launch	Late 2027	San Francisco Union Square (25,000 sq ft, 10 scanners, saunas, cold plunges).	Consumer-wellness mapping; bypasses FDA diagnostic pathway.
Phase III: Expansion	2028	Rollout of third-generation scanners with custom silicon across multiple cities.	Accumulation of observational clinical data to support FDA diagnostic filings.
Phase IV: Fleet Scale	By 2031	50,000 scanners globally; target of 1 billion scans per month.	Full therapeutic and diagnostic integration across medical systems.

The first physical facility is planned to open near Union Square in San Francisco in late 2027. The spa is designed as a 25,000-square-foot space housing nine or ten scanners alongside high-end wellness amenities, including hot tubs, saunas, cold plunges, and a gym. This model aims to integrate full-body scanning into a casual wellness routine, encouraging users to undergo regular, repeating scans as a standard part of their self-care and longevity routines.

During this initial consumer phase, Midjourney plans to refine its reconstruction algorithms and gather large-scale observational datasets. In 2028, the company plans to transition to its third-generation scanner, which will introduce custom silicon to improve resolution and image quality. This hardware update is intended to support formal FDA submissions, with the goal of securing clearances that allow the system to perform active medical diagnoses.

The long-term roadmap is highly ambitious, aiming for a global fleet of over 50,000 scanners by 2031 with the capacity to conduct one billion scans per month. Holz claims that widespread, early-stage scanning could eventually prevent up to 30 percent of all global deaths and cut 50 percent of healthcare costs. He also suggests that over a ten-year horizon, these devices could expand from diagnostic imaging into therapeutic applications. This therapeutic capability points toward the potential integration of high-intensity focused ultrasound (HIFU) or micro bubble targeted therapies directly into the submersion bath, transforming the scanner from a diagnostic tool into an active treatment system.

Competitive Analysis of Whole-Body Screening

The consumer-wellness scanning market is divided into two distinct approaches: high-end diagnostic magnetic resonance imaging (MRI) and multi-sensor, non-diagnostic physical mapping. Midjourney Medical plans to position itself at the intersection of these two models, aiming to offer the deep-tissue capabilities of an MRI with the speed and lower cost of a sensor-based wellness scan.

Feature / Dimension	Midjourney Scanner (Midjourney Medical)	Prenuvo Whole-Body MRI	Ezra Proactive MRI	Neko Health Body Scan
Imaging Modality	Ultrasonic CT (Tomographic reconstruction)	Whole-Body MRI (Magnetic Resonance)	Whole-Body MRI & low-dose CT	3D Body Scan, Optical, Infrared, & localized Ultrasound
Physical Mechanism	Water submersion gantry with 500k sensors	Closed magnet bore, dry environment	Closed magnet bore, dry environment	Light-based imaging chamber, dry environment
Session Duration	Under 60 seconds	Approximately 60 minutes	Approximately 60 minutes	10 to 15 minutes
Direct Session Cost	Low-cost positioning (To Be Disclosed)	$999 to $2,499 per scan	$1,350+ per scan	$250 to $350 (£299 / €250-300)
Primary Focus	Volumetric anatomical & body composition maps	Early-stage tumor detection, spinal & organ health	Multi-organ cancer screening & brain health	Skin cancer monitoring & cardiovascular risk factors
Clinical Validation Strategy	Spa-based consumer data loop transitioning to FDA filings	10-year, 100k-participant study (Hercules Research, Boston)	3+ years of clinical observational data collection	Primary care integration, clinical studies (0.2mm skin tracking)
Geographic Footprint	San Francisco launch (Union Square, 2027)	Expanding across USA, Canada, and London	Expanding across major metropolitan areas in the USA	Limited availability (Stockholm and London)

In the premium MRI screening market, companies like Prenuvo and Ezra target high-income individuals and wellness advocates. Backed by high-profile investors and celebrities, these platforms provide detailed scans of internal organs and spinal health.

However, their high pricing ($999 to $2,499 per session) and standard 60-to-90-minute scan times limit their accessibility. To establish clinical validity, Prenuvo launched a 10-year, 100,000-participant research study at the Hercules Research Center in Boston to track how whole-body MRI can predict significant diagnoses in asymptomatic populations. Similarly, Ezra has been compiling observational clinical data for over three years to evaluate the diagnostic value of screening asymptomatic populations.

In contrast, Neko Health, co-founded by Spotify's Daniel Ek, operates at a lower price point, charging between $250 and $350 for a 15-minute exam. Neko Health relies on a dry optical chamber and localized sensors to track skin changes (down to 0.2 millimeters), vascular health, and basic cardiovascular metrics. While highly accessible, Neko Health does not perform whole-body internal cross-sectional imaging, meaning its diagnostic capabilities are primarily limited to superficial skin conditions and basic blood abnormalities.

By utilising Butterfly Network's semiconductor-based ultrasound chips, Midjourney plans to offer cross-sectional, volumetric anatomical imaging at a speed and cost structure that directly challenges these models. The success of this approach depends on whether its 60-second submersion scanner can generate high-quality internal imagery that matches the clinical value of standard MRI systems.

Systemic Clinical Concerns and the Incidentaloma Dilemma

The proposal to deploy thousands of high-resolution whole-body scanners for regular consumer wellness use faces significant skepticism from the clinical community. The American College of Radiology (ACR) has issued clear statements advising against total body screening for asymptomatic individuals, noting that there is currently insufficient evidence to justify the practice. Professional medical organisations argue that there is no clear evidence that routine, whole-body screening is cost-effective or successful in prolonging life.

The primary clinical concern centres on "incidentalomas", benign, asymptomatic abnormalities that are naturally present in healthy human anatomy, such as non-progressive renal cysts, benign liver hemangiomas, or harmless thyroid nodules. High-resolution imaging systems will inevitably identify these harmless variations, initiating a sequence of clinical challenges.

                     [ Asymptomatic Consumer Scan ]
                                   │
                                   ▼
                       [ Incidentaloma Flagged ]
                    (Benign cyst, non-progressive nodule)
                                   │
                    ┌──────────────┴──────────────┐
                    ▼                             ▼
         [ Patient Anxiety ]        [ Secondary Diagnostic Triage ]
       (Psychological stress)     (Blood panels, specialist visits)
                    │                             │
                    └──────────────┬──────────────┘
                                   ▼
                      [ Invasive Follow-up Procedures ]
                         (Needle biopsies, surgeries)
                                   │
                                   ▼
                    [ Risk of Clinical Complications ]
                      (Infections, procedural harm)

Final Thoughts

Because an imaging scan alone cannot reliably differentiate between a benign structural variation and a malignant lesion, flagging an incidental finding often triggers significant patient anxiety. To rule out serious pathology, the patient is referred back into the traditional medical system for secondary diagnostics, such as specialised laboratory testing, contrast-enhanced CT scans, or serial MRIs. In many cases, these findings lead to invasive needle biopsies or exploratory surgeries, which carry real risks of procedural complications, including infections, bleeding, and localised tissue damage.

Critics argue that by offering open access to whole-body scanning in a spa environment, Midjourney's model could generate a high volume of false positives and clinically insignificant findings. Proponents of proactive screening argue that these systems establish individual anatomical baselines, allowing clinicians to detect real pathological changes earlier.

However, from a public health perspective, the downstream costs of investigating incidentalomas could strain existing healthcare infrastructure. Resources could be redirected toward managing benign findings in worried, asymptomatic consumers, potentially reducing the capacity of clinics and hospitals to care for symptomatic patients with acute medical needs.

Nelson Advisors > European MedTech and HealthTech Investment Banking

Nelson Advisors specialise in Mergers and Acquisitions, Partnerships and Investments for Digital Health, HealthTech, Health IT, Consumer HealthTech, Healthcare Cybersecurity, Healthcare AI companies. www.nelsonadvisors.co.uk

Nelson Advisors regularly publish Thought Leadership articles covering market insights, trends, analysis & predictions @ https://www.healthcare.digital

Nelson Advisors publish Europe’s leading HealthTech and MedTech M&A Newsletter every week, subscribe today! https://lnkd.in/e5hTp_xb

Nelson Advisors pride ourselves on our DNA as ‘Founders advising Founders.’ We partner with entrepreneurs, boards and investors to maximise shareholder value and investment returns. www.nelsonadvisors.co.uk

#NelsonAdvisors #HealthTech #DigitalHealth #HealthIT #Cybersecurity #HealthcareAI #ConsumerHealthTech #Mergers #Acquisitions #Partnerships #Growth #Strategy #NHS #UK #Europe #USA #VentureCapital #PrivateEquity #Founders #SeriesA #SeriesB #Founders #SellSide #TechAssets #Fundraising #BuildBuyPartner #GoToMarket #PharmaTech #BioTech #Genomics #MedTech