Midjourney’s new Ultrasound Scanner vs. Theranos and Elizabeth Holmes: What’s the Difference?
A few days ago, I saw the Twitter/X discourse about the new Midjourney full-body ultrasound scanner (more info to follow). To say people…
Midjourney’s new Ultrasound Scanner vs. Theranos and Elizabeth Holmes: What’s the Difference?
A few days ago, I saw the Twitter/X discourse about the new Midjourney full-body ultrasound scanner (more info to follow). To say people are split on this topic is an understatement.
I reposted a doctor who changed his mind about the concept, stating that if one could scan themselves weekly, they could view the actual changes over time instead of just counting on one single scan. I responded that it would be “interesting” to see where this scanner goes in the next few months.
“Interesting” is certainly a neutral adjective. I was not claiming it to be the new magical advancement saving millions of lives, nor was I claiming they’re evil and immoral. Just, “interesting.” And the responses I received in return were also undeniably “interesting” as well, some comparing Midjourney’s new scanner to the largescale fraud committed by Elizabeth Holmes and her investor-backed company, Theranos.
Due to this wild conflation of two massively different types of companies and technologies, I wanted to compare and contrast these two cases, and hopefully add more nuance to the conversation than Twitter/X users seem to be allowing themselves.
I will first explain what/who Midjourney, Elizabeth Holmes, and Theranos are, and then dig into the actual debate about Midjourney, why skepticism against their new scanner is ultimately still a completely valid take, but why it diverges from the Theranos comparison.
I hope to help reframe what the debate should be about, instead of focusing on one inaccurate conflation.
Relevant XKCD: https://xkcd.com/1053/
What is Midjourney?
Midjourney is self-funded AI research lab founded in 2022. Its flagship product is a closed-source image generation tool that uses diffusion techniques (it refines images from noise using a prompt), similar to other tools like Stable Diffusion.
Midjourney is completely bootstrapped. This means they do not have outside investors; they are not funded by venture capital. Despite this, they have massive revenue (estimated $200–500mil ARR between 2023–2025), 100+ employees, and are profitable. They sell access to their product, but unlike other research labs like Anthropic or OpenAI, they do not supplement their development with VC funding.
Midjourney did not pivot; they have a new medical division in their company.
What was the scanner Midjourney just announced?
On June 17th, 2026, Midjourney publicly announced their new Midjourney Medical division and their new hardware project. Here are the details of their claimed technology:
- Full-body ultrasound done by submerging the client in water
- No radiation (x-rays) or magnets (MRIs) but still high-quality using “ultrasound tomography”
- Full 3D maps of interior body composition
- Very fast, about 60 seconds
- NOT for public use yet, but estimated to be available in San Francisco in 2027
- NOT FDA-cleared for diagnosis yet (aiming for layering FDA approval on incrementally)
- NOT using generative AI; they say their image reconstruction is classical/computational (i.e. no deep learning/neural networks for reconstruction yet)
Their vision for how this product would be used is the client would go to a spa-type facility instead of a traditionally doctor’s office to complete frequent scans and receive time-series data for preventative health.
Why can’t doctors do this? Their standard technology is way too expensive and resource-intensive to get regular (e.g. weekly) scans.
What makes this different from just getting yearly preventative scans at the doctor’s office? This could provide frequent time-series data, something that is not possible at a large scale right now for healthy patients.
What is the actual status of the product thus far?
As stated, it is not available yet. So far they have been doing longer trials (20 minutes instead of 60 seconds) on a dozen people. The product is not FDA-approved for diagnosis, and they are very transparent about this fact and their goals for later FDA layers added incrementally. That means the first users would be using it at their own risk, without proper diagnosis yet.
Who is Elizabeth Holmes and what is Theranos?
Elizabeth Holmes dropped out of Stanford to found Theranos in 2003, when she was 19. The stated goal is similar to Midjourney — medical testing is expensive, inconvenient, and can be painful, so her stated goal was to improve this. That is where the similarities end.
Her claim was that their product, a machine that could process a single finger prick of blood, could run a plethora of lab tests for a fraction of the time and cost.
The scandal was not necessarily that the tech didn’t work, it’s that she raised over $700mil by misleading investors about the technology and the company’s financial performance. She went to jail for fraud.
Comparing and contrasting Midjourney vs. Theranos
The comparison is not completely unfounded, but the issue is what is at the center of the debate. Here are the similarities and differences in the two cases:
Similarities:
- Silicon valley company using “democratize medicine” arguments to get attention for potential tech products
- Hyping up not-yet-validated “sounds too good to be true” medical tests
- Due to hype, people may confuse what the technology can do today with what the vision for the future is
Differences:
- Theranos took on investors, Midjourney is bootstrapped. Midjourney has no one to commit fraud against (yet). No investors, no consumers yet. They simply announced a product they are still developing.
- Theranos had customers already, Midjourney has none yet.
- Theranos was very closed off about its limitations in an attempt to raise more money, Midjourney has thus far been very transparent about it.
- Midjourney’s proof of concept is a real but early prototype, and they do not claim the current status to be more than that. They explicitly tell the public the current stage of development. Theranos advertised a menu of over 200 tests they claimed they could already perform, but the SEC found its proprietary device could actually run only about a dozen of them and that the vast majority of patient tests were secretly run on modified commercial machines from other manufacturers.
- Midjourney is partnering with hardware manufactured by the Butterfly Network whereas Theranos claimed they were developing their own proprietary hardware; they were instead using commercially available machines for most of the tests they did perform (which itself is not the issue, it was lying about it that was)
- Midjourney’s scanner is not for public use while still in the early stages of development, while Theranos was already claiming to be selling real clinical tests.
In other words, Midjourney’s scanner doesn’t have any outside involvement yet, and they are being transparent about their limitations so far. Contrasting that, Theranos was making false claims about a product they were already selling, lying to investors, and trying to raise more money through all the fraudulent claims.
A federal jury convicted Holmes of intentionally defrauding investors; she was convicted with wire fraud, which requires proof of the intent to deceive.
So what is the real debate?
With the differences between Midjourney and Theranos out of the way, there is still a real, valid, scientifically-grounded debate between physicians on this product that may even lean more skeptical than accepting of this product. Why?
- The claim that the scanner can produce results similar to an MRI or other medical imaging bothers radiologists, because all different types of medical imaging measures different physical properties. Ultrasound has physical limitations that resolution alone cannot fix with current technology. So anyone implying the Midjourney scanner is an actual “replacement” for other medical imaging is making too strong a claim.
- The big one: Even if the technology is highly accurate, screening healthy people can increase health anxiety about issues that never materialize. Everyone has something weird going on in their body (e.g. in a study with full-body MRIs, ~95% of patients had at least one abnormality with <2% actually being cancer); one scan may cause panic about nothing.
- Repeated scanning can amplify false positives; a client is more likely to receive a false positive at least once if they get testing done weekly.
- The technology is has not undergone clinical trials or any type of FDA approval.
- Clients may be more likely to use these scans as a form of diagnosis or self-diagnosis, even with the transparency that the technology cannot produce diagnoses yet.
But the main rebuttal is that with proper use and interpretation, these scans could potentially help clients discover and prevent health problems by correct utilizing the primary novel contribution: frequent time-series data.
The vision the physician I reposted had for this product looks like this: If there is an abnormality in a scan, do not assume something is wrong yet. Get a few more weekly scans. If nothing changes, there may not be an issue, and the client could simply bring it up with their PCP at their next regularly scheduled appointment. If an abnormality is changing or growing in an unexpected way, that is when the client should be concerned and consider scheduling a closer appointment with their PCP.
But these are all “if”s! The Midjourney scanner really is still un-validated medical technology.
The whole point of this article is to inform and reframe where the debate should be. Midjourney has no outside involvement yet, so they are not defrauding investors and lying to customers, because they have no investors or customers yet.
The debate should be in the science of medical imaging, the psychology around medical anxiety for no meaningful improvement to one’s life, and lack of proper interpretation.
The Theranos comparison is inaccurate and a strawman argument made by skeptics who want to be skeptical without digging into the actual nuance of this debate.
Sources
[embed]Preliminary Thoughts On The Midjourney Scanner Edit descriptionwww.astralcodexten.com
Stock image of full-body scan art
메타데이터
- post_id
- 13d482e1baac
- slug
- midjourneys-new-ultrasound-scanner-vs-theranos-and-elizabeth-holmes-what-s-the-difference-13d482e1baac
- url
- https://medium.com/@leahchilders/midjourneys-new-ultrasound-scanner-vs-theranos-and-elizabeth-holmes-what-s-the-difference-13d482e1baac
- canonical_url
- https://medium.com/@leahchilders/midjourneys-new-ultrasound-scanner-vs-theranos-and-elizabeth-holmes-what-s-the-difference-13d482e1baac
- author_url
- https://medium.com/@leahchilders
- status
- ok
- fetched_at
- 2026-07-16 11:36:58
