Description

NNOX is a fraud. My research suggests it is worth approximately $50M versus the current $1.75bn market cap. It is a well-researched fraud. The links to the Citron and Muddy Waters short reports are below. Their reports focus on 1) likely fake customers/distributors. the future order book showing that customers believe in the product is bogus. 2) that strategic partners are mostly bribed by receiving significant stock options. 3) The company was bought for $6.1M at the end of 2018, suggesting there is no value to the IP and the order book. 4) The company is possibly faking the images it claims their machine produces.

https://citronresearch.com/wp-content/uploads/2020/09/NANO-X-a-Complete-Farce-on-the-Market.pdf

https://www.muddywatersresearch.com/research/nnox/mw-is-short-nnox/

I felt that these short reports were missing a detailed critique of the scientific claims and did not do enough to address why a couple reputable scientists are involved. I think NNOX true technology is misunderstood. it is not bogus. I believe they have a working machine. it is just that their current machine is mundane and the one they claim to have on the way will be a dud in my view. 

I hope my analysis adds more evidence and conviction to the short case.

How did the company achieve a $1.725bn valuation

1 – A 15k unit order book – there is significant evidence the order book is fraudulent.

2 – Reputable partners.

Some have some skin in the game. The SK Telecom collaboration is preliminary. It is an agreement to explore a definitive agreement by year end 2020, whereby SK would help make the chips for the machine. There is no definitive collaboration yet. That said, SK Telecom did invest $20M into NNOX and Foxconn invested $26M. so they are taking on real financial risk. However, as detailed by MW, the SK Telecom deal was brokered by one of their employees who was given a big grant of NNOX stock.

Hadassah and USARAD have call options and take zero risk. The partnerships with Hadassah and USARAD are really call options to the partner. Hadassah gets $250k cash upfront, $500k stock upfront and gets a royalty of any revenue if the tech works. It is riskless to Hadassah. Worst case, they get $250k. USARAD will act as a sales lead, recommending the machine to US federal government hospitals in the hopes of selling 3,000 machines.  USARAD will receive a royalty on any revenue generated. NNOX will pay for all development costs. Again, this is a riskless call option to USARAD.

3 – Reputable advisors.

This worried me the most. I wanted to understand why a credible scientist would get involved with this company. I conclude that their presence is not an endorsement. I spoke to colleagues of theirs to understand why they joined. I believe that some advisors are happy to collect a check for little to no work. A couple are interested in working on new ways to commercialize tomosynthesis technology (a cross between X-ray and CT, this is what NNOX is actually working on). I believe the involvement of these advisors began with a genuine curiosity about the commercial use of Sony’s technology in medical imaging. I believe the credible advisors have minimal interaction with the company, barely know the promotional management and do not support the promotional claims.

One key advisor is John Nosta. He was involved since 2012 and owns stock in the company. I feel he is important because he is a well followed tech/healthcare marketer and promotes their stock on Twitter. He has a reputable background but seems like nothing more than a marketer for companies introducing novel technology into healthcare. After reading his public comments my conclusion is that he is well intentioned but has absolutely no understanding of what the company actually does and instead believes in their promotional claims.

What NNOX promotes themselves as doing

NNOX promotes themselves as revolutionizing radiology. They will replace a CT scan with a much cheaper machine and do high quality 3D images of the entire body. This would indeed be revolutionary. I am sure the world would love to replace $300k machines with a $10k machine. The hope is that such a cheap machine can expand the CT TAM massively to poor countries and new use cases where the current cost is prohibitive.

They make it seems as if they have truly revolutionary technology. “100 million nano guns”, MEM, FED technology etc…. in reality their IP is either mundane or pure hope.

What their real technology is and why it has nearly zero value

Their real technology comes in two forms. First, a simple single source 2D X-ray machine. Nothing novel here. Second, a 3D tomosynthesis machine. The value of the company lies in this tomosynthesis machine. They claim it can replace CT images

If you look at the picture of the Nanox.ARC it seems to show about 12 X-ray sources (the little black dots on the arc). My research concludes that this is roughly how the machine works: 10+ X-ray sources moving in an arc across your body. They claim a 360-degree view, but it seems like 180 degrees to me.

Let’s compare that to X-Rays and CT scans. An X-ray is a single source taking a single view from a single point that produces a 2D image. A CT scan is a single source that rotates so fast it takes 1,300 views per rotation and makes 225 rotations per minute. A CT machine must rotate so fast because we move ever so slightly while inside the machine and you need many quick images to piece together what the 3D image should really look like. A CT machine then uses powerful software and processors to reconstruct an accurate 3D image of the body.

What can the NanoX.ARC tomosynthesis machine actually do and what can’t it do

NNOX claimed technology are mostly meaningless. they repeat MEM and FED most often. it says a lot in my opinion that they focus on meaningless terms as the cornerstone of their IP. MEM stands for micro electro mechanical semiconductor cathodes. it is a mature technology. all “MEM” means is that it uses small mechanical parts to make the nano tips that shoot electrons. nothing novel at all and a highly generic unspecified term. FED stands for field emission display technology. this is Sony’s technology that lost out to LCD TVs. FED was used to convert light onto TV screens. it has no obvious connection and commercial application in X-rays, which require much greater energy than visible light. FED uses conventional optical technology and speeds it up to create X-rays. It is one of many ways to create X-rays. I can find no evidence that this technology shows any promise.

What NNOX does is 3D tomosynthesis. it’s important to understand what tomosynthesis is. Think of it as halfway between a 2D X-ray and a 3D CT scan. It's decades old technology and the only current commercial use case is in mammograms. it is quicker and can more accurately detect a mass in the breast.  NNOX is trying to take this proven technology and find a hardware structure to extend the use cases to large parts of the body. the problem is, tomosynthesis only works in low energy areas: small areas, close to the surface. that is why the  only other commercial use case I could find is for dental X-rays. the hope is that tomosynthesis can improve dentists’ ability to detect cavities. the most successful companies I could find are both private and I estimate their private valuation at $20M for one and $100M for the other. They are named Carestream and Surround Medical. the latter is much closer to commercialization than NNOX, suggesting a fair value of perhaps $20-50M.

NNOX claimed technology does differ from traditional tomosynthesis. There are approximately 10 X-rays sources, which pass across the whole body and cover 360 degrees. Tomosynthesis in the breast uses a single X-ray source and covers 15 degrees.  this X-rays source is set in motion in a similar way to the NanoX.ARC technology. But tomosynthesis in mammograms is very different than a CT scan. 1) it collects much less data than CT machines which capture 1,300 images per rotation and 250 rotations per minute. 2) tomosynthesis is much less powerful, it emits fewer X-rays. the result is that the image is of lower quality than traditional CT. however, because breast tissue is near the surface and is not dense, you can get a reasonable quality picture. A mammogram is designed to detect a tumor, all the image has to do is signal a large mass that is different from surrounding matter. Tomosynthesis works well for this very narrow use case. You can see parallels between this use case and dental X-rays.

In contrast to existing commercial tomosynthesis technology, NNOX wants to use approximately 10 X-ray sources. This is far more complex than it seems, and my work suggests the world is very far off from developing this into a commercial technology. When you use 10 X-ray sources, creating an image becomes very complex. If any one of the 10 X-ray sources fail, the image will have blind spots. this unreliability is exacerbated by NNOX choice of technology: millions of little nano tips firing one electron each. Each tip can fail very easily, because they are small and delicate, which can lead to more blind spots on an image. furthermore, the spectrum generated by each X-ray source can be slightly different, resulting in an incoherent image. furthermore, the data analytics required to put together an image from 10 sources is huge. it is a very complex process and there is no evidence they have made a dent in solving the barriers to commercialization. In addition, it is much harder to maintain. instead of worrying about one point of failure you have 10. I think the maintenance costs will be high.

In contrast to CT technology, I believe the NanoX.ARC cannot produce high quality images. 1) NNOX uses too little power. so far, NNOX has only released 2D images for this reason. the 3D images will not be of comparable quality. The images they do show are of 50k volts, that is the level of energy needed for a 2D X-ray. CT requires 120k volts. they have not demonstrated they can surpass X-ray technology yet. 2) NNOX does not use any rotation. the reason a CT rotates is to take many cuts of a body as a way to create a useful image when the patient is moving slightly. Based on my research, without rotation you cannot get high quality 3D images. Because of the lack of rotation, NNOX’s 10 X-ray sources is far less powerful than the single X-ray source in a CT machine. Think of the NanoX.ARC as an arc that has 10 laser beams pointed on the body. it moves up and down the body taking pictures. you cannot get a complete picture whatsoever this way. this is why a CT machine rotates and rotates extremely quickly – to cover every single millimeter or mass in your body and not miss a thing. the NanoX.ARC will miss a lot of information. to get high quality images it will need hundreds of X-ray sources or it will need to rotate. I spent a lot of time talking with physicists working on medical imaging devices to understand why the NanoX.ARC cannot work well. It comes down to basic laws of physics. first, there is the principle of energy conservation. Energy in equals energy out. in the case of a CT scan,99% of the time, an electron that is fired into your body does nothing of value and gets dispersed as heat. 1% of the time it helps create a medical image and here is what happens in this case: The machine creates a lot of energy to fire an electron at super high speed. the electron has to create a photon to create an X-ray that is then used to form an X-ray image. it is very hard to create a photon because most of matter is empty space. it is therefore very hard to get a direct hit. Matter is made up of many atoms. Atoms have a nucleus at the center and then a bunch of electrons orbiting the nucleus just as planets orbit the sun. each atom is 99.9% empty space. to create an X-ray, the electron must arrive at massive force and hit an orbiting electron, which makes up 0.1% of the space an atom takes up. that electron must then escape all the other atoms to get detected as an input to form an image. Most of the time an electron does not hit another electron or it hits one and then bounces around and slowly loses energy in the form of heat before it gets detected. The reason CT machines require so much energy is that producing an image is like sending a laser beam into outer space and hoping to hit a planet – the bulk of the universe is empty space. the odds you can create an X-ray are very low. it is simple physics. this is longwinded but essential to understand: if you use less energy there is basically no way to get the same number of X-rays. less energy = fewer X-rays = lower quality image. Period. When NNOX says they can use 1-10% of the energy a CT scan uses, they are effectively saying they are producing a much lower quality image.

There is a lot of evidence that shows NNOX’s technology is a dud

First, there are many companies and leading universities working on similar technology and they have made little progress and go about solving the problem differently.  in 2018, GE healthcare and Stanford creates a machine with 32 sources. they were able to produce 2D images using the same rotation as a CT machine. I have found several university science papers over the last 6 years that document that their chosen path is to use rotation and start with a 2D image of something very small and simple. The reason they avoid 3D is that would require too much energy. in 5-10 years, they will be lucky to achieve a fraction of what NNOX claims they’ve already achieved. 

Second, the company has spent a cumulative $7M on R&D. A single university study trying to replicate a fraction of what NNOX has done costs over $10M-$20M. There are many precedents suggesting $40-100M is the minimum required. Companies that have some form of commercial technology in CT and tomosynthesis have typically spent upwards of $50M over 10 years. Accuray spends $55M a year. etc…

Third, the company has no meaningful patents. I could find 6 patents for LCD technology. a few vague 5 year old patents about electron emitting technology.  I could find no patents covering their nano tips, none covering image reconstruction, none discussing the anode material and none discussing their detector. My conversations with leading scientists indicate you definitely need patents in all those areas to do anything novel.

Fourth, they provide no details. no 3D CT images (only 2D X-rays), no scientific papers, no details about their tech (just vague terms like MEM and FED. nothing about their conversion process, nothing about the anode material etc…). No details about how they will recreate this very complex image… nothing. Two leading scientists in this field looked at the company’s website and materials for me and could not understand the technology behind the machine. if you go to the website of any rival they provide extremely detailed tech specifications and background info and evidence.

Fifth, they are woefully lacking in relevant scientific expertise. The breakthroughs they claim are bleeding edge technology. All peers employ physicists who come from universities that lead in medical imaging: Stanford, Boulder, Max-Born-Institute of Berlin, etc. NNOX also needs R&D engineers with extensive experience in medical imaging. NNOX needs talented software engineers to solve the data analysis problems. the CEO and CSO (chief science officer) both come from Powermat, that company makes mobile wireless receivers and transmitters. the CSO needs a background in physics. Quite simply, NNOX does not have the team that is capable of discovering a bleeding edge technology.

Sixth, it is absolutely impossible to build a useful machine for $10k. The $10k cost NNOX refers to is for their low end X-ray machine, not a 3D tomosynthesis machine.  it’s clear to me that NNOX is intentionally misleading about this, many have the impression their 3D machine will cost $10k, their prospectus makes it very clear that $10k is for their 2D X-ray machine. Accuray’s machines costs over $1M to make and a CT costs MIN $100k.  Each NNOX X-ray tube costs $5k and the 3D tomosynthesis machine likely uses 10 X-ray sources for a total cost of $50k. if they want to produce high quality 3D images they will need more energy and that means more heat, there is no way around it. the added safety costs to deal with all that heat and X-rays costs MIN $20k. the chips that are used as “nano guns” can indeed cost very little. The caveat is that only happens at massive scale. the initial production cost will be very high. My work concludes that the  bulk of the cost is in the detector and the processing. it’s a huge cost.

Risks

-          Short squeeze. The stock could easily double, it is dominated by day traders. In my experience the borrow cost and availability can swing significantly. If you find it hard to borrow and expensive you might find the opposite in a week etc.

-          The CEO is clearly as promotional as they come. he may have another story to tell that can prop up the stock.

-          I think you can handle these risks through position sizing.

-          I could be wrong but I see no risk of a takeout or true breakthrough technology.

Catalyst

-          Commercialization – I expect their orders will prove fraudulent and never translate into actual deliveries and revenue.

-          In two weeks they will present their machine at a virtual radiology conference. the stock rose 45% on the news as it was viewed as rebutting the short thesis – the company will show their machine to the public. What the short sellers underappreciate is that NNOX has a real machine. it is mundane. the company can show 2D images. they might be able to show some 3D images. But I believe they have no ability to produce a high quality 3D image. That said, the stock is a day trader’s stock and could move significantly up or down on this event, without much relationship to the fundamentals of what is presented.

-          Regulatory or activist short sellers. The company seems like a blatant fraud and promote to me. it should be easy for an activist to get prominent scientists on the record, to doubt that NNOX technology has any merit.