Nano-X Imaging

Why imaging machines stay expensive, and who pays for it

A classic X-ray tube heats a metal filament until it gives off electrons. It works, but it needs a vacuum, cooling, and ongoing maintenance of a part that wears out. The result is not only a high purchase price but a continuing operating cost. That is why a clinic in a small town or in a developing country cannot justify a machine of its own.

The problem has a second half that people tend to forget: even where there is a machine, someone has to read the scan. A radiologist is a scarce and expensive resource, and the real bottleneck once the number of scans grows. A company that only makes the hardware cheaper produces a pile of images that nobody reads. A company that only develops an algorithm depends on a machine that may not exist where it is needed.

This is what makes Nanox unusual, and also the source of the controversy around it: it promised to attack both halves at once, with a cheaper radiation source and software that reads the result. A double promise like that explains both the unusual attention a relatively small company received, and why every statement it made was examined closely, by investors and by regulators.

Cold cathode: what is really different about the technology

Instead of a heated filament, the Nanox radiation source pulls electrons out of a silicon chip using an electric field, at room temperature. The company calls it Nanox.SOURCE. The "cold cathode" idea removes the need for heating, for cooling, and for the wear that comes with them.

It is important to be precise about what the machine itself does, because this is where mistakes happen: Nanox.ARC is not a CT machine. It is a multi-source digital tomosynthesis system. It takes a series of exposures from different angles, and three-dimensional slices are reconstructed from them. The advantage over a regular two-dimensional X-ray is less overlap between structures that hide each other in a flat image.

The multiple sources are part of the idea. Instead of a mechanical arm that moves around the body, as in CT, an array of fixed sources fires in sequence. This design allows a smaller, more static unit. The second generation, Nanox.ARC X, received marketing clearance in April 2025. According to the company, it is a single integrated unit with a much smaller floor footprint, a one-day installation, and the option to add capabilities through a remote software update, after future regulatory approvals, not before them.

What exactly the FDA cleared, and what it did not say

Nanox.ARC X was cleared on 17 April 2025 through the 510(k) pathway of the U.S. Food and Drug Administration. The clearance covers producing tomographic images for general use (the musculoskeletal system, the lungs, the abdomen and the sinuses) in adults only, and as an addition to regular X-ray imaging, not a replacement for it. The word "addition" is not legal wording: it defines where the tool actually sits in everyday clinical work. The software arm came from an acquisition: in November 2021 Nanox completed its merger with the Israeli company Zebra Medical Vision, in an all-stock deal valued at about $110 million at closing, plus up to $84 million more in stock tied to performance milestones, and the acquired company was renamed Nanox AI.

The software arm, Nanox AI, holds separate clearances of its own. HealthFLD, cleared in February 2024, automatically analyzes the weakening of the CT signal in the liver from routine chest and abdominal scans of people aged 18 to 75, to help identify fatty liver, a finding linked to a fairly common metabolic disease.

It is worth remembering what the 510(k) pathway establishes and what it does not. It establishes that the device is sufficiently equivalent to an existing cleared device, so it can be marketed. It is not proof that using it improves patient outcomes. That is a separate question, tested in clinical studies, and in some areas it is still open.

2023: the U.S. Securities and Exchange Commission sues the company and its founder

On 29 September 2023, the U.S. Securities and Exchange Commission filed a lawsuit against Nanox and its former CEO, Ran Poliakine. According to the complaint, between August 2020 and May 2021 the company and the CEO claimed that Nanox.ARC could be mass-produced at a cost of $8,000 to $12,000 per unit, while higher estimates received from senior people at the company, including senior engineers, were not taken into account.

This estimate appeared in filings submitted to the Commission before the company raised $165 million in its IPO, kept appearing in later financial reports, and was repeated in investor calls and interviews.

The matter ended in a settlement, without admitting or denying the claims: a civil penalty of $650,000 for the company and $150,000 for Poliakine, plus disgorgement of $266,836 from him, with interest. It is worth being precise about the limits of the case: it dealt with what investors were told about the production cost, not with whether the machine works. Poliakine is no longer CEO. The role is held by Erez Meltzer, who also serves as acting chairman.

Where the revenue actually comes from, and it is not the machine

The revenue breakdown the company itself published for the fourth quarter of 2025 tells the story better than any description. Total revenue that quarter was $3.7 million. Of that, $3.1 million came from remote imaging-reading services, about half a million from AI and software solutions, and only $49,000 from selling imaging systems. The imaging-systems line even posted a gross loss of $2.6 million that quarter, meaning the hardware currently costs far more than it brings in.

So almost all the revenue comes from a remote radiology service the company bought, not from the X-ray machine around which the whole story was built. Nanox got the remote-reading service by acquiring USARAD and MDWEB in November 2021, for about $20 million in shares and cash. It is also the only line that makes a gross profit for the company.

For the whole of 2025, the company reported revenue of about $13 million and a net loss of about $75 million. Cash and investments fell to $60 million at the end of 2025, compared with $83.5 million a year earlier, and the company states explicitly that it expects to need additional funding.

36 systems in the field against a target of 360

As of a report from April 2026, about 36 Nanox.ARC systems are at various stages of deployment, and 17 more are expected to be installed in a limited pilot the company calls the Nanox Imaging Network. In parallel, distribution agreements were signed for about 360 systems in the United States over the next two to three years, including an agreement with Howard Technology Solutions to deploy 300 systems over three years.

The company itself adds caveats: the timing depends on regulatory, operational and marketing factors, and most of the installed systems do not yet generate revenue. Its target for 2026 is $35 million, compared with about $13 million in 2025, a gap that shows how much everything depends on the actual pace of activation.

At the same time, the company changed its manufacturing setup: the chip line in Korea was restructured, with a one-time write-off of $17.5 million and a move to production by outside contractors. In November 2025 it bought a medical software company called Vaso Healthcare IT, and today it presents itself as an integrated platform: hardware, algorithms, cloud, remote reading and medical information systems. Clinical work continues, including a collaboration with Cedars-Sinai to examine a solution for aortic valve calcification.