I used an incredible X-ray machine to look inside my gadgets — let me show you

But Lumafield isn’t building these machines to satisfy our curiosity or to help reverse engineer. Primarily, it rents them to companies that need to dissect their own products to make sure they don’t fail — companies that could never afford the previous generation of industrial CT scanners.

A decade ago, Eduardo Torrealba was a prizewinning engineering student who’d prototyped, crowdfunded, and shipped a soil moisture sensor that ScottsMiracle-Gro eventually took off his hands. (Fun fact: his fellow prizewinners were behind Microsoft’s IllumiRoom and Disney’s Aireal we once featured on The Verge.) Torrealba has been helping people prototype products ever since, both via the Fuse 1 selective laser sintering 3D printer he developed as a director of engineering at Formlabs and as an independent consultant for hardware startups after that.

Throughout, he ran into issues with manufactured parts not turning out properly, and the most compelling solution seemed to be a piece of lab equipment: the computed tomography (CT) scanner, which takes a series of X-ray images, each of which shows one “slice” of an object. Good ones, he says, can cost a million dollars to buy and maintain.

So in 2019, he and his co-founders started Lumafield to democratize and popularize the CT scanner by building its own from scratch. It’s now an 80-person company with $67.5 million in funding and a handful of big-name clients including L’Oréal, Trek Bikes, and Saucony.

“If the only cars that existed were Ferraris, a lot less people would have cars. But if I’m going to the corner store to get a gallon of milk, I don’t need a Ferrari to get there,” he tells The Verge, pitching the Lumafield Neptune as an affordable Honda Civic by comparison.

He admits the Neptune has limitations compared to a traditional CT, like how it doesn’t readily scan objects larger than a bike helmet, doesn’t go down to one micron in resolution, and probably won’t help you dive into, say, individual chips on a circuit board. I found it hard to identify some digital components in my scans.

But so far, Lumafield’s “gallon of milk” is selling scanners to companies that don’t need high resolution — companies that mostly just want to see why their products fail without destroying the evidence. “Really, we compete with cutting things open with a saw,” says Jon Bruner, Lumafield’s director of marketing.

Bruner says that, for most companies, the state of the art is still a band saw — you literally cut products in half. But the saw doesn’t always make sense. Some materials release toxic dust or chemicals when you cut them. Many batteries go up in flames. And it’s harder to see how running impacts a running shoe if you’ve added the impact of slicing it in half. “Plastic packaging, batteries, performance equipment… these are all fields where we’re replacing destructive testing,” Bruner adds.

When L’Oréal found the bottle caps for its Garnier cleansing water were leaking, it turned out that a 100-micron dent in the neck of the bottle was to blame,something the company discovered in its very first Lumafield scan — but that never showed up in traditional tests. Bruner says that’s because the previous method is messy: you “immerse in resin, cut open with a bandsaw, and hope you hit the right area.”

With a CT scanner, there’s no need to cut: you can spin, zoom, and go slice by digital slice to see what’s wrong. Lumafield’s web interface lets you measure distance with just a couple clicks, and the company sells a flaw detection add-on that automatically finds tiny hollow areas in an object — known as porosity; it’s looking for pores — which could potentially turn into cracks down the road.

But only select firms like aerospace contractors and major medical device companies could normally afford such technology. “Tony Fadell said [even Apple] didn’t have a CT scanner until they started working on the iPod nano,” Bruner relates. (Fadell, creator of the Apple iPod and co-founder of Nest, is an investor in Lumafield.)

Torrealba suggests that while you could maybe find a basic industrial CT scanner for $250,000 with $50,000 a year in ongoing software, maintenance, and licensing fees, one equivalent to the Neptune would run $750,000 to $1 million just in upfront costs. Meanwhile, he says, some clients are paying Lumafield just $54,000 a year ($4,500 a month), though many are more like $75,000 a year with a couple of add-ons, such as a lower-power, higher-resolution scanner or a module that can check a part against its original CAD design. Each scanner ships to your office, and the price includes the software and service, unlimited scans, and access for as many employees as you’d like.

How can Lumafield’s CT scanner be that much less expensive? “There’s never been market pressure within the industry to push costs down and make it more accessible,” says Bruner, saying that aircraft manufacturers, for example, have only ever asked for higher-performance machines, not more affordable ones, and that’s where Lumafield finds an opportunity.

Torrealba says there are plenty of other reasons, too — like how the company hired its own PhDs to design and build the scanners from scratch, assembling them at their own facilities in Boston, writing their own software stack, and creating a cloud-based reconstruction pipeline to cut down on the compute they needed to put inside the actual machine.

Even after a pair of interviews, it’s not wholly clear to me just how successful Lumafield has been since it emerged from stealth early last year. Torrealba says the team has shipped more than 10 but fewer than 100 machines — and would only say that the number isn’t 11 or 99, either. They wouldn’t mention the names of any clients that aren’t already listed on their case studies page.