Last June, a British-Brazilian scientific team boarded a bus that had been turned into a makeshift laboratory and headed out to tour six cities across northeastern Brazil.

The researchers were there to find mosquitoes infected with the Zika virus and sequence its genome in their blood, since the evolution of the viral genome contains clues to the epidemic’s origins. But rather than collect insects and send the samples back to a central lab, they’d outfitted the bus with everything they needed to do the research. The most important item: a DNA sequencer about the size and weight of a deck of cards that runs off a laptop USB plug and costs just $1,000.

The instrument, called the MinION, works by pulling DNA through around 500 nanoscopic pores and reading it as it passes through by measuring an electrical signal produced by each nucleotide, or DNA letter. Oxford Nanopore Technologies, a private British company that has spent 12 years and about $200 million developing the invention, foresees its cheap DNA sequencers providing a way to study life in real time (see "50 Smartest Companies.") Every living thing has its own distinctive DNA, and being able to read it provides a powerful tool to identify and profile microbes in great detail. In the last year, the same gadget has seen action in Antarctica to check for life in frozen valleys, on the International Space Station to sequence genes in space for the first time, and deep below the earth in a Welsh coal mine called the “Big Pit.”

A central question for Oxford as a business is whether such portable sequencing is widely useful. Privately valued at around $1.5 billion, Oxford introduced the MinION in 2014, but its sales remain negligible, at around $6 million last year, and the company sometimes gives away the supplies needed to run the instrument. Still, it is betting that users of the device will create new applications, thereby expanding demand for the technology and allowing the company to conquer the broader sequencing market.

Globally, the market for high-speed DNA sequencers and the chemicals to run them is currently about $3 billion a year, according to the consultancy DeciBio—roughly the amount that a single blockbuster drug brings in. The market is dominated by Illumina, a San Diego company whose top-of-the-line machine costs $1 million, takes up as much space as a large filing cabinet, and weighs 498 pounds. It can decode 35 human genomes per week, at high accuracy, for less than $1,000 apiece. In the hands of big academic centers and companies, these machines are powering genetic research into the causes of disease and, increasingly, the quest for new types of cancer diagnostics and prenatal tests.

All those activities might one day be run instead using nanopore sequencing—or at least Oxford hopes so. Right now, the MinION is a good way to identify and study bacteria and viruses. But last year, the first human genomes were sequenced on a MinION as well. (It’s still a cumbersome process and not as inexpensive, however—taking close to $20,000 worth of disposable cartridges at $500 each.)

Three years ago, when Oxford began sending out the first MinIONs to select labs to try out, the devices’ performance was shaky at best. Early models were error prone and often didn’t work at all. But these early mistakes, Oxford says, were the price of a strategy to recruit tech-savvy biologists who have helped it improve the device, create buzz, and explore potential applications. And Oxford has been improving the MinION rapidly—in two years, the device has become far more accurate and the DNA strands now barrel through each pore at 450 letters per second. And there’s headroom for further improvements. “I can see it happening. It’s developing by leaps and bounds,” says Benedict Paten, a bioinformatics expert at the University of Southern California. “I really believe they are about to dominate sequencing.”

For now, the MinION isn’t as fast or accurate as Illumina’s machines. But sequencing typically involves observing DNA with big, costly microscopes and elaborate chemistry. In nanopore sequencing, by contrast, DNA strands are sucked through minuscule ring-shaped biological pores borrowed from the surface of bacteria. As the strings of DNA letters—known as A, G, C, and T—pass through, each generates a faintly shifting electrical signal that’s usually strong enough to let the machine identify it.

This April, Oxford flew many of its key users to London for its annual event, a kind of biological developers’ conference. At the conference, chief technology officer Clive Brown took the stage to alternately boast about and apologize for a parade of half-finished ideas for making the MinION better and creating future products.

Starting with what Brown now calls the “notorious” announcement of the MinION itself back in 2012 (unexpected delays meant it didn’t appear for another year and a half), Oxford has frequently made big promises and then missed deadlines. Twice since 2016 it dodged patent infringement lawsuits brought by Illumina and another competitor, Pacific Biosciences, by changing key aspects of the MinION. In both cases, the updates ended up making the device better, not worse.

“There seems to be a halo around this company,” says its patent lawyer, Martyn Andrews.

MinION can do some unusual tricks. One is to read out extraordinarily long stretches of consecutive DNA letters. Illumina’s instruments read DNA in short fragments of 150 letters. With a MinION, readouts of 10,000 letters are common. Its record: 882,000 DNA letters in a row.

Being able to read long sequences of letters eases the job of digitally reassembling a genome from its parts: imagine a puzzle with thousands of pieces to fit together rather than millions. That’s a crucial factor when exploring an organism’s genome for the first time. It also makes it simpler to study patients’ DNA for certain disorders and cancers caused when genes get duplicated or deleted—what scientists call “structural variation,” as opposed to mutations that affect single letters.

While MinION is mostly suitable for analyzing bacteria and viruses, it’s a second device Oxford is developing, the Promethion, that Brown believes will be Oxford’s “Illumina killer,” because it will be suitable for sequencing humans and other organisms with large genomes. A printer-size instrument that can route DNA through tens of thousands of pores at once, the Promethion will read a trillion DNA letters in a few hours, Brown claims—similar to Illumina’s high-end $1 million sequencing instrument, called the HiSeq X—but won’t cost nearly as much to buy.

Like many of Oxford’s ideas, Promethion is a work in progress. And it’s already well behind schedule. Oxford now plans to start selling it later this year. “Just sit tight,” Brown told scientists at Oxford’s meeting. “The raw horsepower is enough to put you ahead.”

One team in possession of an early model is ZF-Screens, a company in the Netherlands, where scientists hope to decode the genome of the country’s national flower, the tulip. Knowing this commercially valuable flower’s genetic details could speed up the creation of new varieties, which can take as long as 30 years because the tulip’s maturation time is so long. But the plant’s genome is huge—about 10 times the size of a human’s—and so highly repetitive that it can’t easily be decoded using existing sequencing methods. Team member Hans ­Jansen says he has tested the Promethion and “it kind of works and kind of doesn’t work.” The software to capture data is poorly worked out, and the first batch of pore cells arrived broken.

But Jansen plans to stick with it. That’s because he thinks the longer DNA readouts may be the only way to conquer the tulip’s labyrinthine genome. “We need this technology,” he says.