SPECIAL EDITION
Best in Tech
inTech
the way we live decades from now.
50 Smartest Companies
Our picks of the most audacious and ambitious companies,
innovating in the areas of energy storage, gene therapy, virtual
reality, autonomous cars, e-commerce, and more.
Innovators Under 35
Our young innovators are individually tackling the trickiest
technology problems of our time. Collectively they offer us a
sense of how we can solve our greatest challenges.
68 Visionaries
74 Inventors
84 Entrepreneurs
90 Pioneers
98 Humanitarians
Enjoy,
Jason Pontin
CEO, Editor in Chief, and Publisher
MIT Technology Review
GUIDO VIT TI
CEO, Editor in Chief, and Publisher CORPORATE ADVERTISING SALES CONSUMER MARKETING
Jason Pontin Chief Operating Officer VP, Sales SVP, Consumer Revenues and Marketing
Elizabeth Bramson-Boudreau Laurie Hironaka Doreen Adger
EDITORIAL laurie.hironaka@technologyreview.com
VP, Licensing and Communities VP, Consumer Revenues and Marketing
Editor 415-659-2977
Antoinette Matthews Bruce Rhodes
David Rotman
Digital Sales Strategy Manager
VP, International Business
Executive Editor Ken Collina Director of Marketing and Communications
Nicola Crepaldi
Brian Bergstein ken.collina@technologyreview.com David W.M. Sweeney
Assistant to the Editor in Chief
617-475-8004
Deputy Editor Senior Marketing Associate
Giovanna Bortolamedi
Megan Barnett Business Development Managers Julie Swanson
Assistant to the Chief Operating Officer
Senior Editor, Business
Debbie Hanley
Katie McLean debbie.hanley@technologyreview.com
Assistant Consumer Marketing Manager
Nanette Byrnes Katya Hill
Manager of Information Technology 214-282-2727
Senior Editor, MIT News Colby Wheeler
Denise Thayer MIT ENTERPRISE FORUM, INC.
Alice Dragoon Office Manager denise.thayer@technologyreview.com Executive Director
Senior Editor, AI and Robotics Linda Cardinal 617-475-8030 Antoinette Matthews
Will Knight
FINANCE Midwest Director of Chapter Leadership and Process
Senior Editor, Mobile Manager of Accounting and Finance Maureen Elmaleh Gaylee Duncan
Rachel Metz Enejda Xheblati maureen.elmaleh@technologyreview.com
303-975-6381 Director of Communications
Senior Editor, Biomedicine General Ledger Manager Joyce Chen
Antonio Regalado Olivia Male New York, New England, and Southeast
Barry Echavarria Chairman
Senior Editor, News and Commentary Accountant
barry.echavarria@technologyreview.com Jason Pontin
Michael Reilly Letitia Trecartin 603-924-4546
President
San Francisco Bureau Chief Mid-Atlantic
BOARD OF DIRECTORS Kathleen Kennedy
Tom Simonite Clive Bullard
Martin A. Schmidt
Senior Web Producer cbullards@cs.com CUSTOMER SERVICE AND
Judith M. Cole
Kyanna Sutton 845-231-0846 SUBSCRIPTION INQUIRIES
Jerome I. Friedman
Joichi Ito France National: 800-877-5230
Senior Writer
David Talbot Israel Ruiz Philippe Marquezy International: 903-636-1115
David Schmittlein philippe.marquezy@espacequadri.com
Business Editor +33-1-4270-0008 E-mail: technologyreview@
Alan Spoon
Elizabeth Woyke pubservice.com
Germany
Managing Editor PRODUCT DEVELOPMENT
Michael Hanke Web: www.technologyreview.com/
Timothy Maher Chief Digital Officer and michael.hanke@heise.de
VP, Product Development customerservice
+49-511-5352-167
Copy Chief
Erik Pelletier MIT Records: 617-253-8270
Linda Lowenthal China
Product Manager
Vincent Chen (alums only)
News and Commentary Editor Vanessa DeCollibus +86-185-1033-0513
Jamie Condliffe Reprints:
User Interface/Digital Designers
Associate Editors
Japan techreview@wrightsmedia.com
Emily Waggoner Akiyoshi Ojima
Emily Mullin 877-652-5295
Jon Akland ojima@media-jac.co.jp
Mike Orcutt +813-3261-4591 Licensing and permissions:
Engineers
Associate Web Producer Shaun Calhoun licensing@technologyreview.com
Spain and South America
J. Juniper Friedman Molly Frey Cecilia Nicolini
Social Media Editor Jason Lewicki cecilia.nicolini@opinno.com
Julia Sklar DJ Hartman +34607720179
10 Breakthrough
Breakthrough Availability
Immune Engineering. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1-2 years
Precise Gene Editing in Plants. . . . . . . . . . . . . 14 5-10 years
Technologies 2016
Conversational Interfaces. . . . . . . . . . . . . . . . . . . . . . 16 now
Reusable Rockets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 now
Robots That Teach Each Other.. . . . . . . . . . 22 3-5 years
Which of today’s emerging technologies have a DNA App Store. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 this year
chance at solving a big problem and opening up new SolarCity’s Gigafactory. . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 next year
JESSICA SVENDSEN
opportunities? Here are our picks. The 10 on this list Slack. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 now
all reached an impressive milestone in the past year or Tesla Autopilot.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 now
are on the verge of reaching one. Power from the Air. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 2-3 years
Immune
Engineering
Genetically engineered immune cells are saving the lives
of cancer patients. That may be just the start.
By Antonio Regalado
The doctors looking at Layla Richards saw a little girl with leu-
kemia bubbling in her veins. She’d had bags and bags of che-
motherapy and a bone marrow transplant. But the cancer still
thrived. By June 2015, the 12-month-old was desperately ill.
Her parents begged—wasn’t there anything?
There was. In a freezer at her hospital—Great Ormond
Street, in London—sat a vial of white blood cells. The cells had
been genetically altered to hunt and destroy leukemia, but the
hospital hadn’t yet sought permission to test them. They were
the most extensively engineered cells ever proposed as a ther-
apy, with a total of four genetic changes, two of them intro-
duced by the new technique of genome editing.
Soon a doctor from Great Ormond was on the phone to
Cellectis, a biotechnology company with French roots that
is now located on the East Side of Manhattan. The company
owned the cancer treatment, which it had devised using a
gene-editing method called TALENs, a way of making cuts
and fixes to DNA in living cells. “We got a call. The doctors
said, ‘We’ve got a girl who is out of T cells and out of options,’”
André Choulika, the CEO of Cellectis, remembers. “They
wanted one of the vials made during quality-control testing.”
The doctors hoped to make Layla a “special,” a patient who
got the drug outside a clinical trial. It was a gamble, since the
treatment had been tried only in mice. If it failed, the com-
8
8
8
pany’s stock and reputation could tank, and even if it In November 2015, Great Ormond announced
succeeded, the company might get in trouble with reg- that Layla was cured. The British press jumped on the
ulators. “It was saving a life versus the chance of bad heartwarming story of a brave kid and daring doctors.
news,” Choulika says. Accounts splashed on front pages sent Cellectis’s stock
price shooting upward. Two weeks
later, the drug companies Pfizer and
10
10
11
11
Researchers are building on decades of research lem has been friendly fire. So far, easy ways to target
(and several Nobel Prizes involving immunology) only cancer cells are lacking. Lim has founded his own
that worked out many important details, including startup, Cell Design Labs, to commercialize his engi-
how T cells recognize invaders and go in for the kill. neering ideas. He declined to say how much money he
Seen through a microscope, these cells display almost has raised, but he says everyone working with T cells is
animal-like behavior: they crawl, probe, then grab stunned by the kind of money being thrown at the idea.
another cell and shoot it full of toxic granules. “What’s “It’s a ‘wow’ type of situation,” he says.
exciting is they have the ability to move all around;
they’re autonomous,” says Wendell Lim, a synthetic Googling cures
biologist who is also at UCSF. “Immune cells talk to The search to expand immune therapy now involves
other cells, they deliver poisons, they can change what not only the world’s largest drug companies but also
happens in a microenvironment, they have a memory, tech firms. Sharp says that in 2015 Google held two
and they make more of themselves. I think of them as summits at MIT of top immune oncologists and bio-
little robots.” engineers to determine what parts of the problem
Lim is now breaking new ground in what he calls could be “Googlified.” Attendees say the search giant
“synthetic immunology.” This year and last, he pro- paid special attention to new research techniques
duced some futuristic T cells. Tested only in mice so that fingerprint cells from a tumor biopsy in rapid-
far, the cells deploy their targeted search-and-kill fire fashion. These methods might generate big data
behavior only if a specific drug is added—a feature that about what immune system cells are actually doing
could be used to turn the cells on at specific places and inside a tumor, and new clues about how to influence
times, which Lim calls “remote control.” Another T them. So far, Google’s life science unit, named Verily,
cell he designed is a two-stage affair, which kills only hasn’t revealed its plans in cancer immunotherapy.
if it locates not one but two different markers on a But in New York’s Union Square, I met Jeffrey Ham-
cancer cell; it is like a dual authentication method for merbacher, a former Facebook employee who now
the enemy cell. Lim thinks of it as a sensing circuit or runs a lab that is part of Mount Sinai, the hospital
“advanced Google search.” and medical school. With 12 programmers in a light-
Such work is critical because targeting T cells to soaked loft—the nearest thing to blood and guts is a
tumors of the liver, lung, or brain is dangerous, and photo of an exhausted surgeon on the wall—he’s also
some patients have been killed in trials. The prob- spending time on T cells. He’s developing software to
Big Deals
T-cell companies have sought agreements with drug
companies and specialists in gene editing.
August 2012 January 2015 June 2015 November 2015 January 2016 January 2016
Swiss drug giant Novartis buys Biotech firm Drug companies Food maker Nestlé Juno pays $125
Novartis forms a CRISPR gene- Celgene pays Pfizer and Servier pays $120 million million to buy
sweeping alliance editing rights Seattle-based pay Cellectis $40 to a startup named AbVitro, a Boston
with the University from Intellia Juno $1 billion for million for rights Seres for bacteria company that can
of Pennsylvania, Therapeutics. a slice of its T-cell to the first “off pills able to ward sequence the DNA
site of early Juno and Editas treatment portfolio. the shelf” T-cell off infection and inside individual T
successes using Medicine later treatment for immune disorders. cells.
engineered T cells. strike a similar deal leukemia.
for $25 million.
12
12
interpret the DNA sequence in a patient’s cancer Lin says years of scientific work have finally
and predict from it how to goose the response of resulted in a level of mastery that makes therapeu-
killer T cells. tic products seem practical. He thinks the treatments
A clinical trial by Mount Sinai was scheduled to will go beyond leukemia, and beyond cancer. “We
start this year. The patients receive a dose of abnor- think that this fundamental principle, engineering
mal protein fragments that Hammerbacher’s software human cells, could have broad implications,” he says,
predicts will train T cells to attack the cancer. “What “and the immune system will be the most convenient
vehicle for it, because they can
move and migrate and play such
important roles.”
“Where the technology Researchers are already work-
ing on autoimmune disorders, like
stands, it’s a pretty radical diabetes, multiple sclerosis, and
lupus. Infectious disease is also
treatment.” in the sights of T-cell engineers.
Edward Berger, a virologist at the
National Institutes of Health who
helped discover how HIV enters
was fun was that what we submitted to the [U.S. Food human cells, thinks it may be possible to permanently
and Drug Administration] was not a molecule but an keep the virus in check, a so-called “functional cure.”
algorithm,” he says. “It might be one of the first times In February, he says, he will start giving monkeys T
the output of a program is the therapy.” cells genetically programmed to find and destroy any
In January 2016, Juno Therapeutics (see “Biotech’s cell in which the simian version of HIV is replicating.
Coming Cancer Cure,” July/August 2015) paid $125 The actual process isn’t as simple as the theory.
million to acquire AbVitro, a Boston-area company Berger is sure that years of missteps and do-overs lie
that specializes in sequencing the DNA inside single ahead. Also, most protocols involving engineered T
T cells. Now Juno is trying to locate T cells that are cells require patients, or monkeys, to take drugs that
active inside cancers and study their receptors. Juno’s temporarily kill off their own T cells, which isn’t with-
chief scientist, Hyam Levitsky, says an experiment out risks. “Where the technology stands, it’s a pretty
that used to take seven months now takes seven days. radical treatment,” Berger says. “You aren’t going to
And data is piling up: an average experiment gener- use it on a cold sore.” But despite all the progress that
ates 100 gigabytes of information. “A lot of what is has been made treating HIV, a better approach is still
happening is technology-driven,” he says. “The ques- needed. Because the virus hides in the body even after
tions have been there for a while, but there was no treatment, patients have to take antiretroviral drugs
way to get at the answers. Now we’re visualizing them for life. With immune engineering, maybe not. Berger
with new technology in ways we never could before.” sees the chance of a one-time treatment that can hold
the virus in check for good.
Beyond cancer “I was totally inspired by the cancer work,” he says.
In March 2016, Pfizer appointed John Lin to head its “They cured leukemia, and we’ve borrowed it from
San Francisco biotech unit, which develops cancer drugs them. The extension of those ideas for engineering the
and recently started making engineered T cells. He says immune system against other things that ail people
the company had been negotiating with Cellectis well is a major front. I think HIV is the best candidate in
before the news of Layla’s treatment and that no one infectious disease. If you talk to the HIV community,
there was even aware the girl had been treated before it they are crying for a cure—a treatment that, ideally,
hit the news. “The publicity was a big surprise,” he says. you do once and never again.”
13
13
Photographs by RC Rivera
14
Precise Gene
Editing in Plants
CRISPR offers an easy, exact way to alter genes to create
traits such as disease resistance and drought tolerance.
By David Talbot
Breakthrough A new gene-editing method is providing a precise tory process,” Kamoun says. “But the pathogens
The ability to cheaply way to modify crops in hopes of making them yield don’t sit and wait for you; they keep evolving and
and precisely edit more food and resist drought and disease more changing.”
plant genomes with-
effectively. Research in the past year has shown that A version of CRISPR he co-developed paved
out leaving foreign
DNA behind.
the resulting plants have no traces of foreign DNA, the way for recent work on barley and a broccoli-
making it possible that they will not fall under like plant at the John Innes Centre, a plant science
Why It Matters existing regulations governing genetically modified research center also in Norwich. Kamoun and col-
We need to increase organisms and will sidestep many of the consumer leagues showed that the second generation of some
agricultural pro-
concerns over these GMOs. of the edited plants contain none of the foreign
ductivity to feed
the world’s growing The technology is known as CRISPR (see “10 DNA that had been used to create the first genera-
population, which is Breakthrough Technologies 2014: Genome Edit- tion. (Though CRISPR doesn’t require inserting
expected to reach 10 ing”), and plants modified with it are sprouting in foreign genes, it does typically use bits of bacterial
billion by 2050. laboratory greenhouses around the world. Already, genetic material to target the editing.) Meanwhile,
a lab in China has used it to create a fungus- a group at Seoul National University has avoided
Key Players in
Engineering Crops resistant wheat; several groups in China are using leaving any foreign genetic material even in first-
- The Sainsbury the technique on rice in efforts to boost yields; and generation plants.
Laboratory and a group in the U.K. has used it to tweak a gene in Big and small companies alike are jumping in.
John Innes Centre, barley that helps govern seed germination, which DuPont Pioneer has already invested in Caribou
Norwich, U.K.
could aid efforts to produce drought-resistant vari- Biosciences, the CRISPR startup cofounded by
- Seoul National
University eties. Indeed, because it’s so easy to do and the Jennifer Doudna, one of the inventors of the tech-
- University of plants could avoid the lengthy and expensive regu- nology, and is using it in experiments on corn, soy-
Minnesota latory process associated with GMOs, the method beans, wheat, and rice. It hopes to sell seeds bred
- Institute of Genetics is increasingly being used by research labs, small with CRISPR technology in as little as five years.
and Developmental
companies, and public plant breeders unwilling The big question is whether CRISPR crops will
Biology, Beijing
to take on the expense and risks of conventional be governed by the same regulations as GMOs. The
genetic engineering. U.S. Department of Agriculture has already said
The gene-editing technique could be critical some examples of gene-edited corn, potatoes, and
in helping scientists keep up with the constantly soybeans (edited using a different method, known
evolving microbes that attack crops, says Sophien as TALENs) don’t fall under existing regulations.
Kamoun, who leads a research group at the Sains- But both the United States and the more restric-
bury Lab in Norwich, England, that is applying the tive European Union are now conducting reviews
GRANT CORNET T
technology to potatoes, tomatoes, and other crops of today’s regulations. And Chinese authorities have
to fight fungal diseases. “It takes millions of dollars not said whether they will allow the crops to be
and many years of work to go through the regula- planted.
15
16
17
makes voice interfaces more practical and useful. That could offer a glimpse of a graceful future in which there’s less need to
make it easier for anyone to communicate with the machines learn a new interface for every new device.
around us. Baidu is making particularly impressive progress, espe-
“I see speech approaching a point where it could become cially with the accuracy of its voice recognition, and it has the
so reliable that you can just use it and not even think about it,” scale to advance conversational interfaces even further. The
says Andrew Ng, Baidu’s chief scientist and an associate pro- company—founded in 2000 as China’s answer to Google,
fessor at Stanford University. “The best technology is often which is currently blocked there—dominates the country’s
invisible, and as speech recognition becomes more reliable, I domestic search market, with 70 percent of all queries. And it
hope it will disappear into the background.” has evolved into a purveyor of many services, from music and
Voice interfaces have been a dream of technologists (not movie streaming to banking and insurance.
to mention science fiction writers) for many decades. But in A more efficient mobile interface would come as a big help
recent years, thanks to some impressive advances in machine in China. Smartphones are far more common than desktops
learning, voice control has become a lot more practical. or laptops, and yet browsing the Web, sending messages, and
doing other tasks can be painfully slow and frus-
trating. There are thousands of Chinese characters,
The systems offer a glimpse and although a system called Pinyin allows them to
be generated phonetically from Latin ones, many
of a future in which there’s people (especially those over 50) do not know the
system. It’s also common in China to use messaging
less need to learn a new apps such as WeChat to do all sorts of tasks, such as
paying restaurant tabs. And yet in many of China’s
interface for every device. poorer regions, where there is perhaps more oppor-
tunity for the Internet to have big social and eco-
nomic effects, literacy levels are still low.
“It is a challenge and an opportunity,” says Ng,
No longer limited to just a small set of predetermined who was named one of MIT Technology Review’s Innovators
commands, it now works even in a noisy environment like Under 35 in 2008 for his work in AI and robotics at Stanford.
the streets of Beijing or when you’re speaking across a room. “Rather than having to train people used to desktop comput-
Voice-operated virtual assistants such as Apple’s Siri, Micro- ers to new behaviors appropriate for cell phones, many of them
soft’s Cortana, and Google Now come bundled with most can learn the best ways to use a mobile device from the start.”
smartphones, and newer devices, like Amazon’s Alexa, offer Ng believes that voice may soon be reliable enough to be
a simple way to look up information, cue up songs, and build used for interacting with all sorts of devices. Robots or home
shopping lists with your voice. These systems are hardly per- appliances, for example, could be easier to deal with if you
fect, sometimes mishearing and misinterpreting commands could simply talk to them. The company has research teams
in comedic fashion, but they are improving steadily, and they at its headquarters in Beijing and at a facility in Silicon Valley
81%
been driven by the surge of mobile 75%
600 69% 557 80
Internet use in China. 66%
500
61%
420
Mobile Internet users, in millions 450 60
356
Proportion of Internet users on 40% 303
mobile devices
300 233 40
24%
118
150 20
50
18
Mandarin or Cantonese.
mistakes, thus gradually training the system to perform better.
“In the future, I would love for us to be able to talk to all of
engine.
2016, if you were to say ‘Hi’ to your microwave oven, it would
rudely sit there and ignore you.”
19
Reusable Thousands of rockets have flown into space, but not until
2015 did one come down upright on a landing pad, steadily
firing to control its descent, almost as if a movie of its
Rockets launch were being played backward. If this can be done reg-
ularly, spaceflight could become a hundred times cheaper.
Two tech billionaires made it happen. Jeff Bezos’s Blue
Rockets typically are destroyed on their maiden voyage. But Origin first pulled off a landing in November 2015; Elon
now they can make an upright landing and be refueled for Musk’s SpaceX did it the month after. The companies are
another trip, setting the stage for a new era in spaceflight. quite different—Blue Origin hopes to propel tourists in cap-
sules on four-minute rides, while SpaceX already launches
satellites and space station supply missions—but both need
By Brian Bergstein reusable rockets to improve the economics of spaceflight.
Blasting things into space has been expensive because
rockets cost tens of millions of dollars and fly once before
20
21
Robots
That Teach
Each Other
What if robots could figure out more things
on their own and share that knowledge
among themselves?
By Amanda Schaffer
22
23
24
Eventually, robots confronting a crowded shelf will be able to er’s experience. Her collaborator Ashutosh Saxena, then at
“identify the pen in front of them and pick it up,” Tellex says. Cornell, taught his PR2 robot to lift small cups and position
Projects like this are possible because many research robots them on a table. Then, at Brown, Tellex downloaded that
use the same standard framework for programming, known information from the cloud and used it to train her Baxter,
as ROS. Once one machine learns a given task, it can pass which is physically different, to perform the same task in a
the data on to others—and those machines can upload feed- different environment.
back that will in turn refine the instructions given to subse- Such progress might seem incremental now, but in the next
quent machines. Tellex says the data about how to recognize five to 10 years, we can expect to see “an explosion in the abil-
and grasp any given object can be compressed to just five to 10 ity of robots,” says Saxena, now CEO of a startup called Brain
megabytes, about the size of a song in your music library. of Things. As more researchers contribute to and refine cloud-
Tellex was an early partner in a project called RoboBrain, based knowledge, he says, “robots should have access to all the
which demonstrated how one robot could learn from anoth- information they need, at their fingertips.”
25
26
DNA App Store of what it costs other companies. That’s why Helix can afford
its second gambit: to generate and store this type of data for all
customers, even if they initially make only one specific genetic
An online store for information about your genes will query—such as whether they have the sweet tooth gene or a
make it cheap and easy to learn more about your health risk for a certain disease. Maybe two guys in a garage will write
risks and predispositions. a $10 app that shows you how old you’ll look in 10 years, or
which celebrity you are most closely related to. Kao says the
tactic will make genetic information available to consumers “at
By Antonio Regalado an unprecedentedly low entry price.”
The engine to power the app store is being assembled a
mile from Illumina’s San Diego headquarters, in a building
where workmen were still bending sheet metal and laying floor
tiles last January. Several miles of data cables strung through
the ceiling will be connected to a large farm of sequencing
machines, able to process the DNA from a million samples a
While driving and listening to National Public Radio one day, year. Illumina’s CEO, Jay Flatley, also chairman of Helix, has
Justin Kao heard about the discovery of a “sweet tooth gene” said it could be the largest sequencing center anywhere.
that makes you more likely to crave sweets. “Oh my God,” Helix plans to launch the store by 2017. Customers will
thought Kao, who has always loved cookies. “I would pay $5 to control their data by deciding who sees it. There’s even a
know if I had that.” “nuclear button” to erase every A, G, C, and T. But key details
Kao is hoping that millions of other people will be just as are still being sorted out. Will people be able to download their
eager to spend a few bucks for tidbits revealed in their DNA. DNA information and take it elsewhere? Probably, though
He is a cofounder of Helix, a San Francisco–based company they might pay extra for the privilege.
that in 2015 secured more than $100 million in a quest to cre- One company working with Helix is Good Start Genet-
ate the first “app store” for genetic information. ics, a startup in Cambridge, Massachusetts, that offers pre
Our genomes hold information about our health risks, conception testing. These DNA tests tell parents-to-be if they
our physical traits, and whom we’re related to. Yet aside from share a risk for passing on a serious genetic condition, such
ancestry tests that provide a limited genetic snapshot, there’s as cystic fibrosis. Jeffrey Luber, Good Start’s head of business
not a mass market for DNA data. Helix is a bet by Kao’s former development, says it hopes to reach a larger audience with an
employer, the buyout firm Warburg Pincus, and Illumina, the app that can report a few important risks. As with browsing
leading manufacturer of ultrafast DNA sequencing machines, on Amazon, he thinks, people will discover things they “didn’t
that what’s been missing is the right business model. know they needed but that [are] targeted to them, and that
Helix’s idea is to collect a spit sample from anyone who they want.”
buys a DNA app, sequence and analyze the customers’ genes, A looming question mark is the U.S. Food and Drug
and then digitize the findings so they can be accessed by soft- Administration, which has kept close tabs on gene tests and
ware developers who want to sell other apps. Helix calls the will decide how much information Helix apps can reveal.
idea “sequence once, query often.” (The company says cus- Right now, says Keith Stewart, director of the Center for Indi-
tomers will find these apps on websites and possibly in the vidualized Medicine at the Mayo Clinic, most apps that return
Android and Apple app stores.) real medical information—your chance of cancer, say, not just
With its ties to Illumina, Helix thinks it can decode the how much Neanderthal is in your DNA—would need agency
most important part of a person’s genome—all 20,000 genes approval, or at least a doctor in the loop.
and a few other bits—at a cost of about $100, about one-fifth “The bottom line is going to be: What are the regulatory
constraints on information that is truly useful?” says Mirza
Cifric, CEO of Veritas Genomics. His company has been offer-
Breakthrough Why It Matters Key Players in ing since last fall to sequence a person’s entire genome and is
A new business model Your genome Consumer Genomics creating its own app to explore the data, complete with a but-
for DNA sequencing determines a great - Helix
ton to get a FaceTime appointment with a genetic counselor.
that will make genetic deal about you, - Illumina
Cifric hasn’t decided whether to create an app with Helix, but
JAVIER JAÉN
27
SolarCity’s In an industrial park near the shore of Lake Erie, hard by the
Buffalo River, the future of the solar power industry is under
construction. SolarCity’s sprawling Buffalo factory, built and
Gigafactory paid for by the state of New York, is nearing completion and
will soon begin producing some of the most efficient solar pan-
els available commercially. Capable of making 10,000 solar
A $750 million solar facility in Buffalo will produce panels a day, or one gigawatt of solar capacity a year, it will be
a gigawatt of high-efficiency solar panels per year the largest solar manufacturing plant in North America and
and make residential panels far more attractive to one of the biggest in the world.
homeowners. When production begins, SolarCity, already the leading
installer of residential solar panels in the United States, will
become a vertically integrated manufacturer and provider—
By Richard Martin doing everything from making the solar cells to putting them
Photographs by Gus Powell on rooftops. At a time when conventional silicon-based solar
panels from China have never been cheaper, investing in a new
28
SolarCity CEO
Lyndon Rive
Breakthrough
Highly efficient solar
panels made using a
simplified, low-cost
manufacturing process.
Why It Matters
The solar industry
needs cheaper and
more efficient technol-
ogy to be more compet-
itive with fossil fuels.
Key Players in
Photovoltaics
- SolarCity
- SunPower
- Panasonic
29
30
31
The factory is
situated on a former
Republic Steel
manufacturing
site, not far from
downtown Buffalo.
mont, California, had tested at just over 22 percent efficiency. from the Buffalo factory from the first quarter of 2017 to later
Today’s commodity silicon-based solar panels have efficien- in the year.
cies of between 16 and 18 percent. SolarCity competitor But the real risk lies in the rapid advance of solar technol-
SunPower previously led the market with cells that can reach ogy: a record-setting panel today might look relatively inef-
21.5 percent. ficient three or five years down the road. Soon after SolarCity
Efficiency matters because the panels themselves repre- showed off its high-efficiency panels last October, Panasonic
sent only 15 to 20 percent of the cost of the full installation. topped its rival by claiming that its new panels would reach
Much of the rest comes in what’s known as balance-of-system efficiencies of 22.5 percent. Meanwhile, efficiencies in the lab
costs: inverters to connect to the grid, materials to house the are even higher: researchers have made exotic solar-cell mate-
array, nuts and bolts to attach it to the roof, the labor to install rials with efficiencies of up to 40 percent. “I think that within
it, and so on. SolarCity’s installation, says the company, will 10 years, most manufacturers will be producing panels over 20
require one-third fewer panels to produce the same amount of percent efficiency, with the best commercial panels reaching
electricity as conventional installations. “Fewer panels means over 23 percent,” Green says.
fewer bits and pieces, less wire, less days on the roof to install,” O’Sullivan adds: “For now, SolarCity is moving the boat
says Francis O’Sullivan, the director of research and analysis at out as far as it can with, generically speaking, contemporary
the MIT Energy Initiative. technology. But we’re beginning to approach a choke point for
SolarCity uses a deposition manufacturing process that the economics of any silicon-based technology”—including
reduces the number of steps required to make the cells from the new cells SolarCity is bringing online. Future advances, he
two dozen or more to just six. It also replaces silver, one of the says, will entail much lighter, flexible panels that offer much
most expensive elements of conventional solar cells, with less higher efficiencies and are even cheaper to install—and thus
expensive copper. produce electricity at a much lower cost.
But the difference in performance between solar panels At that point, the solar panels coming out of the gigafac-
produced in a small facility like SolarCity’s Fremont plant tory may seem as conventional as commodity panels produced
and in a large factory like the Buffalo one could be signifi- in China today. It is, however, SolarCity’s willingness to take
cant. And scaling up production could be particularly tricky on such risks that makes the Buffalo facility so ambitious. Over
COURTESY OF SOLARCIT Y
given SolarCity’s lack of manufacturing experience. Rive the last 10 years, the Silicon Valley company has made residen-
acknowledges that there could be “small risks around the tial solar a popular choice for many consumers through smart
actual time line” in getting the products coming out of Buf- marketing and attractive financing. Now it wants to transform
falo to match the efficiencies achieved at small scale. Already, solar manufacturing. Whether SolarCity succeeds or fails, it is
SolarCity has pushed back the target date for full production once again pushing the possibilities of solar power.
32
33
TODAY
51%
2008
12%
TODAY
37%
1995
9%
Estimated average number of e-mails sent and received by business users each day
TODAY
122
2011
105
34
Slack
A service built for the era of mobile phones and short text
messages is changing the workplace.
By Lee Gomes
Breakthrough The intra-office messaging system known as Slack School of Management, points out that Slack fun-
Easy-to-use commu- is often described as the fastest-growing workplace nels messages into streams that everyone who
nication software that software the world has ever seen. It surpassed two works together can see. That “allows you to ‘over-
is supplanting e-mail
million daily users less than three years after its hear’ what is going on in an organization, which
as a method of get-
ting work done.
launch in 2013. research has shown can lead to business impact,” he
But what, exactly, makes it so popular? says. “It’s a kind of ambient awareness that you just
Why It Matters Slack gives you a centralized place to commu- don’t get from e-mail.”
In many kinds of nicate with your colleagues through instant mes- Kristina Lerman, a specialist in social com-
workplaces, the
sages and in chat rooms, which can reduce the puting at the Information Sciences Institute at
“water cooler” effect
that lets people over- time you have to spend on e-mail. Whether you’re the University of Southern California, notes that
hear their colleagues’ on a mobile device or a desktop computer, you can Slack messages tend to be short and casual, much
conversations can upload files, get and manipulate information stored more like the mobile text messages that people are
enhance productivity. in spreadsheets or other business applications, and increasingly favoring over e-mail in their personal
easily search through past conversations. But many life. This creates the perception that keeping in
Key Players in
Communication of the core features have been around since the touch with coworkers is effortless. “You get the feel-
Software 1990s. And there have been other “Facebook for the ing that you are quickly responding to everything
- Slack office” software packages that resemble Slack and that is happening around you,” Lerman says.
- Quip have failed to generate anything close to the same In fact, Slack makes it so easy to create mes-
- Hipchat
level of enthusiasm. sages that it might end up placing as many
- Microsoft
The reason for its success lies in part with big demands on people’s time as e-mail traditionally
MIGUEL PORLAN; DATA SOURCES: EMARKETER, GALLUP, RADICATI GROUP
trends: more and more people now get work done has, albeit with a hip and friendly interface. “There
on mobile devices, in collaboration with people who are limits to the amount of time that we have to
aren’t always in the same office at the same time. interact with each other, and Slack doesn’t really
But Slack’s specific design choices have also been cure that,” Lerman says. Software might take some
important. Gerald C. Kane, associate professor of of the friction out of getting work done, but it is
information systems at Boston College’s Carroll still work.
35
36
Tesla Autopilot
The electric-vehicle maker sent its cars
a software update that suddenly made
autonomous driving a reality.
By Ryan Bradley
Photographs by Julian Berman
37
38
will reshape not just the car and our relationship with it but identify objects up to 16 feet away—but also because humans
the road and our entire transportation infrastructure. are awful in traffic. We are bad at estimating distances to
Which is why I jumped at the chance to borrow a car begin with, and we are constantly trying to switch lanes
with Autopilot for a few days and drive it—or let it drive when the next one looks faster, causing accidents in the pro-
me—around Los Angeles. cess. With Autopilot, I no longer had to stare at the bumper
Everyone wanted to know what it felt like, the strange ahead of me, and I could look around to see the variety of
surrender of allowing a car to take control. The only bad decisions drivers make, stopping and starting and stop-
moments that seemed like magic were when the car parked ping again. Meanwhile, my car accelerated and slowed more
itself or changed lanes, mostly because watching a steering smoothly than it ever could have with me in charge.
wheel turn all on its own was unnatural and ghostly. Other With its incremental approach, Tesla stands in con-
than that, I was amazed by how quickly I got used to it, how trast to Google and other companies that have small test
inevitable it began to feel. As a Tesla engineer told me—on fleets gathering data in hopes of someday launching fully
condition of anonymity, because the company won’t let any- autonomous cars. For Tesla, its customers and their partially
one but Musk speak publicly these days—the thing that autonomous cars are a widely distributed test fleet. The
quickly becomes strange is driving a car without Autopilot. hardware required for true autonomy is already in place, so
“You’ll feel like the car is not doing its job,” he said. the transition can play out in software updates. Musk has
The car can’t start in Autopilot; it requires a set of cir- said that could be technically feasible—if not legally so—
cumstances (good data, basically) before you can engage the within two years.
setting. These include clear lane lines, a relatively constant The day after I returned the Tesla, my fiancée and I were
speed, a sense of the cars around you, and a map of the area on an L.A. freeway and saw someone, speeding, cross three
you’re traveling through—roughly in that order. L.A.’s abun- lanes, cutting in front of several drivers. As the traffic stopped,
dant highway traffic is the ideal scenario for Autopilot, not the car behind us came in way too fast and crashed into our
simply because of all the data it makes available to the ultra- bumper, which fell right off. The future, I thought, was practi-
sonic sensors—which use high-frequency sound waves to cally here, and it couldn’t arrive soon enough.
39
40
Washington Shyamnath Gollakota and his colleague vested from the Wi-Fi signals it is backscat-
- Texas Instruments Joshua Smith have proved that weak radio tering, or by feeding on other signals such as
- University of
signals can indeed provide all an Internet TV and radio broadcasts.
Massachusetts,
Amherst gadget needs, using a principle called back- The researchers believe that tiny passive
scattering. Instead of generating original sig- Wi-Fi devices could be extremely cheap to
nals, one of their devices selectively reflects make, perhaps less than a dollar. In tomor-
incoming radio waves to construct a new row’s smart home, security cameras, tem-
signal—a bit like an injured hiker sending perature sensors, and smoke alarms should
an SOS message using the sun and a mirror. never need to have their batteries changed.
41
Not all emerging technologies will alter the business landscape—but some do have the potential to
disrupt the status quo, improve the way people live and work, and enable new levels of innovation.
We can help you make sense of and harness today’s emerging technologies to deliver more value
for your business. pwc.com/digitalandtech
© 2016 PwC. All rights reserved. PwC refers to the PwC network and/or one or more of its member firms, each of which is a separate legal entity. Please see www.pwc.com/structure
for further details. This content is for general information purposes only, and should not be used as a substitute for consultation with professional advisors.
Smartest
Companies
CONTENTS Each year we identify 50 companies that are “smart”
The List......................................................................................................................... 44 in the way they create new opportunities. Some of
23andMe.................................................................................................................. 46
this year’s stars are large companies, like Amazon and
Toyota.............................................................................................................................. 48
Alphabet, that are using digital technologies to redefine
Didi Chuxing..................................................................................................... 54
24M........................................................................................................................................ 56
industries. Others are wrestling with technological
Bosch................................................................................................................................ 60 changes: companies like Microsoft, Bosch, Toyota, and
Intel. Also on the list are ambitious startups like 23andMe,
a pioneer in consumer-accessible DNA testing; 24M,
a reinventor of battery technology; and Didi Chuxing,
a four-year-old ride-hailing app that’s dominating the
Chinese market. Still, despite the excitement of recent
advances in such fields as artificial intelligence and
genomic medicine, technology has failed to energize the
overall economy. In our opening essay, we explore why
that is so and what needs to change.
4343
The List
14. ENLITIC
A number of Australian radiologists are now
using the company’s deep-learning software to
analyze x-rays.
Enlitic claims its algorithm read chest CT images
Over the past year, these 50 companies have best 50 percent more accurately than experts in its
own test.
combined innovative technology with an effective
business model. The list is the product not of a formula 15. FACEBOOK
Its Oculus Rift technology is the first truly
but of our editors’ judgment. high-quality virtual-reality headset for
consumers.
Rift sells for $599.
44
45
23andMe
The consumer genetic-testing startup has amassed one of the world’s
largest databases of DNA. Now it is sifting through it for clues to new drugs.
By Antonio Regalado
Welcome to You.” So says the the privilege. That idea so appeals to The company almost didn’t survive
genetic test kit that 23andMe investors that they have valued the still- to build its database. In 2013, the U.S.
will send to your home. Pay unprofitable company at over $1 bil- government forced 23andMe’s flag-
$199, spit in a tube, and several weeks lion. “Money follows data,” says Barbara ship health test off the market when
later you’ll get a peek into your DNA. Evans, a legal scholar at the Univer- it charged, in one of the angriest let-
Have you got the gene for blond hair? sity of Houston, who studies personal ters the Food and Drug Administration
Which of 36 disease risks could you pass genetics. “It takes a lot of labor and cap- has ever sent to a private company, that
to a child? ital to get that information in a form the company’s gene predictions were
Run by entrepreneur Anne Wojcicki, that is useful.” inaccurate and dangerous for those
the ex-wife of Google founder Sergey
Brin, and until 2015 housed alongside
the Googleplex, the company created a GENETIC BIG DATA
test that has been attacked by regula- Despite a government ban on its flagship health product,
tors and embraced by a curious public. 23andMe has used ancestry tests and consumer surveys to build
It remains, nine years after its intro- up one of the world’s largest troves of DNA and personal data.
duction, the only one of its kind sold
directly to consumers. 23andMe has 1 million customers
managed to amass a collection of DNA FDA crackdown
information about 1.2 million people, 800,000 bans 23andMe
which last year began to prove its value from marketing
when the company revealed it had sold health test
600,000
access to the data to more than 13 drug
DAVID BISKUP; DATA FROM 23ANDME
46
who might not fully understand the nology captures the big picture of
results. HOW IT WORKS which genes a person has. It allows
Wojcicki apologized and contin- People who sign up for 23andMe submit 23andMe to tell you, for instance,
ued offering more limited ancestry a spit sample. The DNA in stray cheek that your eyes are probably blue
tests. But she never really changed skin cells is analyzed for some 650,000 rather than brown.
genetic markers. These markers reveal
her idea. By last fall, the govern- To gain the volume of infor-
which common version of each human
ment agreed to allow some health gene a person has, about 20,000 genes
mation necessary to study specific
information back on the market— in all. Such “genotypes” may explain diseases, 23andMe has recruited
for example, letting customers know many physical traits and disease risks, patients by giving the test away for
whether they’re carriers of risk although not all. free. One person who joined the
genes like the one that causes cys- database is Amy Caron, who was
tic fibrosis. Wojcicki has vowed she WHAT YOU LEARN diagnosed with lupus, an autoim-
“will not sleep” until the full results Inherited Disease Risk
mune disorder, at age 22. Caron
(which once included estimates of a Thirty-six genes that put your children at risk agreed to submit her DNA as part of
person’s risk for diabetes, macular for inherited disease, including: a study of lupus financed by Pfizer.
degeneration, and breast cancer) are Very little is known about the dis-
- Niemann-Pick - Usher syndrome
available again. disease ease, and filling out surveys “is a
- Sickle-cell anemia
To some, 23andMe’s strategy is - Cystic fibrosis safe, low-risk way to get involved
- Bloom syndrome
controversial for the way it treats - Tay-Sachs disease and contribute,” she says.
personal data as a commodity. But In spring 2016, 23andMe
“prescient” may be a better word. Traits also opened a drug lab, where it
Twenty-two genes that explain your
Even the U.S. government is catch- will begin testing some of its own
appearance or characteristics, including:
ing up. President Obama’s Precision treatment ideas. It’s the first time
Medicine Initiative began inviting - Cheek dimples - Asparagus-odor the company has done work at a
- Cleft chin detection
citizens to join its own one-million- lab bench rather than a computer
- Unibrow - Bald spot
strong database this year. And just screen, says Joyce Tung, the com-
- Earlobe type - Bitter-taste
like 23andMe, it must find ways to perception pany’s vice president of research. To
entice the public to join. - Widow’s peak some observers, finding drugs is the
For now, though, 23andMe’s bio- only way 23andMe can justify the
Wellness
bank is the world’s largest repository Six genes that reveal differences related value investors have given it, since
of DNA samples that also contains to food, exercise, and sleep, including: the company has never turned a
extensive health information, will- profit from its tests.
- Sensitivity to - Lactose
ingly provided by customers who alcohol intolerance Another reason 23andMe can’t
answer survey questions like “Do - Preference for - Muscle stand still is that genetic technol-
you like cilantro?” and “Have you caffeine composition ogy keeps advancing, and it keeps
ever had cancer?” 23andMe says its getting cheaper. That means lots of
Ancestry
customers supply it with as many companies are offering low-priced
The overall composition of your genes
as two million new facts each week. reveals a person’s ancestry, including: gene tests. One even promises to
These surveys are proving valuable fully decode a person’s genome for
- Countries of origin - Percentage of
to drug investigators. This year the Neanderthal $1,000. Yet unlike 23andMe’s test,
- Relatives who
company found genetic variations share DNA genes these must be ordered by a doctor, in
strongly linked to whether custom- order to avoid regulations covering
ers consider themselves early risers, Banned in U.S. direct-to-consumer medicine.
offering a clue about how to develop The U.S. still bars certain genetic findings Wojcicki still believes the pub-
from being provided directly to consumers,
drugs that modulate alertness. lic is able to deal with the sort of
including:
When it receives a spit sample, complex information that can be
23andMe examines about 650,000 - BRCA breast - Risk of gleaned from DNA “without a mid-
cancer gene Parkinson’s
locations in its customer’s genomes. disease dleman in a white coat delivering it,”
- Blood-thinner
That’s not as detailed (or expensive) sensitivity - Response to as she recently told the Wall Street
as generating a complete, letter-by- - Risk of Alzheimer’s hepatitis C Journal. Instead, it’s 23andMe that’s
letter genome map. Yet the tech- disease treatment in the middle.
47
Toyota
The world’s largest automaker is finally getting serious
about self-driving technologies.
By George Anders
MICHAEL KIRKHAM
48
oyota spends $10 billion a year develop that technology, Toyota lags Determined to make up for lost
T on research, more than any behind several of its fellow carmak- time, Toyota’s 60-year-old CEO, Akio
other automaker except Volks- ers and Silicon Valley upstarts such as Toyoda, is spending $1 billion for a new
wagen. That pays for endless incremen- Google and Tesla Motors. It’s possible Toyota Research Institute with offices
tal improvements in everything from that a generation from now everything in Michigan, Silicon Valley, and Cam-
lithium batteries to seatbelt design, but from roadway design to driver certifi- bridge, Massachusetts, that will focus
such tweaks may not be enough any- cation will be radically reshaped by the on autonomous cars and robotics. He
more if Toyota is to remain the world’s ubiquity of semi- or fully autonomous has recruited Gill Pratt, a top robotics
top seller of cars. vehicles, and carmakers without the req- researcher, to run the institute, giving
The development of autonomous uisite technology will be as imperiled as him authority to hire hundreds of engi-
vehicles now threatens to change the the sellers of silver-halide film in the age neers and scientists. At the same time,
very essence of driving. In the race to of the digital selfie. Toyota is striking up partnerships with
49
Toyota is testing
a “Teammate
Concept” car that
could do some
highway driving on
its own by 2020.
Stanford, the University of Michigan, looking cameras that could help assess ware to anticipate trouble that could
and MIT to rethink cars’ capabilities, drivers’ alertness. If drivers get drowsy emerge from an obscured side street, a
even if provocative new approaches or stop paying attention to the road, wobbling bicycle, or an angry motorist
might take a decade or longer to show then automated safety systems could who is switching lanes in a dangerous
up in dealer showrooms. help keep the car safe while nudging the way? Teams of Stanford researchers are
It’s clear that Toyota, like most driver to get back on task. testing out approaches. One initiative,
established carmakers, isn’t making As cars take on more and more tasks, led by John Duchi, an assistant profes-
an all-out bid to match Google’s efforts just how smart can they get? Fei-Fei Li, sor of statistics and electrical engineer-
to build fully autonomous vehicles. a Stanford computer science professor ing, is starting with known hazards,
Instead, Toyota envisions drivers and who is heading her department’s $25 such as erratic bicyclists, and then try-
software sharing control for years to million alliance with Toyota, says her ing to build prediction software that
come. Pratt is championing “guardian team is applying a wide range of AI tech- could make smart decisions in similar
angel” technology that could find the niques to driving-related challenges. As situations. Another team, led by Li, is
best evasive strategies in an instant if she cheerfully acknowledges, “Our work relying on 3-D vision and pattern rec-
trouble looms. might be relevant to the cars of 2018, or ognition to identify high-risk group-
Similarly, artificial-intelligence 2028, or anywhere in between.” ings. These include pedestrians staring
researchers at Stanford who are work- One area of interest: defensive driv- at smartphone screens, or children play-
ing with Toyota are testing out inward- ing. Is it possible to teach a car’s soft- ing catch near a roadside. Put a human
driver behind the wheel, and it’s easy to
distinguish an alert pedestrian from a
50
tify objects in photos, even if they were selves, the automaker turned down an 33,000 people. As much as 90 per-
in odd poses or in front of confusing offer from Google to coöperate on the cent of those accidents can be traced
backgrounds. Now she is building technology because it was reluctant to human error. Still, it’s unlikely that
on those techniques to analyze road- to share manufacturing know-how. data-rich companies will want to relin-
way photos. Her goal: to ensure that Google has instead taken steps toward quish control of their own hard-won
a car’s software can detect the differ- an alliance with Ford. Even as recently knowledge that—for now, at least—
ence between a pedestrian at the curb as 2014, CEO Toyoda said he wasn’t provides them with a competitive
making eye contact with drivers and a inclined to take autonomous technology edge.
pedestrian listening to music through seriously until a self-driving car could Raj Rajkumar, co-director of the
earbuds and gazing downward at a beat the best humans in a 24-hour test General Motors/Carnegie Mellon
smartphone. on a top German racetrack. autonomous-driving lab, puts GM, Nis-
Li’s group is also creating ways that By now, Google and several auto- san, and the German Big Three (Daim-
self-driving cars can share informa- makers have built up large stockpiles ler, VW/Audi, and BMW) in the upper
tion as instantaneously as possible. On of video and sensor data from years of echelon of global car companies mov-
highways, for example, it should be testing autonomous cars and selling ing toward some degree of autonomous
possible to safely compress the spac- models with some autonomous fea- driving. “Toyota seems to be lagging
ing between cars, helping traffic flow tures, such as lane-departure warnings behind,” he says. “But with the creation
more smoothly, as long as vehicles at and blind-spot detection. Having less of its research institute, it might catch
the back of a convoy can be apprised information to feed into their machine- up quickly.” One of Pratt’s first hires
of any surprises that the front car has learning systems could put Toyota’s soft- at the institute was a former Google
already identified. Even in city traf- ware researchers at a disadvantage. robotics director, James Kuffner, as
fic, crashes could be avoided if cars Both Pratt and Li have been calling chief technology officer.
could instantly communicate with one for car companies to share data from In the university partnerships,
another about hazards that might be autonomous vehicles in the belief that Toyota isn’t the only car company that
invisible to one vehicle but easily rec- pooled knowledge will help all com- could benefit; other manufacturers will
ognizable from a different perspective. petitors make faster progress and gain eventually be able to consult published
public trust. After all, self-driving cars findings, Li says. Still, while the pieces
Getting data could be far safer. More than six million come together, Toyota will enjoy special
In 2012, when Google was testing Toy- motor-vehicle crashes take place each access and collaboration. And car tech-
otas that it modified to drive them- year in the United States, killing about nology’s long adoption curve may give
Toyota the time it needs. Lots of battles
still lie ahead in winning regulatory
approval and customer loyalty, regard-
less of whose early technology is most
promising. History shows that break-
through technologies such as airbags
and advanced transmissions can take
20 years to gain mass acceptance after
their marketplace debut.
Toyota also figures to move faster
now that its boss has repudiated his ear-
lier doubts. In May the CEO urged all
his employees to embrace the “momen-
tous change” associated with auto-
mated driving and robotics. Such new
technologies will be as transformative
to the company, Toyoda vowed, as was
his own grandfather’s decision in 1930
Inside the Teammate car. Toyota and its
research partners are trying to make driving
to create a motor-car division within
safer and more pleasant, but huge challenges what was then a small loom-making
in artificial intelligence have to be worked out. company.
51
Didi Chuxing
Jean Liu explains how her ride-hailing company bested Uber in China
and why data is its biggest asset.
By Christina Larson
54
In May, Apple announced that it was investing $1 billion in What role has data played in developing
your new services?
Didi Chuxing, China’s leading car-hailing service, shining a
When passengers want to go, they want
spotlight on the Beijing-based startup. Didi, which arranges to go in five minutes. We need to be fast
16 million rides a day in cities across mainland China, was and efficient. Analyzing that data, we
facing a rivalry with San Francisco–based Uber at the time. have a very good idea, in a particular
city, what demand will look like in 10
But Didi won the battle easily. Over the summer it acquired
minutes. Even before the 6 p.m. rush
the Chinese operations of Uber (which reportedly had been hour, we can dispatch drivers in partic-
losing $1 billion a year in its unsuccessful effort to get drivers ular directions. We can predict at 6 p.m.
and riders to use its service instead of Didi’s). In an interview on a Wednesday how many people will
[hail cars] from a particular workplace.
with writer Christina Larson, Didi’s president, Jean Liu,
The only way you can match the sup-
attributed her company’s edge to its use of data. In its four ply with the demand is to do intelligent
years of operation, the company has gathered information like dispatching and demand prediction.
common pickup points and destinations, peak demand times,
Uber has been known, in part, for fric-
and frequent routes in 400 different Chinese cities. It’s used
tion with its drivers. Does Didi face simi-
the data for predictive analysis and to create new products lar challenges or r esentments?
such as Didi Bus, a bus-booking service that’s become a We provide business opportunities to
popular alternative to crowded public buses. 14 million registered drivers. If you
count just the drivers whose major-
ity income comes from Didi, that’s 2.5
million on a monthly basis. We actually
help drivers increase their income by
increasing the efficiency of their routes.
Why did you come to join Didi two years cient way to find passengers]; private In Beijing, a lot of the private car driv-
ago from Goldman Sachs, where you car services [like the Uber model] with ers [who find customers via Didi’s app]
were a managing director? higher-, middle-, and lower-end [vehi- earn four times the minimum wage.
I was born and raised in Beijing. I love cles and prices]; a bus service. That resolves the fundamental issue
the city, and I also get stuck [in traffic] for the drivers.
in the city. I studied computer science
for my bachelor’s and master’s degrees, What are your biggest challenges?
[but] I went into banking after gradua-
tion. I stayed in finance for about a
14 million At this stage, we spend a lot of time
talking about how to recruit and retain
dozen years. My last investment oppor- Number of registered top talent.
tunity was Didi. Didi drivers
What’s the story behind the recent
How has the company changed in those Apple investment, and how will Didi use
two years? The bus is like an expanded carpool the money?
When I joined, there was only one busi- shuttle service: instead of taking a pub- Both of us [Apple and Didi] are invested
ness line—that was the taxi [service]. lic bus with many stops, and maybe no in technology; it just seemed very intui-
But the opportunity is so much bigger seats left, and uncomfortable, we offer tive. Both have a big overlap in our cus-
than just taxi-hailing. It’s really a world- a shuttle-like service with typically just tomers in China. We have some ideas.
class dilemma—how to move around one or two stops. All the seats on the bus But bear in mind, this is a speed date:
800 million urban Chinese. are prebooked. We can [use our years of we got to know them in late April. A lot
China has a lot of urban density, data] to determine popular origins and of things are still in discussion. Going
but the public transit system really lags destinations, where commuters are basi- forward, maybe we could use voice
behind. Today our product line includes cally making the same journey in the function [technology] from Apple; a
ADAM DEAN
hailing taxis [working with existing taxi morning. With the scale of this network, big percentage of our users are using
companies to give drivers a more effi- we can pull people together. iPhones.
55
56
24M
The startup’s cheaper way to
make lithium-ion batteries could
make it cost-effective to store
energy from renewable sources.
By Elizabeth Woyke
Photographs by Adam DeTour
57
58
5 6 7
9 Batteries wait
9
before and after
testing. It only
takes a few hours
to go from raw
materials to bat-
teries ready for
testing, accord-
ing to 24M. In
a conventional
lithium-ion fac-
tory that process
would take about
a week. 7
59
Bosch
An old-school manufacturer is building smart
factories to remain globally competitive.
By Russ Juskalian
Photographs by Laetitia Vancon
60
61
62
63
11 12
64
9 10
9 The interior of in parts delivery and production will Two challenges loom over Bosch’s
manufacturing machines on increasingly move around on their own smart-factory project and these broader
the assembly line.
and interact in close physical proximity ambitions. The first is finding enough
10 Ovens treat metal injection- with human workers. workers with the skills to run increas-
molded parts with heat. “What we will learn in inner-city ingly data-driven manufacturing sys-
11 Goods are automatically traffic we can also realize … for inter- tems. The second will be creating
scanned when they pass logistics in a factory,” says Assmann. industry standards so such systems will
this RFID gate. With knowledge derived by adding sen- be able to work together.
12 A board shows current sor technologies to cars and collecting The financial stakes are far too high
performance on the shop the resulting data, “we can make robots to let either of these obstacles get in
floor. have eyes, ears, and feelings.” the way of progress, however. In Ger-
Assmann has ambitions for this data many, labor and energy costs are up, and
even beyond Bosch’s internal applica- increased automation and efficiency are
tions. Increasingly, he says, the company how businesses like Bosch will stay prof-
will sell its know-how in logistics, data itable in the future. As Assmann puts
processing, and manufacturing as a ser- it, embracing connected industry isn’t
vice to others. optional—“it is a must.”
65
35
Innovators
under 35
The people in our 16th annual celebration of working in big companies, or doing research in
young innovators are disrupters and dreamers. academic labs—they all are poised to be leaders in
They’re inquisitive and persistent, inspired and their fields.
inspiring. No matter whether they’re pursuing Hundreds of people were nominated for this
medical breakthroughs, refashioning energy group. After MIT Technology Review’s editors
technologies, making computers more useful, narrowed the list, outside judges evaluated the
or engineering cooler electronic devices—and quality and potential impact of the finalists’ work,
regardless of whether they are heading startups, guiding the selections you’ll find here.
Judges
Polina Anikeeva John Dabiri Rana el Kaliouby Rachel Sheinbein
Assistant Professor of Materials Professor of Civil & Environmental CEO, Affectiva Managing Director,
Science and Engineering, MIT Engineering and Mechanical Makeda Capital
Engineering, Stanford University Jennifer Lewis
Zhenan Bao Professor of Biologically Inspired Leila Takayama
Professor of Chemical David Fattal Engineering, Harvard University Acting Associate Professor of
Engineering, Stanford University CEO, Leia Psychology, UC Santa Cruz
Hao Li
Emily Cole Tanuja Ganu CEO, Pinscreen; Assistant Jennifer West
Chief Science Officer, Liquid Light Cofounder, DataGlen Professor of Computer Science, Professor of Engineering,
JUSTIN TSUCALAS, RC RIVERA, DAMIEN MALONEY, DAMIEN MALONEY, TIMOTHY O’CONNELL
68
Visionaries
By looking at things a bit differently from everyone else,
they find powerful new uses of technology.
Featuring
Jean Yang needs to access—such as your “Just like there cally. “It is a double hull for
location—doesn’t slip out onto information leaks,” Yang says.
Carnegie Mellon University
the Internet. Needless to say, are many ways She has uploaded the
Why don’t computers they sometimes fail, leaving to sink a boat, code to open-source libraries
keep our personal data our data to be exploited by for anyone to use. And this
secure by default? hackers. “Just like there are
there are many fall she begins as an assistant
many ways to sink a boat,” says ways to leak professor of computer science
When programmers create a Jean Yang, “there are many information.” at Carnegie Mellon, where
feature for an app or a website, ways to leak information.” she can try to get her ideas to
even something as simple as That’s why Yang created necessarily have to scrub per- spread further. “Giving people
a calendar, they should code Jeeves, a programming lan- sonal information from their tools to create technology is
in protections so the personal guage with privacy baked in. features, because Yang’s code incredibly empowering,” she
information that the feature With Jeeves, developers don’t essentially does it automati- says. —Patrick Doyle
At the center of Snapchat—the disappearing- send them to your friends. Snapchat, Spiegel
photo social network valued at $20 billion, used has said, is based on the idea that “ephemeral
by 150 million people—sits an exotic-car-driving, should be the default.”
engaged-to-a-supermodel 26-year-old genius. In its six years of existence—an epoch in
Or jerk. Or both—it’s hard to tell. Evan Spiegel startup time—the company has outlasted rivals
is kind of a recluse. The guy behind this new like Poke and Ansa and Gryphn and Vidburn
media empire follows only about 50 people on and Clipchat and Efemr (I swear I’m not mak-
the mobile app he helped create. (One of them ing these up) and Wink and Blink and Frankly
is the magician David Blaine.) He declined to and (I promise you) Burn Note and Glimpse
speak to me, which is fitting, because what and Wickr. It reaches 41 percent of U.S. 18-
Snapchat is, what Spiegel understands bet- to 34-year-olds every day and generates
Evan Spiegel ter than anyone, might be the opposite of revenue from media companies and adver-
Snapchat an interview with a magazine. tisers that publish snaps in dedicated chan-
The cofounder of Snapchat is often compared to Facebook, nels. What did Snapchat do right that others
Snapchat figured out and Spiegel to Mark Zuckerberg. Which makes didn’t? One thing you immediately notice upon
that people wanted sense, especially since Facebook tried to buy downloading the app is how much it requires of
LEONARD GRECO; STEVE JENNINGS/GET T Y
something different from Snapchat for $3 billion before releasing its own you. You can’t just sit back and watch—you, too,
social media. knockoff versions that promptly fell into irrel- must snap. The home screen practically begs
evance. And both founders are college drop- you to take a picture or shoot a video. Photog-
outs (Spiegel from Stanford, Zuckerberg from raphy once was all about capturing a moment
Harvard). But Facebook is a company built on forever; Spiegel’s great insight was that now
making your personal data public and deliv- the best way to make people pay attention is to
ering targeted ads; the whole point of Snap- capture that moment, share it, and watch as it
chat is to delete your images or videos after you disappears. —Ryan Bradley
69
Nora Ayanian
University of Southern California
Nora Ayanian calls robots people. It’s not some weird affecta-
tion; it helps her with her work.
She’s a computer scientist who thinks machines should
work together to get things done. Let’s say a farmer wants to
have drones autonomously survey crops and take soil samples.
You couldn’t program each drone with the same set of com-
mands, because each would have a different task and would
have to solve different problems as it navigated. You know
what is good at solving problems on the fly, in a group that
draws on various skills from different individuals? People.
So Ayanian studies robot coördination by studying people.
One way is by having groups of humans play a simple video
game that limits their senses and stifles communication. They
need to figure out how to do “something meaningful” together,
as she puts it, such as arranging their on-screen figures into a
circle. Ayanian watches how people coöperate on such tasks
with as little information as possible.
Why not just create a dictator
Distributed robot—one machine that sees the
and diverse whole field and directs other drones?
teams are Well, Ayanian counters, what hap-
pens when the dictator robot runs
always out of power? Or crashes? Distrib-
better at uted and diverse teams, she says, are
always better at problem-solving,
problem- once they learn to work together.
solving. —Ryan Bradley
“My research began in graduate school do amazing things because of her visual
when I was working on artificial-intelli- thinking abilities, it seemed to me that
gence systems and read Thinking in Pic- the same should be true of AI systems.
tures by Temple Grandin, a professor of “I’ve been taking what we learn from
animal science who talks about how her people on the autism spectrum who
70
71
72
Inventors
These innovators are building the stuff of the future, from
a smart sweatband to tomorrow’s memory technology.
Featuring
Alex Hegyi
PARC
74
Hegyi’s
camera
records
parts of the
spectrum
that you
can’t see.
75
Wei Gao
University of California, Berkeley
Evan Macosko Solution
Evan Macosko has helped invent a tech-
Harvard Medical School The engineer has built
nology called Drop-Seq, which allows sweatbands that monitor
A breakthrough in a researcher to look at thousands of
your health.
probing how cells create cells, one by one, to determine how
complex tissues and each is carrying out its genetic instruc-
tions. Such analysis of a single cell can “I grew up in a small village in
organs.
be done with existing tools, but it is typi- Xuzhou, China. When I was a
cally painstaking, expensive work that child I saw a lot of people around
Problem involves dropping individual cells into me dying of different diseases.
To truly understand the human tiny wells. “If you get two cells in a well, Many people don’t realize there’s
genome, we need better insight you’re screwed,” says Macosko.
a problem until it’s too late. I
into how individual cells differ. While To greatly speed up the process,
thought, in the future I should
every cell in a person’s body has Macosko figured out how to take each
design a wearable electronic device
basically the same DNA blueprint, cell he wanted to analyze, break it apart,
and attach the expressed genes to a to monitor health and tell us
there’s great variation in the way that
tiny bar-coded bead. Once material from what’s going on and what’s going
genetic information is actually acted
on, or expressed, at any given time. each cell is labeled, the genes can be wrong before it gets bad.
It’s the reason one cell becomes a analyzed rapidly—all for a cost of just “Our body is generating data
neuron that plays a role in memory, seven cents a cell. all the time. There are so many
while another cell becomes part Macosko says he and his team have wearable devices now—the Apple
of a person’s toenail. Even a given nearly finished profiling hundreds of watch, the Fitbit—but they mainly
organ, like the brain, encompasses thousands of cells spanning most of the track physical activities or vital
different types of cells, and individ- mouse brain. Next stop: the 86 billion
signs. They can’t provide informa-
ual cell types, too, have variations. neurons and innumerable other cells that
tion at the molecular level.
Inadequate knowledge about how make up the human brain. By analyz-
“It came into my mind: what
genes are expressed in different ing the great variation in the cells in our
brains, he hopes to identify the rogue about sweat?”
cells is greatly hampering progress in
cells that are malfunctioning or interfer- This year, Gao made a sweat-
genomic medicine.
ing with normal function in disorders like band that combines sensors with
schizophrenia, autism, and Alzheimer’s. electronic processors and a Blue-
—Michael Reilly tooth transmitter on a flexible
printed circuit board. If you wear
the band, it wirelessly transmits
data about what’s in your sweat to
a cell phone running an app.
Gao’s device has sensors that
interact with chemicals includ-
ing glucose and lactate, causing a
detectable change in their electri-
cal current. Other sensors change
their voltage in response to sodium
or potassium. A recent addition
includes sensors that can pick up
on toxic heavy metals excreted in
the sweat.
The challenge now is to figure
out whether and how these mea-
surements correspond to mean-
ingful changes in health. So Gao
is working with exercise physiolo-
gists on clinical studies to look for
LEONARD GRECO
76
“Our body is
generating data
all the time ...
It came into
my mind: what
about sweat?”
DAMIEN MALONEY
77
Her imaging
solution could
particularly help
people who are
obese, because
fat tissue
can distort
ultrasound
waves.
JUSTIN TSUCALAS
78
Muyinatu
“At the company I cofounded, Skydio, capable of
Lediju Bell we looked at all the things people more. It shows
Johns Hopkins University wanted to do with drones and realized up in the way
that the products are primitive com- it behaves and
Creating clearer imaging
pared to what’s possible. Today the typ- responds in different
to spot cancer earlier ical consumer experience is you take it situations.
and more accurately. out of the box and run it into a tree. “We aren’t saying a lot about our
“We’re building a drone for consum- product yet, but it’ll be a high-end con-
ers that understands the physical world, sumer device smart enough to fly itself
When biomedical engineer
reacts to you intelligently, and can use as well as or better than an expert pilot.
Muyinatu Lediju Bell was an that information to make decisions. It Devices that understand the world and
undergraduate at MIT, her has cameras positioned in a way so that can respond to you and take actions
mother died of breast cancer. computer vision can track its motion will open up things that don’t exist
Bell thought her mother might and understand the 3-D structure of the today. A flying camera that can be any-
have survived if she had been world. It also understands ‘This is a per- where around you would be a very pow-
diagnosed sooner, so she decided son,’ ‘This is a tree.’ We’ve demonstrated erful thing. Drones are likely to be the
to investigate what makes some the ability to fly autonomously in close first widely deployed category of mobile
ultrasound images blurry, a proximity to obstacles such as trees robot. As they start to get out into the
safely and reliably, and to follow some- world and people start to interact with
problem that limits a doctor’s
one walking, running, or cycling. them, it’s going to lead to some inter-
ability to screen for and diagnose
“On a week-to-week basis you can esting places.”
cancer and other diseases.
see the thing getting smarter and being —as told to Tom Simonite
As a doctoral candidate at
Duke University, Bell developed
and patented a novel signal pro-
cessing technique that produces
clearer ultrasound images in real
time. The solution could particu- Adam Bry
Skydio
larly help diagnose problems in
people who are obese, because Building drones that
fat tissue can scatter and distort can navigate the world
ultrasound waves, delaying the and serve as airborne
detection of a serious disease. assistants.
“I think it’s unfair that a long-
standing technology does not
serve a huge group of people that
should be able to benefit from it,”
she says.
Beyond ultrasound, Bell is
now working to improve another
type of noninvasive medical
imaging technique. Called pho-
toacoustic imaging, it uses a
combination of light and sound
to produce images of tissues in
the body. She is especially inter-
ested in using it for real-time
visualization of blood vessels
during neurosurgeries to lower
the risk of accidental harm to the
carotid artery, which supplies
blood to the brain. Her lab at
DAMIEN MALONEY
79
Kendra Kuhl
Opus 12
Growing up in rural Montana, K endra the design of the reactor, which incorpo- Kendra Kuhl’s reactor
Kuhl watched the namesake ice formations rates a family of catalysts she collaborated uses novel catalytic
nanoparticles (black
of nearby Glacier National Park shrink. on during her graduate work at Stanford
square in the bottom
“We could see global warming happening,” University. Sandwiched inside the metal photos). The reactor
she says. The sight drove her professional reactor chamber is an electrode that uses a (bottom, far right)
ambitions. “I liked the idea of putting membrane coated with the catalysts. They has inputs for carbon
atoms together in new ways that are poten- enable the carbon reactions to occur at low dioxide and outputs for
the chemicals.
tially friendly to the environment,” she says. temperature and pressure, without requir-
That’s just what Kuhl hopes to do ing large amounts of energy.
through the startup she cofounded in 2014. Opus 12 is not the first company to
Opus 12 is working on a reactor that will work on converting carbon dioxide into
take the carbon dioxide emitted by power widely used chemicals. But its improved
plants and make useful chemicals from it. catalysts and scalable reactor design set the
At Cyclotron Road, a startup incubator company apart, says Kuhl. Still, the com-
at the Lawrence Berkeley National Labo- pany has far to go before it can begin com-
ratory, Kuhl shows off one of Opus 12’s peting with traditional chemical suppliers.
prototypes, a small reactor with an input By the end of 2017, Opus 12 plans to build
for carbon dioxide and an output spigot a reactor with a stack of electrodes that
connected to an instrument that analyzes can produce several kilograms of product
the products. The key to the technology is a day. —Katherine Bourzac
type of memory. type of material called chalcogen- he could reduce switching time
ide glass between amorphous and to half a nanosecond. He and his
Computer designers have long crystalline states. Potentially, it is coworkers also reduced the size of a
desired a universal memory tech- fast like RAM and nonvolatile like memory-cell bit to just a few nano-
nology to replace the combination of flash. Since 2010, Desmond Loke meters. And he figured out how to
RAM—which is fast but expensive and his colleagues have solved sev- vastly reduce power consumption
and volatile, meaning it requires a eral critical problems holding up its and allow cells to be stacked in
power supply to retain stored infor- commercialization. three dimensions to pack in even
mation—and flash, which is non- As a result of the advances, more memory capacity.
volatile but relatively slow. the Singapore researcher has now —Michael Reilly
80
81
Jiawei Gu
Baidu
getting really good,” says Gu. a radio transmission can be 100 billion from his PhD studies at Stanford so he
Gu’s vision of the future is one times louder than the receiving one, it could commercialize the radio through
in which people can enjoy the ben- was always assumed that outgoing sig- the startup Kumu Networks. Germany-
efits of technology without being nals would invariably drown out incom- based Deutsche Telekom began testing it
captive to cords and notification ing ones. That’s why radios typically send last year, but since Bharadia’s prototype
buzzes. “I want to bring humans and receive on different frequencies or circuit board is too large to fit in a phone,
back to an unplugged age,” he says. rapidly alternate between transmitting it will be up to other engineers to minia-
—Christina Larson and receiving. “Even textbooks kind of turize it. —Ryan Cross
82
Subscribe Now
technologyreview.com/getinsider
Entrepreneurs
Meet seven people who hope to turn innovations into
disruptive businesses.
Featuring
Ari Roisman
Glide Roisman believes video
messaging will flourish on
Why the future of communication watches, whose screens
could be on your wrist. are too tiny for typing.
Ari Roisman clearly covets human con- nonverbal emotion, you can use Glide’s app working, especially as Instagram and
nection. Minutes after meeting me, the to send video messages with a single button Facebook pour millions into their own
32-year-old CEO of Glide gladly settles push. To illustrate, Roisman shows a Glide video-messaging plans. Glide laid off 25
into a conversation about the role of Juda- message he sent to his mother, featuring percent of its staff this spring.
ism in his life and how he gave up a prom- his daughter singing at a kindergarten Roisman says he scaled back on mar-
ising career in clean energy to move to event with the seriousness of a brain sur- keting and customer service to ensure his
Jerusalem. “My entire consciousness of this geon. Mom quickly responded with a video startup’s staying power. He wants the com-
world is that it is a gift,” he says, intently. of herself laughing at the performance. “If pany to focus on a technology he says will
Since 2012, Roisman has been striving we’re going to be glued to these devices, at make visual messaging the primary mode
to create a more human alternative to text least we should be connected in a way that of communication: the smart watch. He is
DAMIEN MALONEY
messaging. Rather than typing short mes- is more authentic,” he says. convinced that having a small screen just
sages on tiny smartphone keys, sometimes A few million people are using Glide, a wrist away will do for video messaging
adding emojis in a desperate stab to impart he says, but that’s a pittance in social net- what the PC did for e-mail. —Peter Burrows
84
85
Christine Ho
Imprint Energy
86
Meron Gribetz his sunglasses and made him one super-mobile package.
$949
Meta realize how he would do it. Within five years, he imagines,
Since then, he’s managed AR headsets will be reduced to
An augmented-reality to raise $73 million in fund- a strip of glass over your eyes
dreamer tries to turn his ing to go up against rivals like
PRICE OF THE META 2
that’s “nearly invisible.”
vision into a business. Microsoft and its HoloLens Meta is building software
device. Why the excitement? meant to be more intuitive
Meron Gribetz has a hard time This year, Gribetz unveiled to navigate than windows
sitting down when he’s talking the company’s latest headset, Both the Meta and the and icons. Gribetz believes
about his augmented-reality the Meta 2, which sells for oloLens are aimed at soft-
H so deeply in AR’s promise, in
startup, Meta. Grinning, he less than a third of what the ware developers, who will fact, that he’s pushing his own
stands or paces as he explains HoloLens headset is going have to come up with appli- employees to stop using com-
that he had always wanted for. It lets you do things like cations. But Gribetz, who was puter monitors and mouses
to create a way to bring digi- grab and prod 3-D imagery raised in Israel by American with their laptops by next
tal information into the real with your hands, or conduct a parents, is aggressively opti- spring; instead, the company
world to make it easier to video call with another Meta mistic about the technology will rely on Meta 2 and its
absorb. Then in 2011, sunlight user, who can hand you a vir- because he thinks it will let hand-tracking capabilities to
shimmering through an air- tual object that you can then us ditch devices like laptops, help them get their work done.
plane window hit the lens of inspect from any angle. smartphones, and tablets for —Rachel Metz
87
Two of GreyOrange’s
Butler robots, which
are designed to be
warehouse workhorses.
Problem Solution
“There are significant differences in the Bowerman is CEO of a startup, Dot Labo-
ways that men and women experience ratories, that is developing a cheap and
many diseases and drugs, and until this easy way to test female sex hormone lev-
problem is solved, women will be forced to els and track them online. A patient spits
make do with therapies that may be of lim- into a tube at specific times and mails the COURTESY OF SAMAY KOHLI ; COURTESY OF HEATHER BOWERMAN
ited benefit,” says Heather Bowerman. tubes to Dot Laboratories. The company
For example, hormones cause plaque then delivers data on hormone levels in an
to form differently in the arteries of men app for the woman or her doctor to review.
and women. Yet drugs to treat cardiovas- It’s still in a beta test; the company plans to
cular disease are tested disproportionately publish data on the efficacy of its methods
on men, and as one consequence, their and release the diagnostic product in 2017.
death rates from that illness are declining Developing more drugs that take hor-
Heather Bowerman faster than women’s. Detailed hormonal monal changes into account will take time.
Dot Laboratories data could help doctors tailor drugs and Even so, Anula Jayasuriya, a doctor who
treatment regimens so that they work bet- invests in life sciences companies, says
Cheap hormone tests could begin ter for women. such tests will help end the “sex bias in
to address gender disparities in basic research and clinical medicine.”
health care. —David Talbot
88
Zephyrus Biosciences
to be distracted by shiny things.
This bioengineer figured out how But it’s important for me to make
to handle a key challenge facing sure this technology is successful.
biotech startups. —Katherine Bourzac
89
90
Pioneers
Pushing the edge of science, these innovators are creating
new approaches to tackling technology challenges.
Featuring
Aleksandra catalysts for water splitting are likely to behave. Vojvod- of success. But today’s super-
and other reactions. The idea ic’s computer experiments, at computers are now capable
Vojvodic behind her work, she explains, the SLAC National Accelera- of doing much more complex
University of Pennsylvania
is to “circumvent the trial and tor Lab, have yielded oxygen- calculations. And Vojvodic has
A computation whiz error of nature”—and of the producing catalysts that match been exceptionally talented at
speeds up the search chemistry lab. or outperform those made of taking advantage of computing
for catalysts that will Splitting water requires expensive materials. power; identifying new ways
make green chemistry two catalysts, one for making Researchers have been to represent electronic proper-
possible. hydrogen and the other for using powerful computers to ties, chemical structure, nano-
making oxygen. “The things try to design better catalysts structure, and other properties
Using enzymes honed over that work efficiently are usu- for years, with varying degrees in mathematical calculations;
hundreds of millions of years ally rare or expensive,” says and writing programs to
of evolution, plants readily Vojvodic. That’s where com- carry them out. Working with
split water into oxygen and putational chemistry comes The idea behind experimentalists, she and her
hydrogen that’s used to fuel in. To predict the behavior coworkers have recently made
metabolic reactions. Humans, of a catalyst, Vojvodic makes her work, she extremely efficient water-
too, could use hydrogen as a computer models that relate explains, is to splitting catalysts that her
fuel and a way to store energy
from intermittent renewable
a material’s functions to its
structure using the rules of
“circumvent the modeling work predicted. The
researchers are now eyeing
sources. But we don’t have mil- quantum mechanics. Chem- trial and error other catalysts, including ones
lions of years to figure out how ists know what functions of nature”— that can convert nitrogen and
to make practical catalysts. the catalyst needs to have, other abundant molecules into
Aleksandra Vojvodic uses and they know how different and of the useful chemicals.
supercomputers to design new kinds of atoms and structures chemistry lab. —Katherine Bourzac
Jia Zhu absorbing lots of sunlight and is turned into steam, what’s
Nanjing University using the energy to generate left are salts or solidified con-
steam that condenses into taminants that can be easily
What to do if there is no clean water. “It only needs two collected.
clean water around. things. The first is water—no He also envisions other
RC RIVERA; COURTESY OF JIA ZHU
91
Qing Cao
IBM Research
Using a macroscale
2006 The first integrated circuit using
model, Zhang a single carbon nanotube is built
shows off his at IBM.
clever trick for
making complex 2008 During his doctoral studies at the
nanostructures. University of Illinois, Qing Cao
invents a way to print circuits of
nanotubes on flexible plastic sub-
Yihui Zhang strates.
Tsinghua University
2013 At IBM, Cao develops a technique
Pop-up nanostructures make it far easier to that applies mechanical force to
fabricate very tiny shapes. push purified nanotubes in water
together into high-density, neatly
Yihui Zhang likes to invite visitors to his office to stretch a piece of highly elas- ordered arrays.
tic silicone that has a soccer-ball-like structure attached to it. Once the silicone
is pulled taut from four corners, the three-dimensional structure becomes a 2015 Cao overcomes a fundamental
roadblock to commercially viable
two-dimensional pattern that looks like a wheel with many adjacent hexagons
nanotube transistors. He devises
and pentagons in the center. When the silicone is relaxed again, the flattened
a way to connect metal wires to
pattern pops back into its three-dimensional shape.
carbon nanotubes by welding
With this trick, Zhang has solved the challenge facing many researchers: metal atoms to the nanotubes’
how to fabricate complex three-dimensional nanoscale structures. Although ends.
the demonstration is done at the macro level, the idea works with nanostruc-
tures, too: easily created two-dimensional patterns can be attached to a sub- 2016 IBM incorporates carbon nano-
strate stretched taut and then buckled into tubes into its in-house semicon-
Easily created three-dimensional structures as the substrate is ductor research line to figure out
relaxed. This process works with a wide range how to refine and scale up the
two-dimensional of materials such as metals and polymers. technology.
patterns buckle The technique could be used to create
2020– IBM aims to have its nanotube
into three- nanostructures for a variety of uses. Ultimately,
2025 transistors ready to replace sili-
Zhang hopes to develop a database or algo-
dimensional rithm that allows researchers to easily map
con transistors. The company
estimates that nanotube tran-
structures as the three-dimensional structures they want sistors will perform two to three
GILLES SABRIE
92
93
JESS KOHL
94
Oriol Vinyals he helped to create an AI bot While he was standing in the market buying
that was able to play StarCraft fruits.” “It worked!” he recalls.
Google DeepMind
unassisted. The bot, forebod- working on “It wasn’t just saying ‘People on
Showing computers ingly dubbed Overmind, rep- the Google AI the street.’ It was reading the
how to learn might seem resented a triumph in machine image with sophistication.” The
like a game, but it’s also learning.
team, creating technology, now being incor-
serious business. Later, while he was work- new techniques porated into Google Image
ing on the Google AI team for translation, Search, allows computers to
When he was 15 years old, creating new techniques for caption images and show them
Oriol Vinyals became obsessed language translation, inspira- inspiration to people who enter relevant
with StarCraft, a video game tion struck. Vinyals decided to struck. search terms.
in which three factions vie for see whether a computer could Vinyals and his coworkers
control of the map—like chess accurately write a descrip- have developed a technology
if it were played not only with tion of an image. It’s a form of now used in Gmail called Smart
black and white pieces but also translation, albeit from pixel Reply, which automatically sug-
with red ones. Vinyals soon to caption. “I remember it gests short replies to e-mails.
became the top-ranked player so well,” he says. “I changed And now, having joined the
in Spain. “I almost knew the a single line of code: instead team at Google DeepMind in
game would return later in my of translating from French, I London, he has come full cir-
life,” he says. “I was fascinated changed my code to input an cle. There, he is working to cre-
by the artificial-intelligence image instead.” The next day, ate computers that can teach
problems it presents.” Vinyals showed his program a themselves how to play and win
It was more than a decade photograph of a busy market complex games—not by hard-
before Vinyals’s premoni- stall, the ground beside it lit- coding the rules but by enabling
tion came to pass. While he tered with bananas. The cap- them to learn from experience.
was studying at UC Berkeley, tion read: “A group of people —Simon Parkin
A polymer
solar cell
printed on
glass.
Problem Solution
Ying Diao Flexible solar cells that are cheap to make Ying Diao is creating printing techniques
University of Illinois could be “printed” on many surfaces, even that bring order to the otherwise chaotic
windows. But the polymers that would be assembly of plastic molecules. She has
She knows how to print
required have so far been lackluster at con- made organic solar cells with double the
perfect plastic solar cells.
verting sunlight to electricity. One reason efficiency of previous ones. Diao came up
is that unlike more efficient solar materials with a microscopic “comb” that controls
such as crystalline silicon, polymer-based the flow of the molecules and lets them
SETH LOWE
materials have a messy molecular structure assemble into orderly structures during
that looks like cooked spaghetti. printing. —Ryan Cross
95
96
Sergey Levine any of the pieces themselves,” ing the task over and over, it solve versions of the task at
University of California, Berkeley
he jokes. eventually attains the goal. But hand—instructing it to screw
One way that the creators the learning process requires on the cap, for example—the
He teaches robots to of AlphaGo trained the pro- lots of attempts, and it doesn’t robot then retrospectively
watch and learn from gram was by feeding 160,000 work with difficult tasks. studies its own successes. It
their own successes. previous games of Go to a pow- Levine’s breakthrough was observes how the data from
erful algorithm called a neural to use the same kind of algo- its vision system maps to the
While serving a nine-month network, much the way similar rithm that has gotten so good motor signals of the robotic
stint at Google, Sergey Levine algorithms have been shown at classifying images. After he hand doing the task cor-
watched as the company’s countless labeled pictures of gives the robot some easy-to- rectly. The robot supervises
AlphaGo program defeated cats and dogs until they learn its own learning. “It’s reverse-
the world’s best human player to recognize the animals in engineering its own behavior,”
of the ancient Chinese game unlabeled photos. But this Levine says. It then can apply
Go in March 2016. Levine, a technique isn’t easily applica-
The robot that learning to related tasks.
robotics specialist at the Uni- ble to training a robotic arm. observes how With the AI technique,
versity of California, Berke- So roboticists have instead
the data from previously insoluble robotics
ley, admired the sophisticated turned to a different tech- tasks have suddenly become
feat of machine learning but nique: the scientist gives a its vision system approachable, thanks to the
couldn’t help focusing on a robot a goal, such as screwing maps to the massive increase in training
DANIEL BERMAN
notable shortcoming of the a cap onto a bottle, but relies efficiency. Suddenly, robots are
powerful Go-playing algo- on the machine to figure out
motor signals getting a lot more clever.
rithms. “They never picked up the specifics itself. By attempt- of its hand. —Andrew Rosenblum
97
Humanitarians
They are taking unconventional routes to bring about a
healthier, cleaner, and more adaptable world.
Featuring
Sonia Vallabh
Broad Institute
Five years ago, Sonia Vallabh just married my husband, Eric “We decided right away I’d
graduated from Harvard Law Minikel. My mom, healthy at 51, get tested. We wanted to know
School and went to work at a had single-handedly organized what we were up against. After
small consulting company. But our beautiful wedding. Then, months in agonizing limbo, a
a stunning medical diagnosis all of a sudden, we were watch- geneticist confirmed our great-
made her change course com- ing her waste away before our est fear: The same change that
pletely: she learned she has a eyes. We had no name for what was found in your mother was
genetic mutation that causes a we were seeing. It was only from found in you.
deadly brain disease. Today she her autopsy that we learned “Knowing the hard truth has
and her husband work in a lab there was a 50 percent chance given us a head start against our
at the Broad Institute of MIT I’d inherited the genetic muta- formidable medical enemy. We
and Harvard and have pub- tion that killed her. waged a campaign to educate
lished research showing a pos- ourselves—taking night classes,
sible pathway to a treatment. As attending conferences, and even-
she told the tale at an event on tually taking new jobs in research
precision medicine with Presi- labs. We retrained as scientists
dent Obama in February: by day and applied what we were
“At the heart of my story is a learning to understanding my
single typo in my genome. disease by night. Four years later,
“We all carry around thou- we’re devoting our lives to devel-
sands of typos in our DNA, most oping therapeutics for my disease.
of which don’t matter much to “We know the road ahead is
our health—but my typo is an “Knowing the uncertain—no amount of hard
unusually clear-cut case. It’s a work can guarantee there will be
single change in a particular gene hard truth a treatment for me when I need
that causes fatal genetic prion has given us one. We are going to do every-
disease, where patients can live
50 healthy years but then sud-
a head start thing we can, hand in hand with
creative allies from every sector,
denly fall into deep dementia and against our to build this bridge as we walk
die within a year. And there’s no formidable across it and develop a treatment
LEONARD GRECO
treatment—at least, not yet. that could save my life, and the
“In 2010, I watched this medical lives of many others.”
disease unfold firsthand. I had enemy.” —Antonio Regalado
98
Vallabh and
her husband,
Eric Minikel, in
the lab where
they are in a
race to solve
puzzles of
prion disease.
99
Ronaldo Tenório
Hand Talk
A deaf person walks into a bar. son, who sees a message on thousands of example sen-
That isn’t the beginning of a
joke, but a potentially frustrat-
ing situation—unless the bar-
6 million the screen that says “Speak to
translate.” As soon as the per-
son starts talking, an animated
tences every month and match
them with 3-D animations of
sign language. They constantly
COURTESY OF EHSAN HOQUE; COURTESY OF RONALDO TENÓRIO
NUMBER OF MONTHLY
tender happens to know sign TRANSLATIONS ON avatar named Hugo begins push these improvements out
language. That’s where Hand HAND TALK signing. through app updates.
Talk comes in. It translates Turning the audio into ani- Tenório plans to roll out
spoken words into sign lan- mations of gestures requires different versions of the ava-
guage that an avatar then con- aldo Tenório. But Brazil alone laborious programming tar in the future so users can
veys on a smartphone screen. has at least 10 million deaf because everything has to be switch the gender or race
For now, Hand Talk can people, one million of whom exactly right, all the way down of their Hugo in an effort
only translate Portuguese have downloaded Hand Talk’s to Hugo’s facial expressions, to broaden the appeal and
into Libras, the sign language mobile app. which also carry meaning in accessibility of having a vir-
used in Brazil—the home of The users hold up their sign language. Tenório and tual translator in one’s pocket.
the program’s creator, Ron- smartphone to a hearing per- his team feed their program —Julia Sklar
100
Jagdish Chaturvedi
InnAccel
101
Kelly Sanders
University of Southern California
A researcher in drought-ridden
California tries to better account
for the ways we use water.
102
Sanders has
developed
better ways
of quantifying
the energy it
takes to supply
water—and
vice versa.
103
30 Years Ago
Communication
Breakdown
A pair of experts mulled whether we’d
ever get machines to talk, let alone think.
MIT Technology Review Best in Tech (ISSN 2474-6606), Reg. U.S. Library of Congress, is an MTR Special Publication published annually by MIT Technology Review, 1 Main Street, Suite 13, Cambridge, MA 02142-1517. Entire con-
tents ©2016. The editors seek diverse views, and authors’ opinions do not represent the official policies of their institutions or those of MIT. Subscriptions to MIT Technology Review can be ordered at www.technologyreview.com.
104