Ji Xu played a key role in building the world’s largest payment platform, which can support more than a billion transactions a day. It’s a boon to commerce, but more important, it enables anyone—especially people without access to traditional banks—to use financial services over the internet.
Originally developed to make payments on Alibaba’s online shopping sites easier and more reliable, Alipay has become a ubiquitous electronic payment app in Chinese e-commerce and brick-and-mortar stores alike. It has 520 million users, who see cash as a thing of the past: whether grocery shopping, paying utility bills, or buying movie tickets, they simply pull out a smartphone and use Alipay to scan a payment code.
As its business grew, Alipay was confronted with two challenges. First of all, it needed to increase the number of transactions it could handle. In addition, it needed to manage a growing variety of funding options. People had started linking all sorts of funds—credit cards, debit cards, electronic cash gifts, and investment portfolios—with Alipay to pay for things, and sometimes one purchase was made using multiple types of funds.
As the chief architect of Alipay’s core payment platform, Xu led a team that increased the system’s capacity from 10 million to 100 million transactions a day, and eventually to one billion. The new system can use servers located anywhere without causing delays during peak hours, which is crucial because the servers consume so much power that no single location can support enough of them to meet the system’s requirements.
The increase in Alipay’s transaction capacity also made it possible to offer online financial services to anyone, regardless of income level. One popular feature of the app lets users invest their leftover cash from online spending in a fund and earn interest at higher rates than they could at a bank.
Growing up near Hangzhou, where the Alipay team is based, Xu was not interested in taking exams. He spent about two years in a computer science program in college before leaving to look for a job, and he joined Alipay at 23. “I wanted to learn technologies,” he recalls. While on the job, he caught up on programming knowledge through online courses.
Natalya Bailey helped develop a way to propel satellites as small as a shoebox or as big as a refrigerator using engines about the size of a dime. It’s based on so-called electrospray propulsion—the idea of using electrical energy to drive small rockets.
Electrospray technology has been in the works for many years. Researchers started studying it in the 1950s, but the work was abandoned because it required very high voltages and because the physics involved was not well understood. Bailey was able to use the technology’s advantages—it’s energy-efficient and doesn’t require toxic propellants or pressurized tanks—to create tiny engines that can be used independently or in tandem with other engines, depending on the size of the satellite.
Bailey founded Accion Systems, just outside Boston, to commercialize the technology. She says the rocket-science field can feel like an old boys’ club but she’s made it work. “Being one of very few women in this field makes me stand out more,” she says. “And I think it probably led to some opportunities that I maybe wouldn’t have had otherwise.”
Jonas Cleveland thinks shopping robots will not only be picking goods off the shelves at massive warehouses but roaming the aisles at local businesses, grabbing products for online orders in stores that are also full of human shoppers.
Cleveland’s company, COSY (for Cognitive Operational Systems), is creating the sensor perception system for those robots. Cameras, AI, and mapping technology help make them smart enough to do their job without interfering with the people around them. So if you’re shopping in a pharmacy or home improvement store, Cleveland’s robots won’t bump into you, and if you’ve ordered online for delivery, the robot that prepares your order will know a six-pack of Diet Coke from a six-pack of Coke Zero.
Elizabeth Nyeko thinks she’s found a solution to one of rural Africa’s key development challenges: how to electrify communities in a way that’s affordable—and efficient.
As CEO of Modularity Grid, a London-based startup, Nyeko builds technologies that improve the performance of mini-grids, small-scale electricity generation and distribution systems that power homes and businesses in areas where extending national grids is too expensive. Yet mini-grids also have limitations. As Nyeko learned at Mandulis Energy, a company she cofounded that built a biomass-fired mini-grid in northern Uganda, the electricity demand of individual customers is very hard to track, which typically leads to overproduction of power, inefficient use of fuels, and inflated electricity prices.
At Modularity Grid, Nyeko designed an intelligent cloud-based platform that enables mini-grid operators to better track and predict individual consumption; it then redirects excess electricity to specific users in need of constant power, called “anchor loads.” At the Mandulis site in Uganda, where Nyeko is piloting her Modularity Grid solution, the anchor loads include the village rice mill—which also provides the rice husks used to fuel the mini-grid itself. “If we can deliver just the amount of electricity to people that they need, and redirect the rest to something that creates value for a rural community, we can make mini-grids viable in a low-income setting,” Nyeko says.
Nyeko, who was born in northern Uganda but fled from civil war there as a child, is now marketing her solution to other mini-grid providers and is set to begin work with Total and Vinci Energies on further power projects across Africa. Eventually, she believes, her solution can also help make national grids more efficient—in Africa and beyond.
—Jonathan W. Rosen
Seven years ago, Yin Qi founded a company called Megvii with two college friends in Beijing. Now people from over 220 countries and regions use Megvii’s face-recognition platform, Face++. The company has more than 1,500 employees.
Face++ has transformed businesses in China, both online and offline. Subways and train stations use face recognition to expedite the screening process; banking apps use it to confirm the identities of their users.
Being in China has given Megvii an edge. While the use of face recognition in the West has mostly been confined to consumer-oriented applications such as unlocking smartphones, in China the same technology enjoyed strong backing from the government and big companies right away. This gives Megvii ample opportunities to commercialize its algorithms for industries as diverse as public security, real estate, finance, and retail.
Yin admits privacy is an issue. He says his products process sensitive raw data on local devices instead of uploading them to the cloud. He’d also like to see an industrywide standard on user privacy. “When there is a good system to manage and run these technologies, the benefits they will bring will outweigh the drawbacks,” he says.
Ashutosh Saxena is the CEO and cofounder of Brain of Things, which developed an AI system called Caspar that turns a home into a sort of robot that we can talk to and interact with. By later this summer, Caspar will have been installed in about 500 apartments in California and Tokyo.
Each of these apartments is outfitted with around 100 devices including motion and humidity sensors, microphones, cameras, thermostats, and automated appliances. All of these feed data about residents’ behavior to Caspar, which uses a number of algorithms to analyze the data so that it gradually learns and adapts to people’s habits and preferences.
If you tend to ask a lot of questions about the packages you are expecting, Caspar will learn to send you alerts when they arrive. It will also learn to tailor its music playlist to what you are doing at the moment.
When asked whether it’s safe to entrust so many intimate details of our lives to a computer, Saxena says the sensitive raw data generated is stored within the home and not uploaded to the cloud.
The idea of creating Caspar came about in 2015, when Saxena and his roommate took home a couple of smart speakers. These devices, such as the Amazon Echo, can play music, order things online, switch the lights on and off, and do many other things around the house. But the roommates struggled to make the gadgets work the way they wanted them to. The virtual helpers sometimes turned off the wrong light, and when their masters’ schedules changed, they couldn’t adjust their control of other devices accordingly.
So Saxena, a robotics researcher, decided to build a better system.
“You no longer need to worry about packages not arriving at your home,” Saxena says. “Caspar notifies you of such things, orders dishwasher soap, or controls your home environment according to your preferences.”
For renewables to work, they need batteries—otherwise, the lights go out when the sun goes down or the wind isn’t blowing. Companies like Tesla and Hyundai are addressing the problem by developing football-field-size lithium-ion batteries in Australia and South Korea.
These massive batteries, however, are expensive.
“There’s a cost floor to lithium-ion, which is dictated by the components that are used,” says William Woodford, the chief technology officer of Form Energy. “No matter how cheaply you put it together, you still have a certain set of active ingredients, and those have costs.” So while Elon Musk can build bigger, cheaper batteries, there’s a limit to how cheap they’ll ever get. Lithium carbonate, for example, can cost as much as $20,000 a ton.
To address this problem, Woodford has identified metal-sulfur chemistries that could beat lithium-ion technologies for long-term storage and cost. As a bonus, sulfur is cheap and abundant: it often goes unused as a waste product of oil and gas production.
Traditional approaches to drug development for diseases like Alzheimer’s, Parkinson’s, and amyotrophic lateral sclerosis (ALS) haven’t offered patients much. Alice Zhang is trying something new. Her company, Verge Genomics, uses artificial intelligence to identify promising compounds, refining the algorithms with high-quality data from patients and lab tests. She hopes this will be a more effective way to find treatments for intractable neurodegenerative diseases.
Zhang’s unorthodox method was inspired when she heard a researcher give a talk detailing how hundreds of genes interact in cancer and wondered whether this “network” approach could apply to neurodegenerative diseases. “Computational biology has provided so much insight about cancer,” she says. “The brain is about 10 years behind.”
Verge is developing machine--learning models that identify key genes within a disease network and predict which compounds might interfere with their activity. It tests these compounds in animal models and nerves grown from patient-derived stem cells. The company then feeds the results back into the machine-learning model to refine it further. Zhang says seven of Verge’s candidate compounds for ALS have slowed cell death in patient neurons in vitro.