Quantum computing, the processing technology that's turbocharged by the weirdness of subatomic physics, is only just becoming a reality. But the newness of the field hasn't stopped AT&T from going another step further.
The telecommunications giant is researching technology to link quantum computers, hoping to amplify their power much the way networking of conventional computers led to massive supercomputers and services spread across the globe. Quantum networking could lead to a similar leap for quantum computers and possibly form the foundation for a quantum internet.
AT&T doesn't expect to bring this technology to market anytime soon. Instead, it's trying to figure out what's possible and bring it closer to commercial reality through a partnership called Intelligent Quantum Networks and Technologies (INQNET). The work involves researchers from the California Institute of Technology, Stanford University, national laboratories, startups, the military and other institutions.
"How do you get it to a point where you can scale it so you can afford to buy one of these things?" said AT&T Chief Technology Officer Andre Fuetsch, sharing details of the effort for the first time at the AT&T Foundry in Palo Alto, California, the research and development lab that's headquarters for the quantum networking project. "We want to make sure we're there and we're relevant."
The mind-bending physics of quantum mechanics, which governs the workings of atoms and smaller particles, are good for more than superhero movie plot twists. They're also well suited to the computing needs of designing new medical drugs, optimizing financial investments and cracking today's encryption.
Qubits, the fundamental element of quantum computing, allow for the transmission of more information than conventional computer processors, which handle data as a bit -- a state of 1 or 0. Quantum computers use qubits, which can store multiple states at the same time. The goal of quantum networking is to link qubits across multiple quantum computers.
"You could allow qubits to interact with each other as if they were next to each other," said Soren Telfer, director of the AT&T Foundry in Palo Alto.
Quantum computing benefits
If researchers crack the challenges, quantum networks could let you:
- Link multiple machines together to tackle larger problems, similar to how multiple classical computers today can share databases or cooperate on calculations.
- Build a network of quantum sensors that feed data to a quantum computer. These could be accelerometers for detecting motion in super-sensitive gravitational research or navigational guidance. Quantum sensing is "much more feasible in the near term" than more elaborate quantum communications, says Yewon Gim, a senior technical staff member at the AT&T Foundry.
- Secure communications. With quantum networks, "you will always know if someone is eavesdropping," said J.D. Dulny, chief scientist at consulting firm Booz Allen Hamilton, which isn't a member of AT&T's alliance. "You can't duplicate quantum state," so it's impossible for an eavesdropper to elbow in on a quantum-network communication link, he said.
- Let competitors cooperate on "secure multiparty computation," in which proprietary data from each could be combined for calculations without the parties revealing their secrets, Telfer said.
Qubits, superposition, entanglement -- the works
Qubits, which in the real world are typically atomic nuclei, photons or other tiny particles, get their power through concepts called superposition and entanglement. Together, they exponentially raise the amount of data qubits can store and process.
Superposition means a single qubit can store not merely a 0 or a 1, as a classical bit does, but in effect both at the same time. (Yes, that's like a light being both on and off simultaneously. I told you quantum mechanics is weird.) Two qubits can store four states simultaneously, three qubits can store eight states, four qubits can store 16 and so on.
Entanglement links these qubits together so that when you perform some kind of processing operation, you perform it on all possible combinations of those qubits' 0s and 1s at the same time. In other words, the difference between classical and quantum comping is that four bits can store one of 16 possible numbers while four qubits can store all 16.
With superposition and entanglement -- and an awful lot of expensive, ultracold hardware to get them to work reliably -- quantum computers can rapidly explore a vast number of possible solutions to a problem. That potential performance boost is why companies like IBM, Rigetti Computing, D-Wave and Google are racing to pack as many qubits onto quantum chips as possible.
Entanglement is important to quantum networking, too, because it works even for separated qubits. If you've heard Albert Einstein's famous objection to quantum mechanics -- "spooky action at a distance" -- that's what we're talking about. (Experimental physics proved Einstein wrong on this point.)
"Linking two quantum computers or quantum chips together while maintaining their quantum states is an active area of research," said Jim Clarke, Intel Labs' director of quantum hardware. (Intel isn't a member of AT&T's alliance.) Academic research has shown evidence that long-range quantum coupling works, too, he said. But because quantum networking "is still very much in the research stage," you should expect progress in isolated quantum computers to come sooner.
First steps in quantum networking
Quantum sensor networks are likely the first quantum networking application. The holy grail, though, is packing the state of a quantum computer into a photon of light transmitted over fiber-optic lines to another quantum computer, an approach that relies on an idea called transduction.
The transduction problem is particularly thorny, said Dan Campbell, a Booz Allen Hamilton scientist who previously worked on quantum computing at the Massachusetts Institute of Technology. To communicate with quantum computers, you use electromagnetic signals at one frequency. But to transfer quantum data over fiber-optic lines, you need photons at a different frequency. Transduction is the technology that will convert from one frequency to another.
That means data sent on quantum networks will need to go through two frequency conversions to send data from one machine to another. "It's hard to think of that process ever becoming perfect," Campbell said.
But, counters AT&T's Telfer, "Hard doesn't mean impossible. There is nothing fundamental that would keep transduction from occurring."
Maybe a quantum internet someday
You can link quantum computers with plain old classical networks -- indeed, that's how today's early quantum computers are used. But converting communication to the classical realm slows it down and sacrifices the special benefits of entangled quantum data.
That's why AT&T's researchers want the quantum links. The ultimate goal of the company's partnership is to build a quantum internet, INQNET leader Pravahan said. Overcoming the challenges of physics and engineering are immense, but success could open up "fundamentally new capabilities in science and technology," in the words of Austrian researchers working on the idea.
In a few years or decades, maybe a quantum network will be in your life, precisely synchronizing actions in your online gaming and locking peeping Toms out of your video chats -- or doing something that's just not even conceivable today.
Originally published June 27, 5 a.m. PT.
Correction, June 28: Fixes the spelling of J.D. Dulny.