Fortune USA 201906

(Chris Devlin) #1

QUANTUM COMPUTING BUSINESS BETS ON A QUANTUM LEAP


168


FORTUNE.COM // JUNE.1.19


Monroe had published a paper that month
in a prominent physics journal, effectively
outlining a road map for how certain de-
vices could help quantum computing leap
forward. When the visitor showed up at
Monroe’s office, he brought the article with
him. This wasn’t just a science paper, the
man said, waving the document in the air.
“This is a business plan!”
That investor was Harry Weller, a
partner at the venture capital firm New
Enterprise Associates (NEA), a legendarily
successful early backer of shopping site
Groupon and a passel of software start-
ups. Monroe, who had been contentedly
sustaining his academic research with
grants from the U.S. intelligence commu-
nity, wasn’t interested at first. Eventually,
though, he came around and accepted
Weller’s funding proposal, founding IonQ
in 2015. (Weller died in 2016.)
IonQ is working on an approach to
quantum computing, described in Mon-
roe’s paper, called the “ion-trap” method. It
activates the qubits in its system by manip-
ulating ions, or charged atoms, with laser
beams. In the ion-trap method, unlike with
superconducting qubits, physical wires are
not needed to send control signals into the
machine. That means the qubits are better
protected from “noise,” or disturbances
that contribute to error, Monroe says. They
sit suspended in a vacuum cushion, like
a maglev train hovering on its tracks. GV,
the venture capital arm of Google parent
Alphabet, joined NEA as an IonQ investor
in 2017. In May, the company added the
former director of engineering of Amazon
Prime as its CEO.
The ion-trap idea has some prominent
converts. Honeywell, the industrial con-
glomerate, last year debuted an ion-trap
approach that it had been working on in
secret for years—a major point of validation
for Monroe’s startup. Honeywell found that
its expertise in areas like vacuum systems,
lasers and optics, microelectronics fabrica-
tion, and other disciplines all converged in
the new field. “If you put all those things
together, you can build a quantum comput-
er,” says Tony Uttley, who leads Honeywell’s
100-person quantum efforts.
The ion-trap method is only one of

One reason the approach is so popular is because it builds atop
decades of advances in the semiconductor industry. These qubits
are created inside specially designed silicon devices; they’re gener-
ated by an electrical current flowing between superconducting
electrodes separated by a thin insulating barrier. (This works only
in cryostatic, ultracold chambers, which helps explain why quan-
tum computers will live for the foreseeable future in labs and data
centers, not on desktops.)
When someone operating a quantum computer enters certain
commands, they can link two qubits together, entwining them in
a state called “entanglement.” If something happens to one en-
tangled qubit, its mate instantaneously reacts. By stitching together
networks of such qubits, a programmer can run massively parallel
operations, meaning a huge number of operations at once. This is
what enables quantum computing’s exponential speedups.
“Superposition,” a related concept, is the other key to quantum
computing. Whereas bits, the building blocks of classical comput-
ing, are limited to representing information as “zeroes” and “ones,”
qubits can assume any combination of gradations between zero and
one. Think of this as the difference between a coin at rest on a table,
displaying heads or tails, vs. one spinning, ballerina-like, on its
edge. The result: Superposition allows qubits to store vast amounts
of data compared with regular bits.
Together, superposition and entanglement give quantum com-
puting its kick—amplified memory tackling complex problems at
remarkable speeds. (The trick only works while no one is watch-
ing, a bizarre but fundamental fact of quantum science. As soon
as someone observes the system, everything collapses.) The act of
measurement causes a cascade of tipped-over qubits that produces a
final state. If the math is right and the machine well-designed, then
that system should tend toward the most probable, most optimal
state—the solution.
Each qubit adds exponential power. But as the quantity of qubits
grows, quality becomes a limiting factor. As with a spinning coin,
even the most minor disturbances, such as heat or vibrations, can
shake up the system, causing errors that manifest as wrong an-
swers. And in today’s machinery, as the number of qubits increase,
so do error rates. Indeed, some practitioners fear there may be
a fundamental law, as yet unknown, prohibiting these machines
from working at scale—like Jenga towers, they may be doomed to
tumble when they get too high. Some skeptics, such as Gil Kalai, a
professor at Hebrew University in Israel, believe that the technol-
ogy will never work as hoped: “My analysis suggests that efforts to
build quantum computers are going to fail,” Kalai says.
That tension explains why IBM and Google are so eager to
demonstrate that they’ve fortified their qubits and lowered their
error rates. It also explains why other scientists are exploring the
possibility of a better way forward.


C


HRIS MONROE, A PHYSICS PROFESSOR at the University of
Maryland, remembers the cold-call email in Febru-
ary 2014 that changed his professional destiny. The
correspondent was an investor who sought a meeting.
Free download pdf