Sign In

Communications of the ACM

ACM News

Quantum Datacenter Intranet Advances

IBM plans to expand its quantum systems from 127 qubits today to 1,121 qubits next year, when its collaboration with Argonne National Labs will enable the connection of separate quantum systems at superconducting speeds for any number of qubit intranets.

Credit: IBM

U.S., European Union, and Japanese researchers recently made advances that they say are necessary for future quantum datacenters, where millions of qubits will be interconnected with a quantum intranet (the private communications grid among separate quantum computers).

In particular, this year U.S. national lab-sponsored researchers have announced the capability to preserve the coherence of quantum superposition states for five seconds, "more than 500 times longer than last year's record," according to David Awschalom, principal investigator and senior scientist at Argonne National Laboratory, director of the U.S. Department of Energy's Q-NEXT next-generation quantum research and science project, and Liew Family Professor in Molecular Engineering and Physics at the University of Chicago. During that five-second interval, as many as 100 million quantum operations and their intermediate results could be communicated over a quantum intranet, according to Awschalom.

Also this year, researchers at QuTech, a collaboration between Delft University of Technology in the Netherlands and TNO, the Netherlands Organization for applied scientific research, demonstrated quantum gates with greater than 99.5% fidelity—the benchmark accuracy needed for quantum error correction, according to Lieven Vandersypen in the QuTech lab (the achievement was confirmed independently at the Riken Center for Emergent Matter Science in Saitama, Japan). "High-fidelity control of quantum bits is paramount for the reliable execution of quantum algorithms and for achieving fault tolerance," said Vandersypen. "Having surpassed the 99% barrier for two-qubit gate fidelity, semiconductor qubits are well positioned on the path to fault tolerance."

In addition, at the end of last year, IBM announced its 127-qubit Eagle quantum computer, to be followed by its 433-qubit Osprey model later this year, and next year by a 1,121-qubit Condor model, according to IBM senior vice president Dario Gil. "Eagle, Osprey, and Condor will truly let us explore uncharted computational territory. …By 2030, we predict that our quantum computer users will be running a trillion quantum circuits per day, each of which will be solving problems that cannot be solved today on traditional digital computers."

Says Who?

"Recently, landmark experiments have carried out specialized tasks beyond the practical reach of supercomputers," said independent quantum expert Hendrik Bluhm at Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen University and the Forschungszentrum Jülich Research Centee in Germany. "However, these demonstrations are still far from reaching the millions of qubits necessary to fulfill the envisioned applications of quantum computers."

As such, these recent breakthroughs must still be considered proofs of concept. Taken together, they demonstrate that the pieces of future quantum datacenters are coming together, according to independent analyst Bob Sorensen, who also is chief analyst for quantum computing at Hyperion Research, which focuses on high-performance computing. According to Sorensen, quantum datacenters will mimic supercomputer datacenters, in that they will amass processor chips with thousands, perhaps millions, of quantum processors working together over intranets.

"We always knew that it would be impossible to fit the millions of qubits necessary for fully mature quantum computers in a single supercooled unit, but connecting them was a problem that remained unsolved until now," said Sorensen. "Now with these long coherence times, error-correction advances, and dedicated industrial partners building the necessary intranets, the time to success may be shorter than we think."

IBM, for one, is pioneering efforts to create ever-larger dilution refrigerators—a necessity when cooling increasingly large intranets of superconducting quantum processors down to the millikelvin realm where errors are minimized. Each of IBM's quantum computers uses a single refrigerator, with the next generation of quantum datacenters connecting them with intranets in a manner similar to the way the world's fastest supercomputer datacenters are connected with optical fibers today.

"We knew that the fault-tolerant quantum computers of the future would require a distributed approach similar to today's supercomputers," said Jerry Chow, IBM's senior manager of quantum system technology at the T. J. Watson Research Center, Yorktown Heights, NY. "This could only be possible if we developed quantum interconnects to link processors together into a quantum intranet."

Quantum Intranet

Unlike a quantum Internet for secure quantum-key distribution resulting in unhackable connections among widely spaced global datacenters, the quantum intranet will connect the millions of supercooled quantum processors within a single quantum datacenter (most likely colocated with a supercomputer datacenter).

Last year, a quantum logic gate between distant quantum-network modules was demonstrated, but the extra level of sophistication required in a quantum intranet is its ability to connect any number of quantum computers while maintaining each qubit's superposition of states during transmission. Superposition is the mechanism by which quantum computers achieve their superiority to classical systems—that is, by simultaneously representing all possible solutions to a problem with coherence. Quantum processors manipulate these quantum states of superposition until the last step of decoherence into the digital solution that would take supercomputers years to calculate by considering each of the possible solutions separately (rather than simultaneously, as with qubits).

Recently, Argonne National Lab's Q-NEXT demonstrated the component needed to make possible a quantum intranet among 10-milliKelvin superconducting quantum computers. Communications of superpositions at 10 milliKelvins (-459 degrees Fahrenheit) is not feasible among superconducting quantum computers. By making the temperature 1,000 times warmer (10 Kelvins, or -442 degrees Fahrenheit), Argonne National Lab's silicon carbide (SiC)-based intranet is feasible, according to the researchers.

"The error rate of quantum communications in SiC at 10 degrees Kelvin is now on par with superconducting gate-level logic among qubits at 10 milliKelvins," said Bluhm. "This was the crucial step needed to vindicate the potential of solid-state semiconductor-based quantum computing. Notably, this demonstrated performance surpasses the theoretical requirements for large-scale quantum computing with real-time quantum error correction."

SiC semiconductors are well known and have proven reliable, according to Bluhm. Historically, SiC began as an industrial material that has been in mass production since 1893 (known for its physical toughness and stability over a wide range of temperatures). This tough material has been molded into myriad shapes and sizes for use in high-endurance applications such as abrasives, automotive braking systems and clutches, and the plates inside bulletproof vests. It also has a long history of uses in electronics, ranging from detectors inside early radios (circa 1907) to a durable receiver of optical waves in large (10-foot) telescopic mirrors fabricated by chemical vapor deposition (CVD).

Just as CVD is a well-known method with which a wide variety of complementary metal oxide semiconductors (CMOS) are fabricated, SiC has likewise diversified into thousands of electronic applications (those too rugged for CMOS). SiC can be operated at high temperatures without deforming (as silicon does), and at higher voltages than silicon can without breaking down, enabling long-lived light-emitting diodes (LEDs) whose heat energy can be quickly dissipated, heating sensors that can withstand temperatures up to 4,000 degrees Fahrenheit, cladding for nuclear fuel, junction-gate field-effect power transistors (JFETs), and many other applications.

Last year, the U.S. Department of Energy announced it would allocate up to $625 million over five years to support Q-NEXT and four other Quantum Information Science (QIS) Research Centers (the Co-design Center for Quantum Advantage at Brookhaven National Laboratory, the Superconducting Quantum Materials and Systems Center at Fermi National Accelerator Laboratory, the Quantum Systems Accelerator at Lawrence Berkeley National Laboratory, and the Quantum Science Center at Oak Ridge National Laboratory) where SiC and at least 10 other next-generation quantum materials are being characterized.

"We expect that the work carried out at these centers will be essential to realizing the quantum intranet," said Chow.

Why SiC?

The key to the unusual capabilities of silicon carbide is its atomic lattice, which combines silicon atoms and carbon atoms in over 250 different lattice structures, all with 50/50 silicon to carbon percentages, but all with different electrical/optical properties. SiC is routinely doped with nitrogen, phosphorus, beryllium, boron, aluminum, or gallium to further fine-tune its electronic and optical properties. SiC lattices also contain divacancies (two vacancies side-by-side), which are used to preserve the coherence of its qubits' five-second-long spin state.

Future architectures envisioned by the research team will use this long-lived amplified electronic spin to support massive transmission rates among quantum computers, to harness long-range entanglement, and for quantum-key distribution during unhackable cryptology that ultimately enables a worldwide quantum Internet. Besides networks, the 10,000-fold amplification of qubits is already being used in the lab to improve the sensitivity of quantum sensors, according to Awschalom.

Of course, SiC could be adopted by quantum computer makers themselves as the gate material, but many hurdles would have to be surmounted to rival the superconducting devices produced by gate-level developers today. For instance, even though SiC readouts of qubits have very long coherence times, so far the one-shot readout operation only works 80% of the time. The developers point out that they are only now beginning the optimization process to improve the one-shot success rate, but Awschalom also points out that since they know immediately whether the one-shot readout worked or not, the data transmission protocol can immediately repeat the process, making it no worse than retransmitting a corrupted data packet over Ethernet.

Kasra Sardashti, an assistant professor of physics and electrical and computer engineering at Clemson University, also calls SiC "a game changer" for connecting remotely located quantum computers. "This research also creates on-chip memories on SiC with a long lifetime. This will be a game-changer because long-distance communication of quantum states requires stable memories to be part of what we call quantum repeaters. This work shows that quantum repeaters may be built on-chip with SiC substrates."

Sardashti does not, however, expect an immediate impact on the SiC wafer manufacturing industry. "In the long term, perhaps one could expect an increase in demand as quantum tech becomes more developed. But the SiC industry is quite well-developed already, and should easily accommodate this demand in the short term, since SiC is highly compatible with CMOS large-scale processing."


R. Colin Johnson is a Kyoto Prize Fellow who has worked as a technology journalist for two decades.


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account