At the latest TechCrunch Disrupt conference IBM provided a visionary speech on the future of compute using quantum computing. IBM Research COO Dario Gil gave a very cogent description of quantum computing and how it will change the computing landscape in the near future.
Quantum computing is a very complex and esoteric technology to try to explain to an audience of entrepreneurs and developers looking to raise money for the next Snapchat. Interestingly enough, there was a quantum computing start up at Disrupt, Rigetti Computing, pitching a quantum computing cloud service. IBM introduced its quantum computing cloud service in May 2016.
Dario tried to make the concepts as clear as possible using common analogies. There are three main quantum concepts that need to be made: superposition, entanglement, and interference. The concept of superposition comes into play because a quantum qubit is more than a digital 1 or 0. The analogy Dario made was a coin – while there is a head and a tail, consider spinning coins where the spin is part of the equation for the state. Superposition also gives us the exponential growth in the amount of data that can be used during an active quantum computation. While 1 qubit contains two pieces of information, 10 qubits contain 210 = 1024 pieces.
Entanglement is a strange concept – two qubits, once they have been entangled will have the same synchronized state no matter how far apart they are. Both superposition and entanglement are part of the famous thought experiment of Schrödinger's cat.
Finally, there is interference where the right answer is amplified by the interference of qubit states and wrong answers are cancelled out. Interference is most commonly associated with wave theory. Interference is important in the development of many quantum computing algorithms.
It’s still a mind-bending concept to think of. The power of quantum computing is that with enough high quality, usable qubits, we may be able to simulate and solve extremely complex problems. In fact, some of these problems are actually quantum in nature, such as in chemistry. Today we approximate chemical reactions and quantum states to keep the problem scalable with classical computing. But with a quantum computer the problem can be more accurately modeled and scaled.
The IBM quantum computer runs at an amazing 15 milli-Kelvin (0.015 Kelvin), only very slightly more that absolute zero temperature (0K), which is actually colder than deep space. To get to that extreme temperature, IBM built a complex, layered cryostat where at the bottom are the qubit processing elements. IBM builds their specialized chips in its own research and development fab.
While many companies talk about how many qubits they have in their quantum computer, there’s an important distinction in the quality and life span of these qubits. Unlike digital computers, quantum qubits have a very limited life. In the 1990s a good qubit might last one nanosecond (a thousandth of a millionth of a second). Today, a high-quality qubit lasts 50-100 microseconds. The lifespan is still short, but long enough to perform calculations. For better computing we may ultimately need to use logical qubits that are stable over time (using hundreds or thousands of today’s physical qubits to make one logical qubit).
While we are still early in the quantum computing evolution, more than 97,000 users of IBM’s free Q Experience 5- and 16-qubit systems have run over 6 million experiments. The commercial IBM Q Network, which has access to a 20-qubit system, and includes hubs, partners, and startups, are working on paving the way for commercial applications of quantum computing. Both users of the free IBM Q Experience and IBM Q Network members can use Qiskit, an open source software platform that provides a level of abstraction so that you don’t need a graduate degree in quantum physics to program the IBM quantum computer.
We can simulate qubits in classic computers, but the number of bits to describe qubits grows exponentially with powers of 2. Inversely, using qubits, we hope to crunch problems that would take an unrealistic number of bits to process. The challenge is not more qubits. It is still an R&D problem to create many qubits with acceptably low error rates. Qubits are still delicate. When comparing different quantum computers, it is important to understand the quality of the qubits.
IBM feels confident we will be getting to the point where quantum demonstrates better performance, and at that point we will reach “quantum advantage,” but that is still at least 3 to 5 years away.
Even then, certain problem will require more than the noisy qubits we work with today. The holy grail is to build a fault tolerant universal quantum computer. One concern about quantum computers is the potential to break today’s factor-based cryptography. To do so reliably will require these fault tolerant systems, which are years, probably decades, away. That’s because factoring two large numbers is an exponentially hard problem. By that point we will need to replace today’s algorithms with a new generation of algorithms (lattice, etc.) that will be “quantum safe.” IBM and others are already actively at work to standardize these new methods.
The question that you may ask is: will quantum computers replace today’s Intel Xeons? The answer is no. Quantum will be part of a new hybrid processing paradigm. Quantum computing elements will solve specific problems but will not replace classic computers (like Intel Xeon and IBM POWER) that run the operating systems like Linux and AIX and control programs. In fact, the future hybrid cloud compute model will likely include classic computing, AI processing, and quantum computing. When it comes to understanding all three of those technologies, few companies can match IBM’s level of commitment and expertise.
Kevin Krewell
Principal Analyst, TIRIAS Research
Twitter: @Krewell
The author and members of the TIRIAS Research staff do not hold equity positions in any of the companies mentioned. TIRIAS Research tracks and consults for companies throughout the electronics ecosystem from semiconductors to systems and sensors to the cloud.
