Earlier this year we wrote that quantum computing was likely 50 years from market penetration, and still not ready for large-scale application in drug discovery, forecasting, and encryption cracking (its most likely first applications). A team of NewtonX quantum computing experts comprised of 3 leading theorists and 2 engineers had indicated three significant barriers to entry for quantum: decoherence, cost, and lack of talent. Today, however, a startup called Rigetti has emerged as a head-to-head competitor to IBM (where the founder of Rigetti previously worked on quantum), and, through a hybrid SDK, has opened up quantum computing to developers. Will the startup push quantum computing into “quantum supremacy,” the point at which quantum algorithms can perform equations faster, better, and cheaper than traditional computing can?
NewtonX revisited the experts we interviewed in early 2018 as well as an additional former senior engineer with IBM’s quantum team, in order to gain insights into advances in quantum computing and new applications for hybrid quantum-classical systems.
(Not sure what quantum is? NewtonX experts provided an explanation in our previous article on the subject, which you can find here).
The New Quantum: Just as Good as Classical, but not Better
In the last NewtonX article on Quantum, one of our experts, formerly a quantum researcher at MIT, stated, “Quantum supremacy is an invented moment when quantum computers will be more efficient and cheaper than traditional ones. This idea motivates people to invest in it, follow it, and work on it. It’s a promise of unlimited power propagated by tech giants.” He highlights that this simplified model is very unlikely to become a reality: “The Quantum Computing curve will most likely not follow the Quantum Supremacy model propagated by Google, it will instead show very uneven success depending on use cases.”
This prediction has thus far proven to be true. Rigetti’s quantum computers and cloud solution for quantum computing are error prone, and chips consists of only 16 qubits – not enough to eliminate the rate of error. The company is tiling them together into systems with 32, and plans to up that to 128 qubits in the coming year. By comparison, Google currently has a 72 qubit system, IBM has one with 50 qubits, and Intel has one with 49 qubits. Despite Rigetti’s low-qubit chips, however, the company has a dozen initial users for its hybrid cloud solution, many of which are pharma companies using the platform for drug discovery.
Drug discovery is one of several applications that does not require a perfect answer, down to the last bit, which makes imprecise quantum solutions useful despite errors. Because drug discovery deals with billions of molecule combinations, if, out of all of those possibilities, a quantum or hybrid quantum solution can can give a shortlist of options to test in the lab faster and cheaper than traditional computing can, then that’s good enough to make the system worth it. Today, drug discovery researchers use classical computers to simulate the interactions between atoms and molecules, but these simulations require massive amounts of computing power, and are therefore extremely time consuming and expensive. Quantum computing can calculate the distribution of the electrons inside molecules, enabling scientists to rapidly simulate molecules for new drugs.
Hybrid algorithms are better at solving mistakes than pure quantum algorithms, and in industries that can tolerate errors, 25-50 qubits can be powerful enough to reach a quantum advantage.
Despite Advances, The Old Barriers Persist
Quantum still performs best in probabilistic situations. It is effective in applications such as predicting the weather, market forecasting, and, as we outlined above, predicting which molecules would work best for certain drugs. However, of the three barriers outlined in our last article on this subject, only one has been addressed: quantum talent.
Rigetti and IBM allow developers to access quantum computers through the cloud, which has expanded the pool of developers who know how to code in quantum. However, Google, Intel, IBM, and Rigetti still struggle with decoherence (Quantum computers need to be isolated from the external world lest the environment cause the system to discohere) and cost (Rigetti has raised over $100M from investors including Andreessen Horowitz). Currently quantum compute centers use superconducting technology, which requires cooling systems at close to absolute zero, and then they manipulate the qubits using microwaves. This process often results in imprecision or errors. While cost will likely decrease with time (just as it did for traditional computing), currently it is still prohibitively high, meaning only a very few elite companies can truly experiment with it.
At the end of the day, while Rigetti’s hybrid cloud solution is a step forward for the industry, and introduces a new player into a scene dominated by tech giants, quantum computing is not necessarily closer to enterprise adoption.