• There’s a race right now in high tech to build the first general purpose quantum computer, with industry leaders IBM, Google, D-Wave Technologies, and Intel each building out very different implementations of a single, revolutionary idea — the use of qubits instead of plain old bits.
• But unlike most races, this one has no clear finish line as we’re still figuring out the best approach to quantum computing or to building software for them. Enter IT services powerhouse Atos, which is backing a pure but as yet simulated idea of quantum computing in an effort to garner what matters most, namely the hearts and minds of future quantum developers.
There’s an awful lot of noise in the technology industry right now regarding the promise of quantum computing. A sizable number of dissimilar technology and platform players, ranging from Intel to Google to Atom Computing (a 2018 startup) are all busy building increasingly capable computers that push and pull qubits rather than bits. And as you might expect from such a diverse cast, there are a lot of differing views on how to build such a beast and how to best put it to use.
What if the best way to win at quantum computing doesn’t depend on the best way to corral the largest number of qubits or the best temperature capable of supporting the longest stable quantum state?. What if the best approach is a bit more academic — literally academic?
Consider European IT services and consulting provider, Atos, which offers Atos Quantum Learning Machine (Atos QLM), a completely theoretical simulation of a working quantum computer. Announced last July, QLM is an appliance designed specifically for quantum software developers that executes code as if it were running on a genuine, quantum computer. And it does this with double digit accuracy, something many vendors are struggling to achieve at scale.
Why choose the theoretical over the tangible? First, a product like QLM let developers learn how to build algorithms that take full advantage of quantum computing without having to deal with any of the constraints (compromises) inherent with current, working quantum computers — namely the short-lived nature of stable quantum states and a lack of data reliability due to extraneous noise. Second, by adhering to a very low level, assembly-style programming language, QLM and its ilk allow developers to build algorithms that are themselves not dependent upon any single hardware implementation. Atos’ QLM even sports a Python pre-compiled extension to allow developers to start with or incorporate more traditional programming paradigms without any worry about software lock in.
The Atos QLM is basically a future-proof proof of concept machine. But it isn’t without its blemishes. Quantum computing is different because it can scale exponentially. Atos’ machine, which is built on a sizable machine (16 CPUs and 24 terabytes of memory) can only scale to emulate a 40 qubit machine. Compare that with Google’s working 72 qubit Bristlecone processor, and already there are use cases where QLM cannot tread. Likewise, as an on-premises only appliance, QLM isn’t for everyone. As mentioned above, Atos has only garnered one customer, the U.S.-based Oak Ridge National Laboratory (ORNL). And QLM is only available as an on-premises appliance, which will hurt Atos’ efforts to drive broad interest among developer communities.
Hopefully more follow for both Atos and its many and varied rivals, because at the end of the day, regardless of the underlying hardware, what matters most is software, or to be more specific – algorithms. The value of quantum computing doesn’t rest in exponential speed hikes but rather in the strange twists and turns native to quantum physics. It needs a new type of developer, one well versed in ideas like entanglement, tunneling, and superposition.