CxO: Quantum Computing for Execs

After artificial intelligence, Quantum Computing is a growing buzzword that promises big returns on investments. This short read is from my experience taking real financial problems and solving them on a real Quantum Computer rather than hype.

Why? What does it mean for my org?

The majority of conventional technologies operate on a processor that follows a sequence of instructions to process your organisation's data - called a Von Neumann architecture - and this includes CPUs, GPUs and the majority of TPUs used in AI for example.

Quantum Computing differs by using atoms* of matter to model probabilities, in fact using a technique called superposition it models all probabilities at once returning a massive speed increase and allows solving modelling of problems too demanding or unviable for conventional computers.

How fast? Let's take a security problem, the maths behind this is simple to find the correct factors that make up a large number. Using Shores algorithm, the largest supercomputers (still a conventional computer) will take 6.4 quadrillion years to break 2048 bit RSA encryption that secures your organisation's data. A quantum computer will take 10 minutes. The same benefits appear for financial modelling with quantum friendly algorithms.

There is the first difference - conventional computing works for data and can solve existing models; quantum computing works for complex mathematical models. In fact the new mathematics is so new that quantum algorithms to solve problems are appearing for financial, AI and security every day. Quantum based approaches are also providing benefits with new solutions to problems on conventional computers.

Skillsets

This leads to the second difference - workforce skillsets. Software development skills are common and in demand as organisations want to process their mountains of data, using languages such as Java, or Python. The process of converting the business problem into a running system is simple and understood, backed by years of research and bitter experience.

The current skillset for Quantum Computing is predominately advanced mathematics in order to convert the business problem or existing modelling algorithm and create the mathematics based on probabilities compatible with a Quantum Computer.

Vendors are already creating tools for developers to use their cloud based Quantum Computers, however the skills that your organisation will need to build is within advanced mathematics. Consultancies dedicated helping organisations with Quantum mathematics already exist, however the issue becomes communication of problems or concepts at the level of maths required.

If you're reliant on modelling for a competitive advantage, I would advise starting with funding university PHD research specifically around developing quantum mathematical approaches to solving the problem for your industry. Next I would create a capable pod of quantum maths capable skills to apply to various problems within your organisation's competitive market modelling - this will also appear through a quantum-enabled software tool vendor providing to a larger organisation. The tooling passes on the benefit without needing a large change in client organisation data science/modelling skillsets but you will not have exclusivity.

For common problems such as providing quantum-safe encryption of organisation data the major vendors such as IBM, MSFT and GOGL are racing to provide profitable solutions.

As each vendor's quantum hardware technology is optimised to solve problems in a specific mathematical approach, any selection criteria must align to the mathematical model.

Timelines

4-6 years with the current rate of technological advancement before the first Quantum Computers large enough to solve current larger financial models and become a true risk encryption. Software tooling vendors will start to provide competitive edges with quantum-enabled features with early adopters needing advanced maths.

5-10 years we will see common reusable software components for common problem solving existing software developer skillsets with more advanced techniques still requiring advanced maths. The problem of modelling existing mountains and realtime streams of data using AI will start in the next few years, however it will only be viable based as the capability of the QC systems increases.

In 10 years the risk to data and competitive advantages will be apparent with common 2048bit RSA encryption broken and quantum algorithms available. Tooling for AI and modelling techniques will continue to grow beyond 15 years as PHD research grows.

It is true that Quantum Computers are in their infancy with infancy-related restrictions to the problems being targeted. However it is also true that it is accelerating faster than conventional computers based on lessons learnt from the last 50 years.

How will this affect my non-modelling organisational data?

Artificial intelligence is being taught maths, AI is being taught to create models using real data whilst data scientists are searching for valuable insights from data as models.

It's possible to create an AI using Quantum Computers and real examples already exist.

It will not be long before an AI or quantum tooling exist to speed up AI training, and then create quantum-friendly mathematical models derived from the mountains of data.

Thank you for reading!

If you liked that, please give me a thumbs up. You can follow us and our HSBC Hackademy 2019 colleagues by searching the tag #CodeLearnInnovate as we look at a second real business problem with Quantum Computing for Phase 2 for the Hackathon.

* - for readability"atom" rather than use terms Cooper Pairs,Einstein-Bose, wave functions etc.