The Last Mile Might be the Toughest

In the telecom industry, there is something known as the “Last Mile Problem”. The telecom companies can lay the latest fiber optic cables from city-to-city and even extend it to a local neighborhood. But the last mile from the local telecom junction box to your home may consist of old copper cable that was originally laid down decades ago. This last mile can limit the ultimate performance of an internet service at your home.

In speaking with a number of folks in the quantum industry we are seeing an increasing number of efforts implementing successful POCs (Proof of Concept) that are getting close to demonstrating quantum advantage. We believe that some of these efforts are getting ready to implement these applications within their organizations and achieve quantum advantage on a production basis. But we are not aware of any organization that has finished this yet. What is the bottleneck? 

It turns out there is a big difference between successfully demonstrating a POC and integrating it with all the classical systems and processes that exist within a large organization. This difference may not be fully appreciated by some in the quantum industry, particularly those who have not spent any time working in a large enterprise. There may be numerous system integration, security, regulatory, training, and testing considerations as well as the human factor of getting the schedulers, warehouse managers, truck drivers, floor traders, etc. trained and using the systems. People can sometimes be reluctant to change so it may take a while to get the operating people to fully embrace the new technology.

Modern enterprise IT systems are very complex and typically consist of many different applications from multiple vendors all integrated and working together to provide the functions needed to operate the enterprise as seamlessly to the end user as possible. Many new quantum applications that are introduced to this environment will require data inputs from one of these classical systems and produce outputs that will be used by another classical systems. Few production quantum applications will be run on a standalone basis. And when one is depending upon these classical systems to run billion dollar corporations, disrupting them in any way or introducing any potential security issue could turn out quite costly. So, CIOs are generally cautious about making sudden changes in their systems. Anyone who has been through a major ERP system upgrade will understand how difficult these changes can often be.

For those of you in the quantum industry who are working with end customers, it is wise for you to take this “Last Mile Problem” into consideration as part of your plans. Your revenues may not increase as fast as you might expect due to customer delays in putting applications into production. And perhaps you could consider offering support programs that would help get your customers through these final steps.

We asked a few quantum industry executives their take on this issue and received the following comments. Zapata has recognized this and has available workflow management platform called Orquestra that can help with the interfaces to the classical computing resources.

"While many enterprises understand they won’t simply 'rip and replace' their high-performance classical computers (HPC) with quantum devices and expect results, just as many seemingly do not. Adding quantum computing to an existing IT stack will result in an unconventional architecture for the foreseeable future – it will not be a swap with even the most powerful HPC resources. Quantum computing doesn’t work in a vacuum; it will always require a hybrid model combining both novel quantum resources and existing classical resources. And for at least the next several years, quantum devices will be accessed remotely through software platforms or directly (Quantum Computing-as-a-Service), exacerbating security, data management and compliance issues. Given that the quantum market is largely driven by hardware providers, this calls for a new approach that prioritizes integration with organizations’ existing classical IT stack.” Christopher Savoie, CEO of Zapata Computing

Some companies are implementing service programs that can help customers with this final stage of production implementation. Here's an example from D-Wave.

“As with every new technology, there is always an important adoption curve that involves everything from re-engineering teams and talent, and managing internal change, to bringing new types of technologies into existing architectures and workflows. Because quantum is a disruptive technology, having the right guide through the process is incredibly important. That’s why D-Wave has developed our services-enabled program, Launch, the on-board program to quantum computing designed to help enterprises go from problem discovery through production implementation. The program is structured into four phases; the last two of which are the deployment of a limited production-scale application to clearly demonstrate value and feasibility of the application, and once that is clear, proceeding with getting the application up and running in production.” Mark Johnson, Vice President, Quantum Technologies and Systems Products at D-Wave Systems

Quantum computing industry revenues today consist entirely of users learning how it could be used, experimenting with the technology, and developing POCs. But this has limits and for industry revenues to truly take off we will need to start seeing end users to start put their quantum applications into regular production use and achieve a significant commercial or scientific advantage that they could not obtain otherwise. So, don't ignore this "Last Mile Problem" and make sure that these promising quantum applications can get smoothly into regular production once they have shown a successful POC

December 28, 2021