Many of our colleagues in the enterprise and IT spaces grapple with the increasing costs of hosting data and business applications in the cloud, especially AI applications. According to a recent report published by UK cloud provider Civo, over a third of organizations surveyed reported that their cloud migrations failed to deliver on the promised cost-effectiveness. As a Technical Founder, I was uniquely positioned not only to foresee the pitfalls of full-cloud migration, but also to take action.

In 2012, I was given a book at a conference, the title of which I cannot recall. The book covered Cloud Computing and heralded the ubiquitous access to information and proliferation of cloud-based services. At the time, it was inspiring, but it wasn't until a year later that I would form a company, and it would take a few more years before we became a deep-tech organization.

Infinite 8's first foray into the cloud came through our dual efforts in Drone & AI Research and Development while also building our first commercial application. In 2014, as I was teaching myself to code with limited knowledge of IT infrastructure, we began with a truly Hybrid Infrastructure. We used powerful simulations that blended Drone hardware-in-the-loop with cloud-based AI technologies. Yet, our first app was built entirely in the cloud, using IBM's BlueMix and specifically a third-party vendor, Mendix.

Through IBM's Global Entrepreneur Program, we were granted $15,000 in cloud credits. However, once our app was about 80% complete and our credits began to run out, we realized that continuing our cloud journey would cost between $1,000 and $2,500 per month—an unsustainable expense for a bootstrapping startup. As a result, we released Dronetrepreneur as a web application without the need for third-party services.

While this web application discovery was useful, two key developments changed our trajectory. First, we were introduced to Linux, and second, we discovered that IBM's Watson was running on Nvidia server architecture. We transitioned to dual-boot systems, running both Windows and Linux, systems which also incorporated Nvidia GPUs. This shift enabled us to build applications and train our AI models locally, avoiding prohibitive cloud costs.

Still, we found this setup insufficient, leading us to abandon Windows and cloud architecture entirely by 2018. We invested in gaming computers and a high-end GPU cluster to extend our infrastructure, enabling us to build AI models and run compute-intensive simulations. However, even with this more powerful hardware, simulations were still bottle-necked, and AI training continued to take days.

This led us to explore Quantum Computing, in 2018, dedicating staff full-time to this emerging vertical. By that time, we had become proficient in programming drones and autonomous vehicles on resource-constrained embedded devices, prioritizing code security and optimization. Our goal was always to develop applications for deployment on physical machines with limited power and memory footprints. We concluded that Quantum Computing was the only way to scale Artificial Intelligence effectively. As early as 2017, we realized that a Quantum Network was essential to run AI applications and advanced simulations. Anything less would introduce too much latency and too many security vulnerabilities for real-time systems such as self-driving cars and drones operating in dense urban environments.

It took half a decade for Infinite 8 to realize its vision of a Quantum Network, which became InfiNET. Our recent project running on InfiNET, ONEAI, is the fulfillment of that vision. We chose a hybrid cloud setup, utilizing local and on-premise infrastructure primarily, with cloud services via API as a secondary option for high-end applications where increased costs are justified. This approach has delivered increased energy efficiency, faster processing and data transfers, enhanced security, and local control, providing us with a scalable architecture that feels limitless. Today, we run InfiNET and ONEAI on clusters of less expensive Nvidia Jetson GPUs locally, with cloud API access to Superconducting and Annealing Quantum Computers.

Now, many companies are facing the same challenges in scaling AI infrastructure, much like we did in 2015. Should you build in the cloud, outside the cloud, or adopt a hybrid approach? The answer depends on the business case, the organization's resources, and a long-term commitment to AI development and commercialization. Fortune 500 companies, global governments, hedge funds, mutual funds, and family offices are beginning to realize what we did years ago—AI is power-hungry, and it needs Quantum. Whether it takes a decade for smaller and midsize firms to reach this same conclusion remains to be seen. But make no mistake: Quantum is coming to save the data center, and it will be integrated into our lives, often without us even realizing it (e.g., GPS, MRI machines).

Wherever you are on your cloud infrastructure journey, Infinite 8 has been there. Together, we can solve these challenges and better serve our customers and the environment in an increasingly technological world. We wish you the best of luck on your cloud infrastructure journey. - Ean Mikale, JD, Founder, Infinite 8 Ecosystem | www.infinite8industries.com