Hybrid cloud is to AI, what broadband was to the internet

Looking back on Think a few weeks ago, there was so much excitement around watsonx and generative AI. It helped me realize something: Hybrid cloud is the accelerant to AI adoption. It will do for AI adoption what broadband did for the internet.

Companies that build enterprise software solutions with hybrid cloud and AI as the foundation will be the ones that succeed. They will be better positioned to use their data effectively, efficiently and ethically. They will be better able to extend AI deep into their businesses and create unique competitive advantages. They will most likely beat those that do not.

There are three ways hybrid cloud is accelerating AI adoption: 

1. Unlocking the value of the data powering AI

 AI is only as good as the data that fuels it. But it’s not just about having the right data. Before it can be used effectively, that data must first be secure and accessible.

The challenge is that for most organizations, data is spread across multiple clouds, on prem, in private datacenters and at the edge. And data complexity is only getting worse. According to IDC, stored data will grow 250% by 2025 (1). Without better data management, finding the right data and putting it to use will only get more difficult.

Leaders must rethink ineffective and monolithic data architectures in favor of infrastructure that allows them to leverage data, wherever it resides. To do that, companies across industries are adopting hybrid cloud for the accessibility and flexibility it enables. Hybrid cloud strategies provide the data foundation needed to scale and operationalize AI, meaning the models the data feeds will be more accurate and enable more informed decision-making.

 2. Making better use of IT spend

A major constraint to AI adoption has been price. A common issue is that the costs of training and running models are hidden from the business, meaning there can be a disconnect between the value generated from a model and the engineering and compute resources needed to train and deploy it. That lack of visibility can prevent IT leaders from understanding and optimizing their spend, which is why many are taking advantage of hybrid cloud to support their AI investments.

We have worked with many organizations to use data to solve complex problems, including a marketing platform that helps retailers and nonprofits predict which consumers will be most receptive to different marketing campaigns. Moving to the cloud increased the platform’s ability to manage its data and leverage all its data assets, both offline and online. In addition to improving the performance of the models by 30%, the new infrastructure gave leaders visibility of training costs, which dramatically reduced from $1,500 per training run to hundreds of dollars per training run (2).

Think about what delivering better insights at a lower cost can do for an organization. That’s the impact that the combination of AI and hybrid cloud creates. 

3. Building trust into AI models

The mass adoption of AI hinges on trust. AI must be explainable, fair, robust and transparent. It also must prioritize and safeguard consumers’ privacy and data.  

Building AI solutions on a hybrid cloud architecture can enable the ongoing AI and data lifecycle management needed to engender that trust by ensuring proper governance and data security.

Compliance and security controls need to be built into a hybrid cloud architecture, so companies can determine who gets access to what and when and automate their compliance controls with the ever-expanding set of regulations. Hybrid cloud can also enable the real-time visibility needed to monitor model performance, detect security threats and mitigate bias -- all situations where system failure can result in the loss of your customers’ trust. 

***

Hybrid cloud has accelerated AI adoption because of the visibility, scalability and security it provides. And it’s only going to become more important as more companies adopt AI-first mindsets and build their businesses around a technology with the potential to radically transform business and society forever.

Because of this, infrastructure is an essential consideration of any enterprise AI strategy. In order for AI to fully deliver its value, how the data is stored, shared and protected can’t be overlooked in favor of what you want your application to do. Your first step must be deciding how to do all that in ways that are fit to your organization, customers and industry: from the integration and cataloging of data to model building and deployment to evaluating performance and finding ways to improve.                                                  

1.   Worldwide IDC Global DataSphere Forecast, 2022-2026, IDC, May 2022.

2.   How to create business value with AI https://www.ibm.com/thought-leadership/institute-business-value/en-us/report/ai-examples