Artificial intelligence (AI) was undoubtedly one of the most discussed topics in the boardrooms of organizations in Belgium and the rest of the world last year. Nevertheless, companies find it quite difficult to integrate the technology.

According to a recent study by Alex De Vries, a doctoral student at the School of Business & Economics in Amsterdam, by 2027 the AI ​​sector would consume as much energy as a country the size of the Netherlands. Other scientists have previously stated that they are concerned about the impact of AI on the environment. Although data centers are undoubtedly among the most advanced players in terms of energy consumption and carbon emissions, there is clearly still a lot of work to be done. Especially now, when interest in AI technology is growing faster than ever before.

Not ready for AI

Generative AI tools like ChatGPT have been the talk of 2023. Companies everywhere have created working groups to explore how AI can advance their businesses. Nutanix’s State of Enterprise AI report shows that companies of all sizes want to adopt AI technology as quickly as possible. But while 90% say AI is a priority for their organization, the report also makes it clear that we’re actually not ready for it yet.

In particular, companies struggle to select the IT environments that are best suited to run different parts of their AI processes and workloads. Or they stumble over finding the right AI applications. In any case, we see several challenges that could also hinder the adoption of AI in 2024.

Given the increasing focus on ESG reporting, it is no surprise that sustainability is a major barrier to implementing AI. Additionally, the State of AI report shows that security and reliability are also important pillars when developing an AI strategy for 90% of respondents. Data security and governance, including the quality and protection of data, are of utmost importance for AI technology and services.

ESG reporting

What is striking is that many survey participants indicate that they need to align their ESG reporting with their AI needs, but only a few companies have the corresponding skills in-house. ESG is one of the key areas where developing AI skills will be critical over the next twelve months.

In Europe, ESG reporting will become mandatory for a large group of companies from January 1, 2024 – a measure that will be felt far beyond the EU. Reporting breaks down emissions from business activities using three “scopes” that distinguish between direct and indirect emissions. Due to the enormous impact of AI on IT infrastructure, the rise of the technology seems rather unfavorable. Companies understand how much energy is required to run compute- and GPU-intensive algorithms and workloads.

Organizations must think and plan carefully before deciding whether to go it alone with custom on-premises AI systems, leverage public cloud solutions, or opt for a standalone edge platform.

Data privacy

Another concern is data protection. To train AI engines, you need data. But how can a company maintain control of its own data when using AI services running in the public cloud? This lack of control is a governance issue because companies cannot be sure where their data ends up. In addition to potential latency issues and costs associated with the flexibility of the public cloud, data protection should be a primary goal.

Another challenge arises from the multi-cloud strategy that has become established in many companies. For this reason, they have distributed their data across different (public) cloud environments such as Microsoft Azure, Google Cloud and AWS. In the worst case, this means information is tied to one provider’s services, making it not easy to bring everything together for a generative AI tool and provide enough capacity, says Nutanix’s Luc Costers.

The cost of a DIY approach, the global shortage of GPUs, and the in-house AI capabilities required make it impossible to do this alone. A standalone AI platform at the edge not only offers data protection and speed, but also reduces the footprint through less energy consumption and lower carbon impact. Companies need to know how to use AI, but the young age of this technology segment means it still lacks strategic best practices, established guardrails or reference architectures.

A new look at infrastructure

Despite the complexity of such a DIY approach, the survey shows that 59% still expect to run AI solutions on-premises or in a private cloud. 51% choose a data center and 44% choose the edge. The challenge of managing these services within the constraints of ESG and data governance will soon determine how these numbers actually perform.

The basis is the infrastructure. As companies pursue an AI future, they must look at infrastructure differently and separate their applications from the underlying infrastructure. We shouldn’t think of AI as an application that we can just throw on the stack of cloud computing applications and everything will be fine.

The stakes are much higher: in addition to data security and the future of the planet (as if that weren’t enough), we also need to consider the profitability of companies using AI systems. By addressing these challenges now, we can prevent companies from making mistakes that ultimately lead to costly damage to the environment or their reputation.

This is a contribution from Luc Costers, Regional Manager Nutanix Belux, CIS and Eastern Europe. Click here for more information about the company’s services.