DEVOPSdigest

Vultr and S&P Global Market Intelligence recently released a report titled The New Battleground: Unlocking the Power of AI Maturity/MultiModel AI. The research — unique in that 72% of the 1,000 decision-makers surveyed have helped their organizations achieve advanced levels of AI maturity — charts the paths mature AI organizations have taken as a guide for those still in the earlier stages.

The report uses a four-stage model to characterize the level of AI usage in the respondents' organizations as follows:

■ Transformational (32%): AI is built into the fundamental aspects of their business and is a value generator for their customers

■ Accelerated (40%): They have many daily uses and are seeking new use cases and deploying AI where possible

■ Operational (27%): They have begun to adopt AI in some of their daily operations

■ Experimenting (1%): They have not deployed any large-scale projects

Here are a few of the key findings:

AI Pioneers Are Prospering Most

Respondents at the transformational (highest) maturity level indicated they are significantly outperforming their peers at a rate 50% higher than those at the operational (third highest) maturity level. They also have greater expectations for growth across core outcomes measured, including customer satisfaction, revenue, and market share.

However, nearly all companies with at least some AI deployed say their 2022/2023 year-over-year performance improved across these metrics, and 49% seek to build on this success through moderate to significant AI spending increases next year.

In fact, 88% of enterprises intend to increase their AI spending in 2025 to support their crucial infrastructure strategies. This growth in AI spending is expected to outpace growth in IT spending.

Welcome to the Multi-Model Era

The number of models actively used within an organization says much about a company's AI maturity level, and on average, organizations across all maturity levels have 158 models in production. This number is expected to reach 176 models within the following year.

As part of the ramp-up, 89% of organizations surveyed expect to deploy advanced AI within two years, and 80% plan to adopt AI across all business functions.

The Tech Driving the Triumph

Survey results suggest the preferred AI infrastructure stack will be hybrid cloud, with 35% of inference taking place on-prem and 38% in the cloud/multi-cloud. For cloud-native AI applications, two-thirds of organizations are either custom-building their models or using open-source models to deliver functionality.

Given the skills shortage, among other factors, 47% of enterprises have partnered (or plan to) with AI specialists (25%) or global system integrators (22%) that can specifically guide the strategy and implementation required to support the deployment of AI at scale. Only 15% are turning to traditional hyperscalers.

Security and compliance, open ecosystems, and technical expertise are the top attributes organizations value most when selecting such partners to scale AI across the organization, dispersed geographies, and to the edge — with these desired attributes outpacing even cost.

Living Deliberately (but Dangerously?) on the Edge

80% of survey respondents expect to grow their AI edge operations in the near future, but specialized requirements can strain existing infrastructure. When considering high-demand AI activities such as real-time inferencing, respondents expressed concern that existing infrastructure would not be up to the task. The top three concerns were insufficient CPU or GPU resources (65%), data locality issues (53%) and storage performance issues (50%).

Other Challenges

As the competition in AI intensifies, organizations will face numerous challenges. Budget constraints, developing or acquiring AI algorithms, a shortage of skilled personnel (as mentioned), and data quality are among the primary obstacles the organizations surveyed say they must overcome to advance to the next stage of AI maturity.

Governance (30%) is the chief concern for enterprises at the transformational maturity level, whereas company culture poses a more substantial challenge for those still in the accelerated (second highest) maturity phase.

A New, Pervasive AI Order

The report makes clear that organizations that continually build on their AI capabilities will likely experience enhanced operational efficiency and competitive advantage. It's no longer about building and training models but rebuilding all apps with AI at their core.

Organizations must integrate all the principles of cloud engineering and CI/CD pipelines before layering in AI. It's no longer enough to be cloud-native. Organizations must be AI-native.