The 2023 State of AI Infrastructure Survey, commissioned by Run:ai, sheds mild on the rising challenges confronted by organizations in AI growth. The survey, which was carried out by World Surveyz Analysis and gathered responses from 450 trade professionals throughout the US and Western EU, reveals that infrastructure and compute, chosen by 54% and 43% of respondents respectively, are actually the first hurdles, surpassing knowledge as the important thing problem going through AI growth. This marks a shift in comparison with final 12 months’s survey by Run:ai, the place the biggest variety of respondents – 61% cited knowledge as their prime problem.
The survey additionally pointed to a different shift over the previous 12 months because the variety of organizations deploying lower than half of their AI fashions in manufacturing elevated from 77%, in accordance with final 12 months’s survey, to 88% this 12 months. Weighted for common, simply 37% of AI fashions make it into manufacturing.
The adoption of cloud providers for AI infrastructure continues to rise, with 73% of surveyed organizations utilizing cloud providers. Nevertheless, the survey discovered a big problem in accessing GPU compute, as solely 28% of respondents reported having well timed and adequate entry to compute energy upon demand. This scarcity of on-demand entry results in frequent GPU allocation points for 89% of respondents who use a ticketing system.
“Regardless of being on the cloud, organizations are nonetheless going through limitations with unlocking the complete potential of their knowledge,” stated Omri Geller, CEO of Run:ai. “This highlights the truth that cloud hasn’t delivered on its on-demand promise and the significance of constructing a strong and scalable infrastructure.”
The survey additionally discovered that as organizations scale and require extra GPUs, they face a proliferation of third-party instruments, making it more and more advanced to handle AI infrastructure and get essentially the most out of it. 77% of respondents indicated they’re utilizing a number of third-party instruments, making it troublesome to get the correct quantity of compute to completely different workloads and end-users.
“Organizations should shift their focus from solely buying extra knowledge to making sure they’ve the right infrastructure in place to successfully course of and put it to use,” added Geller.
Another findings of the survey:
91% of corporations are planning to extend their GPU capability or different AI infrastructure by a median of 23% within the subsequent 12 months. This exhibits that regardless of the unsure financial local weather, corporations are nonetheless investing in AI as a result of potential and worth they see in it.
50% of corporations plan to implement monitoring, observability, and explainability within the subsequent 6-12 months to maintain monitor of their AI fashions.
The second and third priorities had been mannequin deployment and serving (44%) and orchestration and pipelines (34%). This means that corporations are targeted on bringing their AI fashions into manufacturing and streamlining the method to make it extra environment friendly.
Learn the entire 2023 State of AI Infrastructure Survey HERE.
Join the free insideBIGDATA e-newsletter.
Be part of us on Twitter:
Be part of us on LinkedIn:
Be part of us on Fb:
